diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.960159753019871,
+  "eval_steps": 500,
+  "global_step": 3200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.5408,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 1.6226,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.4483,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.5755,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.4079,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.3362,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.3585,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.4116,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 1.1207,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 1.071,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.0267,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.9838,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 0.9071,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.8284,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.8107,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 0.9975,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 0.8088,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 0.7196,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 0.7472,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 0.7957,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.7088,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 0.8995,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 0.7322,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.6688,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 0.7607,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 0.6489,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.6596,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.6976,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 0.6662,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.6572,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 0.6489,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 0.8052,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.651,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 0.6393,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.6384,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 0.5832,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 0.63,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 0.6984,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.6675,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 0.6394,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 0.645,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.6478,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 0.6011,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 0.6373,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.5824,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 0.6449,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 0.6255,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.5936,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5817,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0408163265306126e-05,
+      "loss": 0.6055,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 0.5531,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.122448979591837e-05,
+      "loss": 0.5816,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1632653061224492e-05,
+      "loss": 0.5801,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 0.6879,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2448979591836734e-05,
+      "loss": 0.5773,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.6107,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3265306122448982e-05,
+      "loss": 0.6083,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3673469387755103e-05,
+      "loss": 0.6046,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4081632653061227e-05,
+      "loss": 0.5721,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 0.6235,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4897959183673473e-05,
+      "loss": 0.5901,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5306122448979597e-05,
+      "loss": 0.6093,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5714285714285718e-05,
+      "loss": 0.63,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6122448979591835e-05,
+      "loss": 0.5879,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.653061224489796e-05,
+      "loss": 0.6044,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 0.5547,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7346938775510205e-05,
+      "loss": 0.5784,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.775510204081633e-05,
+      "loss": 0.6148,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 0.5773,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8571428571428574e-05,
+      "loss": 0.7986,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.89795918367347e-05,
+      "loss": 0.5985,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.938775510204082e-05,
+      "loss": 0.6271,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9795918367346944e-05,
+      "loss": 0.5813,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.020408163265306e-05,
+      "loss": 0.6327,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.061224489795918e-05,
+      "loss": 0.5897,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.102040816326531e-05,
+      "loss": 0.6402,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.6197,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.183673469387755e-05,
+      "loss": 0.5846,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.224489795918368e-05,
+      "loss": 0.5615,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.26530612244898e-05,
+      "loss": 0.6572,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.306122448979592e-05,
+      "loss": 0.5902,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.346938775510204e-05,
+      "loss": 0.6059,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.387755102040817e-05,
+      "loss": 0.5948,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 0.5405,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.469387755102041e-05,
+      "loss": 0.5467,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.510204081632653e-05,
+      "loss": 0.6759,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.551020408163265e-05,
+      "loss": 0.5588,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.591836734693878e-05,
+      "loss": 0.5441,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.63265306122449e-05,
+      "loss": 0.5972,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 0.618,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.714285714285715e-05,
+      "loss": 0.564,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.755102040816327e-05,
+      "loss": 0.5565,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.795918367346939e-05,
+      "loss": 0.5825,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.836734693877551e-05,
+      "loss": 0.5789,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8775510204081634e-05,
+      "loss": 0.5945,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9183673469387755e-05,
+      "loss": 0.5823,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.959183673469388e-05,
+      "loss": 0.5709,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4e-05,
+      "loss": 0.6075,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999999015359061e-05,
+      "loss": 0.5627,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999996061437214e-05,
+      "loss": 0.721,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999991138237367e-05,
+      "loss": 0.7032,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999984245764368e-05,
+      "loss": 0.5397,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999975384025003e-05,
+      "loss": 0.5984,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999964553027998e-05,
+      "loss": 0.5805,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999951752784018e-05,
+      "loss": 0.4845,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9999369833056664e-05,
+      "loss": 0.7212,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999920244607485e-05,
+      "loss": 0.5422,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999901536705957e-05,
+      "loss": 0.5477,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999880859619502e-05,
+      "loss": 0.6164,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9998582133684795e-05,
+      "loss": 0.5808,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999833597975188e-05,
+      "loss": 0.5332,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999807013463865e-05,
+      "loss": 0.5697,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9997784598606866e-05,
+      "loss": 0.6157,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999747937193768e-05,
+      "loss": 0.5445,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999715445493162e-05,
+      "loss": 0.558,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999680984790863e-05,
+      "loss": 0.5616,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999644555120801e-05,
+      "loss": 0.5585,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999606156518847e-05,
+      "loss": 0.5568,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999565789022809e-05,
+      "loss": 0.5454,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999523452672437e-05,
+      "loss": 0.5,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999479147509413e-05,
+      "loss": 0.6099,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999432873577365e-05,
+      "loss": 0.5461,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999384630921855e-05,
+      "loss": 0.5449,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999334419590384e-05,
+      "loss": 0.5368,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999282239632393e-05,
+      "loss": 0.5477,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9992280910992614e-05,
+      "loss": 0.5709,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9991719740443043e-05,
+      "loss": 0.5792,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999113888522777e-05,
+      "loss": 0.4995,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999053834591875e-05,
+      "loss": 0.5308,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9989918123107264e-05,
+      "loss": 0.5507,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9989278217404043e-05,
+      "loss": 0.552,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998861862943915e-05,
+      "loss": 0.5466,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998793935986203e-05,
+      "loss": 0.5632,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9987240409341545e-05,
+      "loss": 0.5475,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998652177856588e-05,
+      "loss": 0.5018,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998578346824266e-05,
+      "loss": 0.553,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9985025479098836e-05,
+      "loss": 0.5851,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998424781188076e-05,
+      "loss": 0.5649,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998345046735415e-05,
+      "loss": 0.5466,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9982633446304116e-05,
+      "loss": 0.6372,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998179674953512e-05,
+      "loss": 0.5694,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998094037787101e-05,
+      "loss": 0.6133,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998006433215501e-05,
+      "loss": 0.5308,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9979168613249696e-05,
+      "loss": 0.5302,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9978253222037044e-05,
+      "loss": 0.5155,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997731815941839e-05,
+      "loss": 0.5054,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997636342631442e-05,
+      "loss": 0.5406,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997538902366521e-05,
+      "loss": 0.6507,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997439495243019e-05,
+      "loss": 0.5599,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9973381213588174e-05,
+      "loss": 0.7044,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997234780813733e-05,
+      "loss": 0.5081,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9971294737095184e-05,
+      "loss": 0.4957,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997022200149864e-05,
+      "loss": 0.5439,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.996912960240395e-05,
+      "loss": 0.5219,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.996801754088674e-05,
+      "loss": 0.5821,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996688581804199e-05,
+      "loss": 0.5769,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9965734434984035e-05,
+      "loss": 0.6,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996456339284659e-05,
+      "loss": 0.5236,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996337269278269e-05,
+      "loss": 0.524,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996216233596477e-05,
+      "loss": 0.5907,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9960932323584577e-05,
+      "loss": 0.5289,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9959682656853234e-05,
+      "loss": 0.526,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995841333700122e-05,
+      "loss": 0.4744,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995712436527837e-05,
+      "loss": 0.5518,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995581574295384e-05,
+      "loss": 0.5191,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995448747131616e-05,
+      "loss": 0.4819,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.99531395516732e-05,
+      "loss": 0.5997,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9951771985352174e-05,
+      "loss": 0.6265,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9950384773699646e-05,
+      "loss": 0.5011,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9948977918081526e-05,
+      "loss": 0.5512,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9947551419883064e-05,
+      "loss": 0.5527,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.994610528050883e-05,
+      "loss": 0.5616,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.994463950138277e-05,
+      "loss": 0.5702,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9943154083948156e-05,
+      "loss": 0.5957,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9941649029667564e-05,
+      "loss": 0.5287,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.994012434002295e-05,
+      "loss": 0.5265,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9938580016515593e-05,
+      "loss": 0.534,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.993701606066608e-05,
+      "loss": 0.4914,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.993543247401435e-05,
+      "loss": 0.5216,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9933829258119684e-05,
+      "loss": 0.5464,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9932206414560645e-05,
+      "loss": 0.5121,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9930563944935176e-05,
+      "loss": 0.4726,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9928901850860505e-05,
+      "loss": 0.6841,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992722013397322e-05,
+      "loss": 0.5765,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992551879592918e-05,
+      "loss": 0.5212,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9923797838403596e-05,
+      "loss": 0.5167,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992205726309101e-05,
+      "loss": 0.4858,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.992029707170526e-05,
+      "loss": 0.5018,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9918517265979486e-05,
+      "loss": 0.4893,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.991671784766618e-05,
+      "loss": 0.5577,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9914898818537106e-05,
+      "loss": 0.5006,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9913060180383364e-05,
+      "loss": 0.5139,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9911201935015345e-05,
+      "loss": 0.5073,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.990932408426276e-05,
+      "loss": 0.5418,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9907426629974616e-05,
+      "loss": 0.4805,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.990550957401922e-05,
+      "loss": 0.5284,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.990357291828419e-05,
+      "loss": 0.4808,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.990161666467644e-05,
+      "loss": 0.4821,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9899640815122156e-05,
+      "loss": 0.4994,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9897645371566864e-05,
+      "loss": 0.4918,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9895630335975354e-05,
+      "loss": 0.511,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.98935957103317e-05,
+      "loss": 0.5141,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.98915414966393e-05,
+      "loss": 0.5098,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9889467696920794e-05,
+      "loss": 0.54,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.988737431321814e-05,
+      "loss": 0.5008,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9885261347592576e-05,
+      "loss": 0.5048,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.98831288021246e-05,
+      "loss": 0.4464,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.988097667891402e-05,
+      "loss": 0.5201,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.987880498007989e-05,
+      "loss": 0.4896,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9876613707760564e-05,
+      "loss": 0.485,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.987440286411365e-05,
+      "loss": 0.5176,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.987217245131604e-05,
+      "loss": 0.5453,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.986992247156389e-05,
+      "loss": 0.5069,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.986765292707262e-05,
+      "loss": 0.4924,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9865363820076915e-05,
+      "loss": 0.5009,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.986305515283073e-05,
+      "loss": 0.5107,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9860726927607275e-05,
+      "loss": 0.5,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9858379146699005e-05,
+      "loss": 0.5104,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.985601181241764e-05,
+      "loss": 0.482,
+      "step": 219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.985362492709416e-05,
+      "loss": 0.5139,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9851218493078795e-05,
+      "loss": 0.4753,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.984879251274102e-05,
+      "loss": 0.4778,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9846346988469544e-05,
+      "loss": 0.4746,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.984388192267233e-05,
+      "loss": 0.4961,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.98413973177766e-05,
+      "loss": 0.5301,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.983889317622878e-05,
+      "loss": 0.5112,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.983636950049455e-05,
+      "loss": 0.5174,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.983382629305883e-05,
+      "loss": 0.4736,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9831263556425773e-05,
+      "loss": 0.5132,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.982868129311874e-05,
+      "loss": 0.542,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.982607950568035e-05,
+      "loss": 0.4821,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.982345819667242e-05,
+      "loss": 0.5101,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.982081736867599e-05,
+      "loss": 0.4818,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.981815702429134e-05,
+      "loss": 0.4766,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.981547716613795e-05,
+      "loss": 0.4624,
+      "step": 235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.981277779685452e-05,
+      "loss": 0.5072,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.981005891909895e-05,
+      "loss": 0.4922,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9807320535548374e-05,
+      "loss": 0.4944,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9804562648899104e-05,
+      "loss": 0.4927,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9801785261866674e-05,
+      "loss": 0.4526,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.979898837718581e-05,
+      "loss": 0.521,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.979617199761044e-05,
+      "loss": 0.5487,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9793336125913686e-05,
+      "loss": 0.4432,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9790480764887865e-05,
+      "loss": 0.6212,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.978760591734448e-05,
+      "loss": 0.495,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.978471158611423e-05,
+      "loss": 0.4975,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.978179777404699e-05,
+      "loss": 0.4964,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9778864484011805e-05,
+      "loss": 0.4449,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.977591171889693e-05,
+      "loss": 0.4935,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9772939481609774e-05,
+      "loss": 0.4512,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.976994777507692e-05,
+      "loss": 0.4765,
+      "step": 251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9766936602244124e-05,
+      "loss": 0.5095,
+      "step": 252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9763905966076314e-05,
+      "loss": 0.4575,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.976085586955758e-05,
+      "loss": 0.47,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.975778631569117e-05,
+      "loss": 0.4538,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9754697307499475e-05,
+      "loss": 0.4779,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.975158884802409e-05,
+      "loss": 0.4658,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.974846094032571e-05,
+      "loss": 0.4803,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9745313587484206e-05,
+      "loss": 0.4477,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9742146792598605e-05,
+      "loss": 0.4501,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9738960558787036e-05,
+      "loss": 0.5047,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.973575488918681e-05,
+      "loss": 0.5471,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.973252978695436e-05,
+      "loss": 0.514,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9729285255265256e-05,
+      "loss": 0.4774,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.972602129731419e-05,
+      "loss": 0.4855,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9722737916315e-05,
+      "loss": 0.4857,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.971943511550063e-05,
+      "loss": 0.4211,
+      "step": 267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9716112898123146e-05,
+      "loss": 0.4862,
+      "step": 268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.971277126745374e-05,
+      "loss": 0.4441,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9709410226782735e-05,
+      "loss": 0.4822,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.970602977941954e-05,
+      "loss": 0.5546,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.970262992869268e-05,
+      "loss": 0.4957,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.969921067794978e-05,
+      "loss": 0.507,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9695772030557585e-05,
+      "loss": 0.4626,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.969231398990192e-05,
+      "loss": 0.4711,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.968883655938773e-05,
+      "loss": 0.4595,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.968533974243902e-05,
+      "loss": 0.4803,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.96818235424989e-05,
+      "loss": 0.5217,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.967828796302956e-05,
+      "loss": 0.4403,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.967473300751229e-05,
+      "loss": 0.4494,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9671158679447444e-05,
+      "loss": 0.4628,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9667564982354444e-05,
+      "loss": 0.4317,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.966395191977179e-05,
+      "loss": 0.4417,
+      "step": 283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9660319495257056e-05,
+      "loss": 0.4385,
+      "step": 284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9656667712386875e-05,
+      "loss": 0.4403,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9652996574756934e-05,
+      "loss": 0.4062,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9649306085982006e-05,
+      "loss": 0.4374,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.964559624969588e-05,
+      "loss": 0.4045,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.964186706955141e-05,
+      "loss": 0.4522,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.963811854922051e-05,
+      "loss": 0.4589,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9634350692394124e-05,
+      "loss": 0.4246,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9630563502782235e-05,
+      "loss": 0.466,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.962675698411387e-05,
+      "loss": 0.4301,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.962293114013708e-05,
+      "loss": 0.4519,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9619085974618946e-05,
+      "loss": 0.4694,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.961522149134558e-05,
+      "loss": 0.4488,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.96113376941221e-05,
+      "loss": 0.4522,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9607434586772665e-05,
+      "loss": 0.4444,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9603512173140427e-05,
+      "loss": 0.4606,
+      "step": 299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.959957045708756e-05,
+      "loss": 0.4619,
+      "step": 300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.959560944249523e-05,
+      "loss": 0.4174,
+      "step": 301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.959162913326363e-05,
+      "loss": 0.4548,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9587629533311906e-05,
+      "loss": 0.4691,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.958361064657826e-05,
+      "loss": 0.4091,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.957957247701984e-05,
+      "loss": 0.4351,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.957551502861279e-05,
+      "loss": 0.452,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9571438305352236e-05,
+      "loss": 0.4558,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9567342311252304e-05,
+      "loss": 0.4846,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.956322705034606e-05,
+      "loss": 0.472,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9559092526685565e-05,
+      "loss": 0.4382,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.955493874434184e-05,
+      "loss": 0.4408,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.955076570740487e-05,
+      "loss": 0.4707,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.95465734199836e-05,
+      "loss": 0.504,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.954236188620592e-05,
+      "loss": 0.4669,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.953813111021868e-05,
+      "loss": 0.407,
+      "step": 315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.953388109618769e-05,
+      "loss": 0.4295,
+      "step": 316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.952961184829767e-05,
+      "loss": 0.4394,
+      "step": 317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.95253233707523e-05,
+      "loss": 0.4221,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.95210156677742e-05,
+      "loss": 0.4607,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.95166887436049e-05,
+      "loss": 0.4167,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.951234260250487e-05,
+      "loss": 0.4326,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.95079772487535e-05,
+      "loss": 0.455,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.950359268664909e-05,
+      "loss": 0.458,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.949918892050887e-05,
+      "loss": 0.4729,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9494765954668965e-05,
+      "loss": 0.4555,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9490323793484405e-05,
+      "loss": 0.4898,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.948586244132912e-05,
+      "loss": 0.4579,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.948138190259595e-05,
+      "loss": 0.4622,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.947688218169661e-05,
+      "loss": 0.4743,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.947236328306171e-05,
+      "loss": 0.4289,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9467825211140747e-05,
+      "loss": 0.4184,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.946326797040209e-05,
+      "loss": 0.44,
+      "step": 332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.945869156533298e-05,
+      "loss": 0.4593,
+      "step": 333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9454096000439544e-05,
+      "loss": 0.4505,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944948128024675e-05,
+      "loss": 0.4598,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9444847409298444e-05,
+      "loss": 0.5793,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944019439215733e-05,
+      "loss": 0.4865,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9435522233404956e-05,
+      "loss": 0.4282,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9430830937641726e-05,
+      "loss": 0.4598,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9426120509486875e-05,
+      "loss": 0.4103,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9421390953578485e-05,
+      "loss": 0.4805,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.941664227457347e-05,
+      "loss": 0.4349,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9411874477147574e-05,
+      "loss": 0.4288,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.940708756599537e-05,
+      "loss": 0.4267,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.940228154583024e-05,
+      "loss": 0.4211,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.939745642138439e-05,
+      "loss": 0.4837,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.939261219740885e-05,
+      "loss": 0.4615,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9387748878673416e-05,
+      "loss": 0.4713,
+      "step": 348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9382866469966716e-05,
+      "loss": 0.44,
+      "step": 349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.937796497609618e-05,
+      "loss": 0.4098,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.937304440188802e-05,
+      "loss": 0.5474,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.936810475218723e-05,
+      "loss": 0.4519,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9363146031857595e-05,
+      "loss": 0.446,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9358168245781664e-05,
+      "loss": 0.4371,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9353171398860776e-05,
+      "loss": 0.4565,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9348155496015034e-05,
+      "loss": 0.45,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.93431205421833e-05,
+      "loss": 0.429,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.933806654232319e-05,
+      "loss": 0.4422,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.933299350141107e-05,
+      "loss": 0.4388,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9327901424442094e-05,
+      "loss": 0.4689,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9322790316430105e-05,
+      "loss": 0.4308,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.931766018240772e-05,
+      "loss": 0.4011,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.931251102742627e-05,
+      "loss": 0.4532,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.930734285655583e-05,
+      "loss": 0.4114,
+      "step": 364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.93021556748852e-05,
+      "loss": 0.4602,
+      "step": 365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9296949487521875e-05,
+      "loss": 0.4243,
+      "step": 366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9291724299592085e-05,
+      "loss": 0.4568,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9286480116240773e-05,
+      "loss": 0.4454,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.928121694263157e-05,
+      "loss": 0.4483,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.927593478394681e-05,
+      "loss": 0.4406,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9270633645387525e-05,
+      "loss": 0.4464,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.926531353217344e-05,
+      "loss": 0.3999,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9259974449542946e-05,
+      "loss": 0.4878,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.925461640275313e-05,
+      "loss": 0.4244,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9249239397079725e-05,
+      "loss": 0.4194,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9243843437817185e-05,
+      "loss": 0.419,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.923842853027856e-05,
+      "loss": 0.4457,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9232994679795615e-05,
+      "loss": 0.4606,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.922754189171873e-05,
+      "loss": 0.4543,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9222070171416934e-05,
+      "loss": 0.4217,
+      "step": 380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9216579524277916e-05,
+      "loss": 0.3893,
+      "step": 381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9211069955708e-05,
+      "loss": 0.4804,
+      "step": 382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.920554147113212e-05,
+      "loss": 0.4359,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.919999407599386e-05,
+      "loss": 0.4188,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.91944277757554e-05,
+      "loss": 0.4421,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.918884257589755e-05,
+      "loss": 0.4864,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.918323848191973e-05,
+      "loss": 0.4012,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9177615499339966e-05,
+      "loss": 0.4442,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.917197363369487e-05,
+      "loss": 0.448,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.916631289053965e-05,
+      "loss": 0.4974,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9160633275448104e-05,
+      "loss": 0.4469,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.915493479401263e-05,
+      "loss": 0.3937,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.914921745184417e-05,
+      "loss": 0.422,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.914348125457226e-05,
+      "loss": 0.4526,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.913772620784499e-05,
+      "loss": 0.4061,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9131952317329014e-05,
+      "loss": 0.4043,
+      "step": 396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9126159588709555e-05,
+      "loss": 0.4449,
+      "step": 397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.912034802769036e-05,
+      "loss": 0.4545,
+      "step": 398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.911451763999372e-05,
+      "loss": 0.4242,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.910866843136049e-05,
+      "loss": 0.4095,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.910280040755003e-05,
+      "loss": 0.3849,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9096913574340255e-05,
+      "loss": 0.4098,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.909100793752756e-05,
+      "loss": 0.4341,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.908508350292687e-05,
+      "loss": 0.4612,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.907914027637166e-05,
+      "loss": 0.4024,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.907317826371384e-05,
+      "loss": 0.4526,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.906719747082387e-05,
+      "loss": 0.4056,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.906119790359069e-05,
+      "loss": 0.4487,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9055179567921694e-05,
+      "loss": 0.4093,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.90491424697428e-05,
+      "loss": 0.4273,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.904308661499839e-05,
+      "loss": 0.4021,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.903701200965128e-05,
+      "loss": 0.3952,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.903091865968281e-05,
+      "loss": 0.3855,
+      "step": 413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.90248065710927e-05,
+      "loss": 0.4546,
+      "step": 414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.90186757498992e-05,
+      "loss": 0.4129,
+      "step": 415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.901252620213894e-05,
+      "loss": 0.3878,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.900635793386704e-05,
+      "loss": 0.4575,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.900017095115702e-05,
+      "loss": 0.4035,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8993965260100823e-05,
+      "loss": 0.4575,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8987740866808845e-05,
+      "loss": 0.4316,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.898149777740987e-05,
+      "loss": 0.4428,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8975235998051104e-05,
+      "loss": 0.4807,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.896895553489815e-05,
+      "loss": 0.4073,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8962656394135007e-05,
+      "loss": 0.4156,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.895633858196406e-05,
+      "loss": 0.4175,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.89500021046061e-05,
+      "loss": 0.4196,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.894364696830027e-05,
+      "loss": 0.3797,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8937273179304105e-05,
+      "loss": 0.4426,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8930880743893495e-05,
+      "loss": 0.4353,
+      "step": 429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8924469668362696e-05,
+      "loss": 0.3982,
+      "step": 430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.891803995902432e-05,
+      "loss": 0.4069,
+      "step": 431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8911591622209306e-05,
+      "loss": 0.3783,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.890512466426697e-05,
+      "loss": 0.3987,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8898639091564926e-05,
+      "loss": 0.4217,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8892134910489145e-05,
+      "loss": 0.4635,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8885612127443905e-05,
+      "loss": 0.3965,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.887907074885181e-05,
+      "loss": 0.3822,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8872510781153765e-05,
+      "loss": 0.4433,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.886593223080899e-05,
+      "loss": 0.4145,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.885933510429499e-05,
+      "loss": 0.4525,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8852719408107555e-05,
+      "loss": 0.4272,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8846085148760784e-05,
+      "loss": 0.3669,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.883943233278704e-05,
+      "loss": 0.386,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8832760966736954e-05,
+      "loss": 0.3754,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.882607105717943e-05,
+      "loss": 0.451,
+      "step": 445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.881936261070162e-05,
+      "loss": 0.4189,
+      "step": 446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.881263563390894e-05,
+      "loss": 0.4022,
+      "step": 447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.880589013342505e-05,
+      "loss": 0.3959,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.879912611589183e-05,
+      "loss": 0.5004,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.879234358796943e-05,
+      "loss": 0.3858,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.87855425563362e-05,
+      "loss": 0.4193,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.87787230276887e-05,
+      "loss": 0.4212,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.877188500874172e-05,
+      "loss": 0.4491,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.876502850622827e-05,
+      "loss": 0.3955,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.875815352689953e-05,
+      "loss": 0.3954,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.875126007752489e-05,
+      "loss": 0.4054,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.874434816489192e-05,
+      "loss": 0.4056,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.873741779580637e-05,
+      "loss": 0.3841,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.873046897709217e-05,
+      "loss": 0.4867,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.872350171559142e-05,
+      "loss": 0.421,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.871651601816435e-05,
+      "loss": 0.4229,
+      "step": 461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8709511891689374e-05,
+      "loss": 0.3954,
+      "step": 462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.870248934306304e-05,
+      "loss": 0.4064,
+      "step": 463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.869544837920005e-05,
+      "loss": 0.3855,
+      "step": 464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.86883890070332e-05,
+      "loss": 0.3888,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.868131123351346e-05,
+      "loss": 0.4382,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.867421506560989e-05,
+      "loss": 0.418,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.866710051030966e-05,
+      "loss": 0.4318,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.865996757461805e-05,
+      "loss": 0.385,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.865281626555845e-05,
+      "loss": 0.3874,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8645646590172334e-05,
+      "loss": 0.426,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.863845855551925e-05,
+      "loss": 0.3958,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.863125216867683e-05,
+      "loss": 0.4259,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.862402743674079e-05,
+      "loss": 0.4184,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.861678436682488e-05,
+      "loss": 0.4107,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8609522966060935e-05,
+      "loss": 0.4267,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.860224324159882e-05,
+      "loss": 0.3894,
+      "step": 477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.859494520060646e-05,
+      "loss": 0.4144,
+      "step": 478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.858762885026979e-05,
+      "loss": 0.3851,
+      "step": 479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.85802941977928e-05,
+      "loss": 0.4,
+      "step": 480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.857294125039749e-05,
+      "loss": 0.3759,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.856557001532386e-05,
+      "loss": 0.3816,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8558180499829946e-05,
+      "loss": 0.3844,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.855077271119175e-05,
+      "loss": 0.3578,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.85433466567033e-05,
+      "loss": 0.4172,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8535902343676595e-05,
+      "loss": 0.3815,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.852843977944159e-05,
+      "loss": 0.3865,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.852095897134625e-05,
+      "loss": 0.3974,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.851345992675648e-05,
+      "loss": 0.412,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.850594265305614e-05,
+      "loss": 0.4282,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.849840715764705e-05,
+      "loss": 0.4101,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8490853447948975e-05,
+      "loss": 0.4271,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.848328153139959e-05,
+      "loss": 0.3949,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.847569141545453e-05,
+      "loss": 0.453,
+      "step": 494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8468083107587325e-05,
+      "loss": 0.4288,
+      "step": 495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.846045661528943e-05,
+      "loss": 0.347,
+      "step": 496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.84528119460702e-05,
+      "loss": 0.5167,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8445149107456886e-05,
+      "loss": 0.4333,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.843746810699464e-05,
+      "loss": 0.4458,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.842976895224649e-05,
+      "loss": 0.4397,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.842205165079333e-05,
+      "loss": 0.4042,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.841431621023393e-05,
+      "loss": 0.3641,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.840656263818493e-05,
+      "loss": 0.4027,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8398790942280815e-05,
+      "loss": 0.3808,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.839100113017391e-05,
+      "loss": 0.3995,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.838319320953438e-05,
+      "loss": 0.3919,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.837536718805023e-05,
+      "loss": 0.3865,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8367523073427285e-05,
+      "loss": 0.3611,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8359660873389165e-05,
+      "loss": 0.4312,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.835178059567733e-05,
+      "loss": 0.3542,
+      "step": 510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.834388224805102e-05,
+      "loss": 0.4264,
+      "step": 511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.833596583828727e-05,
+      "loss": 0.3823,
+      "step": 512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.832803137418089e-05,
+      "loss": 0.4286,
+      "step": 513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.83200788635445e-05,
+      "loss": 0.4208,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.831210831420846e-05,
+      "loss": 0.4064,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8304119734020887e-05,
+      "loss": 0.3983,
+      "step": 516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8296113130847674e-05,
+      "loss": 0.3993,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.828808851257245e-05,
+      "loss": 0.3574,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8280045887096585e-05,
+      "loss": 0.4289,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.827198526233916e-05,
+      "loss": 0.4676,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.826390664623703e-05,
+      "loss": 0.3665,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8255810046744694e-05,
+      "loss": 0.4244,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.824769547183443e-05,
+      "loss": 0.3857,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.823956292949615e-05,
+      "loss": 0.3975,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.823141242773751e-05,
+      "loss": 0.3371,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.822324397458381e-05,
+      "loss": 0.362,
+      "step": 526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.821505757807806e-05,
+      "loss": 0.414,
+      "step": 527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.820685324628091e-05,
+      "loss": 0.4218,
+      "step": 528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.819863098727069e-05,
+      "loss": 0.3851,
+      "step": 529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.819039080914336e-05,
+      "loss": 0.4066,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8182132720012546e-05,
+      "loss": 0.382,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.81738567280095e-05,
+      "loss": 0.4282,
+      "step": 532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.816556284128311e-05,
+      "loss": 0.3519,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.815725106799985e-05,
+      "loss": 0.38,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.814892141634387e-05,
+      "loss": 0.3806,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8140573894516853e-05,
+      "loss": 0.4197,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.813220851073812e-05,
+      "loss": 0.4414,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8123825273244586e-05,
+      "loss": 0.3928,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.811542419029071e-05,
+      "loss": 0.3935,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8107005270148555e-05,
+      "loss": 0.6841,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.809856852110772e-05,
+      "loss": 0.5035,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8090113951475395e-05,
+      "loss": 0.4135,
+      "step": 542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8081641569576284e-05,
+      "loss": 0.3698,
+      "step": 543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.807315138375263e-05,
+      "loss": 0.3964,
+      "step": 544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.806464340236424e-05,
+      "loss": 0.388,
+      "step": 545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8056117633788394e-05,
+      "loss": 0.3813,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.804757408641994e-05,
+      "loss": 0.3939,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8039012768671184e-05,
+      "loss": 0.3354,
+      "step": 548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.803043368897196e-05,
+      "loss": 0.3841,
+      "step": 549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8021836855769585e-05,
+      "loss": 0.3395,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8013222277528845e-05,
+      "loss": 0.3712,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8004589962732005e-05,
+      "loss": 0.3619,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7995939919878804e-05,
+      "loss": 0.3435,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.798727215748642e-05,
+      "loss": 0.3787,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.797858668408949e-05,
+      "loss": 0.3567,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.796988350824009e-05,
+      "loss": 0.3798,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7961162638507714e-05,
+      "loss": 0.4257,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7952424083479297e-05,
+      "loss": 0.3667,
+      "step": 558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.794366785175917e-05,
+      "loss": 0.4024,
+      "step": 559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.793489395196909e-05,
+      "loss": 0.3862,
+      "step": 560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7926102392748186e-05,
+      "loss": 0.3991,
+      "step": 561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.791729318275299e-05,
+      "loss": 0.3545,
+      "step": 562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.790846633065741e-05,
+      "loss": 0.389,
+      "step": 563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.789962184515273e-05,
+      "loss": 0.4698,
+      "step": 564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7890759734947586e-05,
+      "loss": 0.3829,
+      "step": 565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7881880008767976e-05,
+      "loss": 0.3959,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.787298267535725e-05,
+      "loss": 0.3296,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.786406774347608e-05,
+      "loss": 0.4067,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.785513522190248e-05,
+      "loss": 0.3382,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.784618511943176e-05,
+      "loss": 0.3823,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7837217444876575e-05,
+      "loss": 0.3813,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.782823220706685e-05,
+      "loss": 0.3867,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7819229414849834e-05,
+      "loss": 0.3959,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.781020907709003e-05,
+      "loss": 0.3444,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.780117120266924e-05,
+      "loss": 0.3588,
+      "step": 575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.779211580048652e-05,
+      "loss": 0.4122,
+      "step": 576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.77830428794582e-05,
+      "loss": 0.3815,
+      "step": 577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7773952448517833e-05,
+      "loss": 0.3776,
+      "step": 578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7764844516616245e-05,
+      "loss": 0.3368,
+      "step": 579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.775571909272147e-05,
+      "loss": 0.4108,
+      "step": 580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.774657618581878e-05,
+      "loss": 0.3458,
+      "step": 581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7737415804910646e-05,
+      "loss": 0.3678,
+      "step": 582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.772823795901676e-05,
+      "loss": 0.3948,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7719042657174006e-05,
+      "loss": 0.3772,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.770982990843645e-05,
+      "loss": 0.3872,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.770059972187535e-05,
+      "loss": 0.3655,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.769135210657912e-05,
+      "loss": 0.3722,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.768208707165334e-05,
+      "loss": 0.4063,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7672804626220735e-05,
+      "loss": 0.3469,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.76635047794212e-05,
+      "loss": 0.3877,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.765418754041173e-05,
+      "loss": 0.4154,
+      "step": 591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.764485291836646e-05,
+      "loss": 0.3471,
+      "step": 592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7635500922476656e-05,
+      "loss": 0.3334,
+      "step": 593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.762613156195066e-05,
+      "loss": 0.3553,
+      "step": 594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7616744846013935e-05,
+      "loss": 0.3564,
+      "step": 595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7607340783909014e-05,
+      "loss": 0.3586,
+      "step": 596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7597919384895546e-05,
+      "loss": 0.3664,
+      "step": 597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.75884806582502e-05,
+      "loss": 0.3661,
+      "step": 598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.757902461326675e-05,
+      "loss": 0.3618,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7569551259255995e-05,
+      "loss": 0.3617,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7560060605545796e-05,
+      "loss": 0.3623,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.755055266148104e-05,
+      "loss": 0.4015,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7541027436423615e-05,
+      "loss": 0.3996,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.753148493975248e-05,
+      "loss": 0.3684,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7521925180863556e-05,
+      "loss": 0.3745,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.751234816916976e-05,
+      "loss": 0.3704,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7502753914101024e-05,
+      "loss": 0.3587,
+      "step": 607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.749314242510424e-05,
+      "loss": 0.3912,
+      "step": 608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.748351371164327e-05,
+      "loss": 0.3861,
+      "step": 609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.747386778319895e-05,
+      "loss": 0.3601,
+      "step": 610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.746420464926904e-05,
+      "loss": 0.3638,
+      "step": 611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7454524319368275e-05,
+      "loss": 0.355,
+      "step": 612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7444826803028294e-05,
+      "loss": 0.3966,
+      "step": 613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7435112109797665e-05,
+      "loss": 0.4125,
+      "step": 614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.742538024924188e-05,
+      "loss": 0.3606,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.741563123094333e-05,
+      "loss": 0.4584,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7405865064501286e-05,
+      "loss": 0.393,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.739608175953193e-05,
+      "loss": 0.3676,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.738628132566829e-05,
+      "loss": 0.4078,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.737646377256028e-05,
+      "loss": 0.4119,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.736662910987467e-05,
+      "loss": 0.3501,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.735677734729507e-05,
+      "loss": 0.4248,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.734690849452192e-05,
+      "loss": 0.3761,
+      "step": 623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.73370225612725e-05,
+      "loss": 0.4011,
+      "step": 624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7327119557280915e-05,
+      "loss": 0.3564,
+      "step": 625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.731719949229806e-05,
+      "loss": 0.387,
+      "step": 626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.730726237609164e-05,
+      "loss": 0.3521,
+      "step": 627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.729730821844613e-05,
+      "loss": 0.3605,
+      "step": 628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.728733702916282e-05,
+      "loss": 0.363,
+      "step": 629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.727734881805975e-05,
+      "loss": 0.3244,
+      "step": 630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7267343594971716e-05,
+      "loss": 0.3896,
+      "step": 631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.725732136975027e-05,
+      "loss": 0.3685,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.724728215226371e-05,
+      "loss": 0.3597,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.723722595239706e-05,
+      "loss": 0.3212,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.722715278005206e-05,
+      "loss": 0.3297,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7217062645147173e-05,
+      "loss": 0.3734,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.720695555761755e-05,
+      "loss": 0.3337,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.719683152741506e-05,
+      "loss": 0.3347,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7186690564508225e-05,
+      "loss": 0.3483,
+      "step": 639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.717653267888226e-05,
+      "loss": 0.3521,
+      "step": 640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.716635788053903e-05,
+      "loss": 0.3129,
+      "step": 641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.715616617949705e-05,
+      "loss": 0.3788,
+      "step": 642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7145957585791495e-05,
+      "loss": 0.3945,
+      "step": 643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.713573210947417e-05,
+      "loss": 0.4618,
+      "step": 644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.712548976061349e-05,
+      "loss": 0.3578,
+      "step": 645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.711523054929449e-05,
+      "loss": 0.3981,
+      "step": 646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7104954485618814e-05,
+      "loss": 0.3153,
+      "step": 647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7094661579704696e-05,
+      "loss": 0.3744,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.708435184168694e-05,
+      "loss": 0.3647,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.707402528171696e-05,
+      "loss": 0.3481,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.706368190996268e-05,
+      "loss": 0.3099,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7053321736608634e-05,
+      "loss": 0.3474,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.704294477185586e-05,
+      "loss": 0.3606,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.703255102592193e-05,
+      "loss": 0.4674,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7022140509040986e-05,
+      "loss": 0.3619,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.701171323146362e-05,
+      "loss": 0.3951,
+      "step": 656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.700126920345696e-05,
+      "loss": 0.3801,
+      "step": 657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.699080843530463e-05,
+      "loss": 0.3146,
+      "step": 658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.698033093730674e-05,
+      "loss": 0.3363,
+      "step": 659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.696983671977984e-05,
+      "loss": 0.3405,
+      "step": 660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.695932579305698e-05,
+      "loss": 0.3206,
+      "step": 661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.694879816748765e-05,
+      "loss": 0.3739,
+      "step": 662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.693825385343778e-05,
+      "loss": 0.3475,
+      "step": 663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.692769286128973e-05,
+      "loss": 0.3338,
+      "step": 664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.691711520144229e-05,
+      "loss": 0.3539,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6906520884310655e-05,
+      "loss": 0.3703,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.689590992032642e-05,
+      "loss": 0.3646,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.688528231993758e-05,
+      "loss": 0.3626,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.687463809360849e-05,
+      "loss": 0.3274,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6863977251819916e-05,
+      "loss": 0.3646,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.685329980506894e-05,
+      "loss": 0.3919,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.684260576386903e-05,
+      "loss": 0.3795,
+      "step": 672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6831895138749955e-05,
+      "loss": 0.379,
+      "step": 673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.682116794025785e-05,
+      "loss": 0.3657,
+      "step": 674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.681042417895515e-05,
+      "loss": 0.3527,
+      "step": 675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.679966386542061e-05,
+      "loss": 0.3376,
+      "step": 676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.678888701024926e-05,
+      "loss": 0.3635,
+      "step": 677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.677809362405244e-05,
+      "loss": 0.3524,
+      "step": 678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.676728371745777e-05,
+      "loss": 0.3901,
+      "step": 679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.675645730110911e-05,
+      "loss": 0.382,
+      "step": 680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.67456143856666e-05,
+      "loss": 0.3679,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6734754981806616e-05,
+      "loss": 0.3526,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.672387910022178e-05,
+      "loss": 0.3672,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6712986751620914e-05,
+      "loss": 0.3946,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6702077946729084e-05,
+      "loss": 0.347,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.669115269628755e-05,
+      "loss": 0.305,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.668021101105375e-05,
+      "loss": 0.3724,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.666925290180132e-05,
+      "loss": 0.3375,
+      "step": 688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.665827837932006e-05,
+      "loss": 0.3353,
+      "step": 689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.664728745441593e-05,
+      "loss": 0.3345,
+      "step": 690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.663628013791106e-05,
+      "loss": 0.3311,
+      "step": 691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6625256440643686e-05,
+      "loss": 0.3193,
+      "step": 692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.661421637346821e-05,
+      "loss": 0.4838,
+      "step": 693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.6603159947255115e-05,
+      "loss": 0.3579,
+      "step": 694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.659208717289102e-05,
+      "loss": 0.3978,
+      "step": 695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.658099806127864e-05,
+      "loss": 0.3418,
+      "step": 696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.656989262333675e-05,
+      "loss": 0.3905,
+      "step": 697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.655877087000024e-05,
+      "loss": 0.3739,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.654763281222003e-05,
+      "loss": 0.3558,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.653647846096311e-05,
+      "loss": 0.3072,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.652530782721251e-05,
+      "loss": 0.3973,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.65141209219673e-05,
+      "loss": 0.3191,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.650291775624255e-05,
+      "loss": 0.3364,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.649169834106938e-05,
+      "loss": 0.3491,
+      "step": 704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.648046268749486e-05,
+      "loss": 0.3319,
+      "step": 705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.646921080658209e-05,
+      "loss": 0.3769,
+      "step": 706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.645794270941013e-05,
+      "loss": 0.3269,
+      "step": 707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.6446658407073995e-05,
+      "loss": 0.3692,
+      "step": 708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.64353579106847e-05,
+      "loss": 0.3684,
+      "step": 709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.6424041231369144e-05,
+      "loss": 0.3453,
+      "step": 710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.641270838027022e-05,
+      "loss": 0.3672,
+      "step": 711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6401359368546707e-05,
+      "loss": 0.3476,
+      "step": 712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6389994207373295e-05,
+      "loss": 0.3056,
+      "step": 713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.637861290794061e-05,
+      "loss": 0.3358,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.636721548145513e-05,
+      "loss": 0.3807,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6355801939139226e-05,
+      "loss": 0.3997,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6344372292231145e-05,
+      "loss": 0.3353,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.633292655198498e-05,
+      "loss": 0.3553,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.632146472967069e-05,
+      "loss": 0.3678,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.630998683657404e-05,
+      "loss": 0.3598,
+      "step": 720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6298492883996635e-05,
+      "loss": 0.3248,
+      "step": 721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6286982883255895e-05,
+      "loss": 0.3208,
+      "step": 722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.627545684568503e-05,
+      "loss": 0.3575,
+      "step": 723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.626391478263307e-05,
+      "loss": 0.3468,
+      "step": 724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.625235670546477e-05,
+      "loss": 0.331,
+      "step": 725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.624078262556071e-05,
+      "loss": 0.3595,
+      "step": 726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.62291925543172e-05,
+      "loss": 0.3129,
+      "step": 727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.62175865031463e-05,
+      "loss": 0.3376,
+      "step": 728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.620596448347579e-05,
+      "loss": 0.3248,
+      "step": 729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6194326506749195e-05,
+      "loss": 0.2983,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.618267258442575e-05,
+      "loss": 0.3561,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.617100272798037e-05,
+      "loss": 0.3032,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6159316948903684e-05,
+      "loss": 0.3529,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.614761525870198e-05,
+      "loss": 0.3812,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6135897668897235e-05,
+      "loss": 0.3527,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6124164191027055e-05,
+      "loss": 0.3347,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.61124148366447e-05,
+      "loss": 0.3412,
+      "step": 737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6100649617319074e-05,
+      "loss": 0.346,
+      "step": 738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.60888685446347e-05,
+      "loss": 0.3533,
+      "step": 739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.607707163019169e-05,
+      "loss": 0.3107,
+      "step": 740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.606525888560577e-05,
+      "loss": 0.3332,
+      "step": 741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.605343032250827e-05,
+      "loss": 0.3617,
+      "step": 742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.604158595254605e-05,
+      "loss": 0.3294,
+      "step": 743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.602972578738159e-05,
+      "loss": 0.2963,
+      "step": 744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.6017849838692864e-05,
+      "loss": 0.3752,
+      "step": 745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.600595811817345e-05,
+      "loss": 0.335,
+      "step": 746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.599405063753239e-05,
+      "loss": 0.3465,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.59821274084943e-05,
+      "loss": 0.3226,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.597018844279928e-05,
+      "loss": 0.3485,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.59582337522029e-05,
+      "loss": 0.3364,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5946263348476263e-05,
+      "loss": 0.3007,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.593427724340592e-05,
+      "loss": 0.4194,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.592227544879385e-05,
+      "loss": 0.3567,
+      "step": 753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5910257976457545e-05,
+      "loss": 0.3661,
+      "step": 754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.589822483822989e-05,
+      "loss": 0.3928,
+      "step": 755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.58861760459592e-05,
+      "loss": 0.3231,
+      "step": 756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.587411161150921e-05,
+      "loss": 0.3341,
+      "step": 757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.586203154675906e-05,
+      "loss": 0.3399,
+      "step": 758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5849935863603276e-05,
+      "loss": 0.3442,
+      "step": 759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.583782457395176e-05,
+      "loss": 0.3374,
+      "step": 760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5825697689729784e-05,
+      "loss": 0.341,
+      "step": 761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.581355522287797e-05,
+      "loss": 0.3408,
+      "step": 762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.580139718535231e-05,
+      "loss": 0.3436,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.578922358912407e-05,
+      "loss": 0.3554,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5777034446179904e-05,
+      "loss": 0.3161,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.576482976852172e-05,
+      "loss": 0.3615,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5752609568166755e-05,
+      "loss": 0.3636,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.574037385714751e-05,
+      "loss": 0.3267,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.572812264751177e-05,
+      "loss": 0.3359,
+      "step": 769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.571585595132258e-05,
+      "loss": 0.3619,
+      "step": 770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.570357378065824e-05,
+      "loss": 0.3436,
+      "step": 771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.569127614761225e-05,
+      "loss": 0.321,
+      "step": 772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.56789630642934e-05,
+      "loss": 0.3533,
+      "step": 773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5666634542825626e-05,
+      "loss": 0.2974,
+      "step": 774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5654290595348106e-05,
+      "loss": 0.3041,
+      "step": 775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.56419312340152e-05,
+      "loss": 0.3007,
+      "step": 776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.562955647099643e-05,
+      "loss": 0.304,
+      "step": 777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.561716631847651e-05,
+      "loss": 0.3597,
+      "step": 778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.560476078865528e-05,
+      "loss": 0.3251,
+      "step": 779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.559233989374773e-05,
+      "loss": 0.3437,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.557990364598399e-05,
+      "loss": 0.3417,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5567452057609296e-05,
+      "loss": 0.3333,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5554985140883995e-05,
+      "loss": 0.3363,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.554250290808352e-05,
+      "loss": 0.3405,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.553000537149838e-05,
+      "loss": 0.3371,
+      "step": 785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.551749254343418e-05,
+      "loss": 0.3199,
+      "step": 786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.550496443621154e-05,
+      "loss": 0.2998,
+      "step": 787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5492421062166164e-05,
+      "loss": 0.3205,
+      "step": 788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.547986243364878e-05,
+      "loss": 0.3873,
+      "step": 789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.54672885630251e-05,
+      "loss": 0.3399,
+      "step": 790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5454699462675886e-05,
+      "loss": 0.2651,
+      "step": 791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.5442095144996886e-05,
+      "loss": 0.2968,
+      "step": 792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.542947562239883e-05,
+      "loss": 0.3215,
+      "step": 793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.5416840907307404e-05,
+      "loss": 0.313,
+      "step": 794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.540419101216327e-05,
+      "loss": 0.2994,
+      "step": 795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.5391525949422036e-05,
+      "loss": 0.3352,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.537884573155424e-05,
+      "loss": 0.3147,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.536615037104534e-05,
+      "loss": 0.2994,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.535343988039572e-05,
+      "loss": 0.3014,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.5340714272120634e-05,
+      "loss": 0.3277,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.532797355875024e-05,
+      "loss": 0.3054,
+      "step": 801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.531521775282957e-05,
+      "loss": 0.3352,
+      "step": 802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.530244686691852e-05,
+      "loss": 0.333,
+      "step": 803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.528966091359181e-05,
+      "loss": 0.3088,
+      "step": 804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.527685990543903e-05,
+      "loss": 0.3259,
+      "step": 805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.5264043855064565e-05,
+      "loss": 0.4465,
+      "step": 806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.525121277508763e-05,
+      "loss": 0.2879,
+      "step": 807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.523836667814222e-05,
+      "loss": 0.4248,
+      "step": 808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.522550557687714e-05,
+      "loss": 0.3281,
+      "step": 809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5212629483955956e-05,
+      "loss": 0.3131,
+      "step": 810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5199738412056984e-05,
+      "loss": 0.3467,
+      "step": 811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5186832373873324e-05,
+      "loss": 0.3211,
+      "step": 812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.517391138211277e-05,
+      "loss": 0.3187,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5160975449497863e-05,
+      "loss": 0.3963,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5148024588765854e-05,
+      "loss": 0.2819,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.51350588126687e-05,
+      "loss": 0.356,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.512207813397302e-05,
+      "loss": 0.327,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5109082565460124e-05,
+      "loss": 0.3031,
+      "step": 818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.509607211992599e-05,
+      "loss": 0.2933,
+      "step": 819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.508304681018122e-05,
+      "loss": 0.3343,
+      "step": 820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.507000664905109e-05,
+      "loss": 0.3572,
+      "step": 821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.505695164937546e-05,
+      "loss": 0.2755,
+      "step": 822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5043881824008816e-05,
+      "loss": 0.3037,
+      "step": 823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.503079718582025e-05,
+      "loss": 0.3997,
+      "step": 824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.5017697747693425e-05,
+      "loss": 0.3166,
+      "step": 825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.5004583522526596e-05,
+      "loss": 0.3137,
+      "step": 826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.499145452323256e-05,
+      "loss": 0.39,
+      "step": 827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4978310762738665e-05,
+      "loss": 0.3372,
+      "step": 828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.496515225398679e-05,
+      "loss": 0.3467,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4951979009933355e-05,
+      "loss": 0.3105,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4938791043549264e-05,
+      "loss": 0.3793,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.492558836781994e-05,
+      "loss": 0.2849,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.491237099574527e-05,
+      "loss": 0.2897,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.489913894033962e-05,
+      "loss": 0.3286,
+      "step": 834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.488589221463181e-05,
+      "loss": 0.3267,
+      "step": 835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.487263083166512e-05,
+      "loss": 0.3317,
+      "step": 836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4859354804497244e-05,
+      "loss": 0.284,
+      "step": 837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.48460641462003e-05,
+      "loss": 0.3069,
+      "step": 838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4832758869860805e-05,
+      "loss": 0.3072,
+      "step": 839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.48194389885797e-05,
+      "loss": 0.3603,
+      "step": 840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.480610451547228e-05,
+      "loss": 0.3002,
+      "step": 841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4792755463668194e-05,
+      "loss": 0.316,
+      "step": 842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.477939184631149e-05,
+      "loss": 0.3455,
+      "step": 843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4766013676560514e-05,
+      "loss": 0.2816,
+      "step": 844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.475262096758797e-05,
+      "loss": 0.3374,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.473921373258087e-05,
+      "loss": 0.3434,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4725791984740515e-05,
+      "loss": 0.3284,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4712355737282516e-05,
+      "loss": 0.3376,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4698905003436746e-05,
+      "loss": 0.2761,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.468543979644736e-05,
+      "loss": 0.3086,
+      "step": 850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4671960129572744e-05,
+      "loss": 0.2954,
+      "step": 851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4658466016085515e-05,
+      "loss": 0.3166,
+      "step": 852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.464495746927256e-05,
+      "loss": 0.3136,
+      "step": 853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.463143450243493e-05,
+      "loss": 0.3103,
+      "step": 854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4617897128887885e-05,
+      "loss": 0.3895,
+      "step": 855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4604345361960895e-05,
+      "loss": 0.3051,
+      "step": 856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.459077921499757e-05,
+      "loss": 0.326,
+      "step": 857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.457719870135571e-05,
+      "loss": 0.3126,
+      "step": 858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.456360383440722e-05,
+      "loss": 0.374,
+      "step": 859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.454999462753818e-05,
+      "loss": 0.3398,
+      "step": 860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4536371094148773e-05,
+      "loss": 0.2941,
+      "step": 861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4522733247653275e-05,
+      "loss": 0.3393,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.450908110148008e-05,
+      "loss": 0.3255,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.449541466907165e-05,
+      "loss": 0.3269,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.448173396388451e-05,
+      "loss": 0.3271,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.446803899938924e-05,
+      "loss": 0.3068,
+      "step": 866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.445432978907047e-05,
+      "loss": 0.2862,
+      "step": 867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.444060634642683e-05,
+      "loss": 0.411,
+      "step": 868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4426868684971014e-05,
+      "loss": 0.3216,
+      "step": 869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.441311681822967e-05,
+      "loss": 0.2935,
+      "step": 870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4399350759743435e-05,
+      "loss": 0.2972,
+      "step": 871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.438557052306696e-05,
+      "loss": 0.3054,
+      "step": 872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4371776121768816e-05,
+      "loss": 0.3079,
+      "step": 873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4357967569431534e-05,
+      "loss": 0.3313,
+      "step": 874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4344144879651585e-05,
+      "loss": 0.2891,
+      "step": 875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.433030806603936e-05,
+      "loss": 0.3355,
+      "step": 876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4316457142219134e-05,
+      "loss": 0.3629,
+      "step": 877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.430259212182911e-05,
+      "loss": 0.3323,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4288713018521356e-05,
+      "loss": 0.3252,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4274819845961794e-05,
+      "loss": 0.3036,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.426091261783022e-05,
+      "loss": 0.3259,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.424699134782025e-05,
+      "loss": 0.3307,
+      "step": 882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4233056049639345e-05,
+      "loss": 0.3486,
+      "step": 883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.421910673700877e-05,
+      "loss": 0.3184,
+      "step": 884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.420514342366358e-05,
+      "loss": 0.328,
+      "step": 885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4191166123352635e-05,
+      "loss": 0.3609,
+      "step": 886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4177174849838555e-05,
+      "loss": 0.2949,
+      "step": 887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.416316961689772e-05,
+      "loss": 0.3194,
+      "step": 888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.414915043832025e-05,
+      "loss": 0.3431,
+      "step": 889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.413511732791001e-05,
+      "loss": 0.2893,
+      "step": 890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.412107029948458e-05,
+      "loss": 0.3186,
+      "step": 891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.4107009366875226e-05,
+      "loss": 0.3345,
+      "step": 892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.409293454392692e-05,
+      "loss": 0.2918,
+      "step": 893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.407884584449831e-05,
+      "loss": 0.3407,
+      "step": 894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.406474328246172e-05,
+      "loss": 0.3387,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.4050626871703094e-05,
+      "loss": 0.2837,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.403649662612203e-05,
+      "loss": 0.3018,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.402235255963175e-05,
+      "loss": 0.3266,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.400819468615908e-05,
+      "loss": 0.3577,
+      "step": 899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.3994023019644445e-05,
+      "loss": 0.334,
+      "step": 900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.3979837574041835e-05,
+      "loss": 0.3125,
+      "step": 901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.3965638363318834e-05,
+      "loss": 0.3412,
+      "step": 902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.395142540145656e-05,
+      "loss": 0.3059,
+      "step": 903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.393719870244969e-05,
+      "loss": 0.3226,
+      "step": 904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.3922958280306396e-05,
+      "loss": 0.3333,
+      "step": 905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.390870414904839e-05,
+      "loss": 0.3049,
+      "step": 906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.389443632271087e-05,
+      "loss": 0.3259,
+      "step": 907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.388015481534252e-05,
+      "loss": 0.3092,
+      "step": 908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.386585964100551e-05,
+      "loss": 0.2849,
+      "step": 909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3851550813775435e-05,
+      "loss": 0.289,
+      "step": 910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3837228347741365e-05,
+      "loss": 0.3236,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.382289225700578e-05,
+      "loss": 0.3371,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.380854255568458e-05,
+      "loss": 0.259,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3794179257907084e-05,
+      "loss": 0.3427,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3779802377815966e-05,
+      "loss": 0.3554,
+      "step": 915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.376541192956729e-05,
+      "loss": 0.3414,
+      "step": 916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.37510079273305e-05,
+      "loss": 0.3081,
+      "step": 917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.373659038528834e-05,
+      "loss": 0.3384,
+      "step": 918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.372215931763694e-05,
+      "loss": 0.3304,
+      "step": 919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3707714738585694e-05,
+      "loss": 0.3352,
+      "step": 920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.369325666235734e-05,
+      "loss": 0.2817,
+      "step": 921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.36787851031879e-05,
+      "loss": 0.3017,
+      "step": 922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3664300075326635e-05,
+      "loss": 0.3782,
+      "step": 923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.364980159303612e-05,
+      "loss": 0.262,
+      "step": 924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3635289670592145e-05,
+      "loss": 0.3569,
+      "step": 925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3620764322283754e-05,
+      "loss": 0.2977,
+      "step": 926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3606225562413186e-05,
+      "loss": 0.3098,
+      "step": 927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3591673405295906e-05,
+      "loss": 0.3271,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.357710786526057e-05,
+      "loss": 0.2894,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.356252895664899e-05,
+      "loss": 0.2911,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.354793669381616e-05,
+      "loss": 0.3255,
+      "step": 931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.353333109113023e-05,
+      "loss": 0.2581,
+      "step": 932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3518712162972474e-05,
+      "loss": 0.3249,
+      "step": 933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.350407992373727e-05,
+      "loss": 0.3111,
+      "step": 934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.348943438783213e-05,
+      "loss": 0.2897,
+      "step": 935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.347477556967765e-05,
+      "loss": 0.3078,
+      "step": 936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3460103483707506e-05,
+      "loss": 0.3097,
+      "step": 937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3445418144368423e-05,
+      "loss": 0.2982,
+      "step": 938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.34307195661202e-05,
+      "loss": 0.3055,
+      "step": 939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.341600776343565e-05,
+      "loss": 0.3472,
+      "step": 940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.340128275080062e-05,
+      "loss": 0.2702,
+      "step": 941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.338654454271396e-05,
+      "loss": 0.3495,
+      "step": 942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.337179315368751e-05,
+      "loss": 0.3218,
+      "step": 943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.33570285982461e-05,
+      "loss": 0.3536,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3342250890927507e-05,
+      "loss": 0.3072,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.332746004628247e-05,
+      "loss": 0.3067,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3312656078874654e-05,
+      "loss": 0.3067,
+      "step": 947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.329783900328066e-05,
+      "loss": 0.3143,
+      "step": 948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.328300883408998e-05,
+      "loss": 0.3458,
+      "step": 949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3268165585905014e-05,
+      "loss": 0.311,
+      "step": 950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3253309273341036e-05,
+      "loss": 0.3433,
+      "step": 951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.323843991102616e-05,
+      "loss": 0.3223,
+      "step": 952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3223557513601376e-05,
+      "loss": 0.2908,
+      "step": 953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.320866209572052e-05,
+      "loss": 0.3294,
+      "step": 954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.31937536720502e-05,
+      "loss": 0.3429,
+      "step": 955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3178832257269895e-05,
+      "loss": 0.308,
+      "step": 956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.31638978660718e-05,
+      "loss": 0.2611,
+      "step": 957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.314895051316097e-05,
+      "loss": 0.2983,
+      "step": 958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.313399021325515e-05,
+      "loss": 0.3162,
+      "step": 959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3119016981084886e-05,
+      "loss": 0.3106,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.310403083139341e-05,
+      "loss": 0.3324,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.308903177893672e-05,
+      "loss": 0.3443,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3074019838483485e-05,
+      "loss": 0.3134,
+      "step": 963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3058995024815087e-05,
+      "loss": 0.3118,
+      "step": 964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3043957352725566e-05,
+      "loss": 0.3162,
+      "step": 965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.302890683702162e-05,
+      "loss": 0.3377,
+      "step": 966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3013843492522625e-05,
+      "loss": 0.2896,
+      "step": 967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.299876733406055e-05,
+      "loss": 0.318,
+      "step": 968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.2983678376479996e-05,
+      "loss": 0.2868,
+      "step": 969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.296857663463818e-05,
+      "loss": 0.3638,
+      "step": 970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.295346212340489e-05,
+      "loss": 0.3236,
+      "step": 971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.293833485766249e-05,
+      "loss": 0.2926,
+      "step": 972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.292319485230591e-05,
+      "loss": 0.3146,
+      "step": 973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2908042122242615e-05,
+      "loss": 0.2899,
+      "step": 974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2892876682392605e-05,
+      "loss": 0.3134,
+      "step": 975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.287769854768839e-05,
+      "loss": 0.2922,
+      "step": 976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2862507733074996e-05,
+      "loss": 0.2888,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2847304253509904e-05,
+      "loss": 0.311,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.283208812396309e-05,
+      "loss": 0.2929,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.281685935941698e-05,
+      "loss": 0.277,
+      "step": 980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.280161797486643e-05,
+      "loss": 0.2926,
+      "step": 981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.278636398531874e-05,
+      "loss": 0.321,
+      "step": 982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.277109740579361e-05,
+      "loss": 0.3226,
+      "step": 983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2755818251323144e-05,
+      "loss": 0.2902,
+      "step": 984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.274052653695181e-05,
+      "loss": 0.3494,
+      "step": 985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.272522227773647e-05,
+      "loss": 0.3309,
+      "step": 986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2709905488746315e-05,
+      "loss": 0.3271,
+      "step": 987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.269457618506289e-05,
+      "loss": 0.2996,
+      "step": 988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.267923438178004e-05,
+      "loss": 0.308,
+      "step": 989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2663880094003956e-05,
+      "loss": 0.2687,
+      "step": 990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2648513336853083e-05,
+      "loss": 0.3036,
+      "step": 991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2633134125458154e-05,
+      "loss": 0.3032,
+      "step": 992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.261774247496219e-05,
+      "loss": 0.2983,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.260233840052042e-05,
+      "loss": 0.2979,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2586921917300336e-05,
+      "loss": 0.2915,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.257149304048164e-05,
+      "loss": 0.3936,
+      "step": 996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.255605178525623e-05,
+      "loss": 0.2781,
+      "step": 997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.25405981668282e-05,
+      "loss": 0.3515,
+      "step": 998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.252513220041382e-05,
+      "loss": 0.3576,
+      "step": 999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.250965390124151e-05,
+      "loss": 0.3182,
+      "step": 1000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.249416328455183e-05,
+      "loss": 0.272,
+      "step": 1001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.247866036559749e-05,
+      "loss": 0.2623,
+      "step": 1002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2463145159643295e-05,
+      "loss": 0.3244,
+      "step": 1003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.244761768196614e-05,
+      "loss": 0.3425,
+      "step": 1004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.243207794785502e-05,
+      "loss": 0.3211,
+      "step": 1005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2416525972611e-05,
+      "loss": 0.3204,
+      "step": 1006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.240096177154719e-05,
+      "loss": 0.3051,
+      "step": 1007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.238538535998874e-05,
+      "loss": 0.3385,
+      "step": 1008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2369796753272816e-05,
+      "loss": 0.3056,
+      "step": 1009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.23541959667486e-05,
+      "loss": 0.306,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.233858301577727e-05,
+      "loss": 0.2936,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.232295791573198e-05,
+      "loss": 0.3731,
+      "step": 1012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.230732068199784e-05,
+      "loss": 0.3191,
+      "step": 1013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.22916713299719e-05,
+      "loss": 0.2809,
+      "step": 1014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.227600987506316e-05,
+      "loss": 0.3482,
+      "step": 1015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2260336332692524e-05,
+      "loss": 0.3201,
+      "step": 1016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.224465071829282e-05,
+      "loss": 0.2776,
+      "step": 1017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.222895304730874e-05,
+      "loss": 0.3331,
+      "step": 1018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.221324333519684e-05,
+      "loss": 0.3387,
+      "step": 1019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2197521597425554e-05,
+      "loss": 0.3154,
+      "step": 1020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2181787849475155e-05,
+      "loss": 0.3037,
+      "step": 1021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2166042106837734e-05,
+      "loss": 0.331,
+      "step": 1022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.215028438501718e-05,
+      "loss": 0.264,
+      "step": 1023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.213451469952921e-05,
+      "loss": 0.2928,
+      "step": 1024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.211873306590128e-05,
+      "loss": 0.2776,
+      "step": 1025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.210293949967266e-05,
+      "loss": 0.2769,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2087134016394315e-05,
+      "loss": 0.2894,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.207131663162899e-05,
+      "loss": 0.2522,
+      "step": 1028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2055487360951126e-05,
+      "loss": 0.2869,
+      "step": 1029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.203964621994686e-05,
+      "loss": 0.2869,
+      "step": 1030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2023793224214043e-05,
+      "loss": 0.2987,
+      "step": 1031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2007928389362166e-05,
+      "loss": 0.3066,
+      "step": 1032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.199205173101241e-05,
+      "loss": 0.3329,
+      "step": 1033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.197616326479758e-05,
+      "loss": 0.2937,
+      "step": 1034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.196026300636211e-05,
+      "loss": 0.3277,
+      "step": 1035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1944350971362044e-05,
+      "loss": 0.2642,
+      "step": 1036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.192842717546502e-05,
+      "loss": 0.3357,
+      "step": 1037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.191249163435026e-05,
+      "loss": 0.3351,
+      "step": 1038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.189654436370855e-05,
+      "loss": 0.2801,
+      "step": 1039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.188058537924224e-05,
+      "loss": 0.3167,
+      "step": 1040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.186461469666517e-05,
+      "loss": 0.3097,
+      "step": 1041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.184863233170275e-05,
+      "loss": 0.2786,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1832638300091876e-05,
+      "loss": 0.3424,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1816632617580914e-05,
+      "loss": 0.2898,
+      "step": 1044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.180061529992971e-05,
+      "loss": 0.3627,
+      "step": 1045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1784586362909574e-05,
+      "loss": 0.3244,
+      "step": 1046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.176854582230327e-05,
+      "loss": 0.2881,
+      "step": 1047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.175249369390495e-05,
+      "loss": 0.2951,
+      "step": 1048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.17364299935202e-05,
+      "loss": 0.3084,
+      "step": 1049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1720354736966006e-05,
+      "loss": 0.3055,
+      "step": 1050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.170426794007072e-05,
+      "loss": 0.3067,
+      "step": 1051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.168816961867406e-05,
+      "loss": 0.2578,
+      "step": 1052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1672059788627084e-05,
+      "loss": 0.2883,
+      "step": 1053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.165593846579221e-05,
+      "loss": 0.2791,
+      "step": 1054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.163980566604314e-05,
+      "loss": 0.2947,
+      "step": 1055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.162366140526488e-05,
+      "loss": 0.4196,
+      "step": 1056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.160750569935375e-05,
+      "loss": 0.2924,
+      "step": 1057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.159133856421731e-05,
+      "loss": 0.3014,
+      "step": 1058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1575160015774375e-05,
+      "loss": 0.2825,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.155897006995502e-05,
+      "loss": 0.339,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1542768742700516e-05,
+      "loss": 0.3633,
+      "step": 1061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.152655604996337e-05,
+      "loss": 0.3144,
+      "step": 1062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.151033200770724e-05,
+      "loss": 0.3106,
+      "step": 1063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1494096631907e-05,
+      "loss": 0.3317,
+      "step": 1064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1477849938548656e-05,
+      "loss": 0.271,
+      "step": 1065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.146159194362938e-05,
+      "loss": 0.3181,
+      "step": 1066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1445322663157445e-05,
+      "loss": 0.3017,
+      "step": 1067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.142904211315226e-05,
+      "loss": 0.3528,
+      "step": 1068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1412750309644314e-05,
+      "loss": 0.3046,
+      "step": 1069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.13964472686752e-05,
+      "loss": 0.3153,
+      "step": 1070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1380133006297526e-05,
+      "loss": 0.3233,
+      "step": 1071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1363807538575015e-05,
+      "loss": 0.2857,
+      "step": 1072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.134747088158238e-05,
+      "loss": 0.2956,
+      "step": 1073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.133112305140536e-05,
+      "loss": 0.3234,
+      "step": 1074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.13147640641407e-05,
+      "loss": 0.2613,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.129839393589613e-05,
+      "loss": 0.3136,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.128201268279034e-05,
+      "loss": 0.3448,
+      "step": 1077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1265620320953e-05,
+      "loss": 0.2996,
+      "step": 1078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.124921686652468e-05,
+      "loss": 0.2777,
+      "step": 1079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.12328023356569e-05,
+      "loss": 0.2471,
+      "step": 1080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1216376744512093e-05,
+      "loss": 0.3255,
+      "step": 1081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.119994010926355e-05,
+      "loss": 0.2949,
+      "step": 1082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.118349244609546e-05,
+      "loss": 0.2926,
+      "step": 1083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.116703377120286e-05,
+      "loss": 0.2521,
+      "step": 1084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1150564100791663e-05,
+      "loss": 0.3146,
+      "step": 1085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.113408345107855e-05,
+      "loss": 0.2474,
+      "step": 1086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.111759183829105e-05,
+      "loss": 0.3018,
+      "step": 1087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1101089278667486e-05,
+      "loss": 0.3074,
+      "step": 1088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1084575788456955e-05,
+      "loss": 0.3572,
+      "step": 1089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1068051383919304e-05,
+      "loss": 0.3029,
+      "step": 1090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1051516081325154e-05,
+      "loss": 0.3043,
+      "step": 1091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1034969896955835e-05,
+      "loss": 0.3211,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.101841284710339e-05,
+      "loss": 0.3453,
+      "step": 1093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.100184494807057e-05,
+      "loss": 0.2752,
+      "step": 1094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.098526621617082e-05,
+      "loss": 0.2551,
+      "step": 1095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0968676667728224e-05,
+      "loss": 0.3215,
+      "step": 1096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0952076319077534e-05,
+      "loss": 0.3636,
+      "step": 1097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.093546518656413e-05,
+      "loss": 0.2712,
+      "step": 1098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0918843286544026e-05,
+      "loss": 0.3097,
+      "step": 1099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.090221063538381e-05,
+      "loss": 0.3137,
+      "step": 1100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.088556724946069e-05,
+      "loss": 0.2678,
+      "step": 1101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.086891314516239e-05,
+      "loss": 0.2698,
+      "step": 1102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0852248338887264e-05,
+      "loss": 0.3373,
+      "step": 1103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.083557284704414e-05,
+      "loss": 0.3024,
+      "step": 1104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.081888668605239e-05,
+      "loss": 0.2802,
+      "step": 1105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.080218987234189e-05,
+      "loss": 0.332,
+      "step": 1106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0785482422353026e-05,
+      "loss": 0.3169,
+      "step": 1107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0768764352536614e-05,
+      "loss": 0.2929,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.075203567935396e-05,
+      "loss": 0.3648,
+      "step": 1109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.07352964192768e-05,
+      "loss": 0.2908,
+      "step": 1110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.07185465887873e-05,
+      "loss": 0.2889,
+      "step": 1111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.070178620437801e-05,
+      "loss": 0.3219,
+      "step": 1112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.068501528255191e-05,
+      "loss": 0.2699,
+      "step": 1113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0668233839822334e-05,
+      "loss": 0.2739,
+      "step": 1114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.065144189271297e-05,
+      "loss": 0.3445,
+      "step": 1115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0634639457757855e-05,
+      "loss": 0.2498,
+      "step": 1116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0617826551501364e-05,
+      "loss": 0.2717,
+      "step": 1117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0601003190498166e-05,
+      "loss": 0.2824,
+      "step": 1118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0584169391313236e-05,
+      "loss": 0.3352,
+      "step": 1119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0567325170521816e-05,
+      "loss": 0.2682,
+      "step": 1120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0550470544709415e-05,
+      "loss": 0.284,
+      "step": 1121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.05336055304718e-05,
+      "loss": 0.2875,
+      "step": 1122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0516730144414937e-05,
+      "loss": 0.307,
+      "step": 1123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0499844403155038e-05,
+      "loss": 0.2817,
+      "step": 1124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.048294832331848e-05,
+      "loss": 0.2815,
+      "step": 1125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0466041921541846e-05,
+      "loss": 0.3268,
+      "step": 1126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.044912521447186e-05,
+      "loss": 0.2976,
+      "step": 1127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.043219821876542e-05,
+      "loss": 0.2902,
+      "step": 1128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0415260951089533e-05,
+      "loss": 0.254,
+      "step": 1129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0398313428121316e-05,
+      "loss": 0.2871,
+      "step": 1130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0381355666548008e-05,
+      "loss": 0.2358,
+      "step": 1131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0364387683066906e-05,
+      "loss": 0.2642,
+      "step": 1132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.034740949438538e-05,
+      "loss": 0.2879,
+      "step": 1133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.033042111722086e-05,
+      "loss": 0.27,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0313422568300798e-05,
+      "loss": 0.3095,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.029641386436264e-05,
+      "loss": 0.3227,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0279395022153872e-05,
+      "loss": 0.3251,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0262366058431945e-05,
+      "loss": 0.2802,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0245326989964263e-05,
+      "loss": 0.2727,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0228277833528186e-05,
+      "loss": 0.3182,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0211218605911025e-05,
+      "loss": 0.2861,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.019414932390999e-05,
+      "loss": 0.2933,
+      "step": 1142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0177070004332195e-05,
+      "loss": 0.2958,
+      "step": 1143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0159980663994628e-05,
+      "loss": 0.3601,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.014288131972417e-05,
+      "loss": 0.2681,
+      "step": 1145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.012577198835752e-05,
+      "loss": 0.2948,
+      "step": 1146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0108652686741233e-05,
+      "loss": 0.2551,
+      "step": 1147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0091523431731674e-05,
+      "loss": 0.278,
+      "step": 1148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0074384240195004e-05,
+      "loss": 0.294,
+      "step": 1149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.005723512900718e-05,
+      "loss": 0.3035,
+      "step": 1150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.004007611505391e-05,
+      "loss": 0.2935,
+      "step": 1151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0022907215230678e-05,
+      "loss": 0.3024,
+      "step": 1152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0005728446442667e-05,
+      "loss": 0.3035,
+      "step": 1153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.998853982560481e-05,
+      "loss": 0.3008,
+      "step": 1154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9971341369641713e-05,
+      "loss": 0.2939,
+      "step": 1155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9954133095487693e-05,
+      "loss": 0.2639,
+      "step": 1156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9936915020086714e-05,
+      "loss": 0.2915,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9919687160392396e-05,
+      "loss": 0.3559,
+      "step": 1158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9902449533368e-05,
+      "loss": 0.2839,
+      "step": 1159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9885202155986405e-05,
+      "loss": 0.2791,
+      "step": 1160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9867945045230067e-05,
+      "loss": 0.2945,
+      "step": 1161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9850678218091064e-05,
+      "loss": 0.3152,
+      "step": 1162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9833401691571e-05,
+      "loss": 0.2912,
+      "step": 1163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9816115482681073e-05,
+      "loss": 0.3056,
+      "step": 1164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.979881960844198e-05,
+      "loss": 0.2728,
+      "step": 1165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9781514085883934e-05,
+      "loss": 0.2436,
+      "step": 1166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9764198932046686e-05,
+      "loss": 0.2992,
+      "step": 1167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9746874163979433e-05,
+      "loss": 0.2591,
+      "step": 1168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.972953979874085e-05,
+      "loss": 0.2808,
+      "step": 1169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.971219585339906e-05,
+      "loss": 0.2626,
+      "step": 1170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9694842345031628e-05,
+      "loss": 0.2499,
+      "step": 1171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.967747929072553e-05,
+      "loss": 0.2764,
+      "step": 1172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9660106707577135e-05,
+      "loss": 0.2514,
+      "step": 1173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9642724612692208e-05,
+      "loss": 0.3105,
+      "step": 1174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9625333023185853e-05,
+      "loss": 0.3361,
+      "step": 1175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9607931956182558e-05,
+      "loss": 0.3058,
+      "step": 1176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.959052142881612e-05,
+      "loss": 0.2964,
+      "step": 1177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.957310145822965e-05,
+      "loss": 0.3041,
+      "step": 1178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.955567206157558e-05,
+      "loss": 0.2655,
+      "step": 1179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.95382332560156e-05,
+      "loss": 0.3232,
+      "step": 1180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.952078505872067e-05,
+      "loss": 0.2569,
+      "step": 1181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9503327486871e-05,
+      "loss": 0.3598,
+      "step": 1182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.948586055765603e-05,
+      "loss": 0.2619,
+      "step": 1183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9468384288274413e-05,
+      "loss": 0.2787,
+      "step": 1184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9450898695934e-05,
+      "loss": 0.3201,
+      "step": 1185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9433403797851822e-05,
+      "loss": 0.3046,
+      "step": 1186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9415899611254074e-05,
+      "loss": 0.2832,
+      "step": 1187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9398386153376086e-05,
+      "loss": 0.2883,
+      "step": 1188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9380863441462338e-05,
+      "loss": 0.2599,
+      "step": 1189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.93633314927664e-05,
+      "loss": 0.2724,
+      "step": 1190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9345790324550952e-05,
+      "loss": 0.2889,
+      "step": 1191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.932823995408774e-05,
+      "loss": 0.3351,
+      "step": 1192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9310680398657577e-05,
+      "loss": 0.2535,
+      "step": 1193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9293111675550333e-05,
+      "loss": 0.3524,
+      "step": 1194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.927553380206488e-05,
+      "loss": 0.2856,
+      "step": 1195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9257946795509115e-05,
+      "loss": 0.3753,
+      "step": 1196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9240350673199925e-05,
+      "loss": 0.2994,
+      "step": 1197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9222745452463174e-05,
+      "loss": 0.2873,
+      "step": 1198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9205131150633677e-05,
+      "loss": 0.2461,
+      "step": 1199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.91875077850552e-05,
+      "loss": 0.2379,
+      "step": 1200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9169875373080434e-05,
+      "loss": 0.3205,
+      "step": 1201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.915223393207097e-05,
+      "loss": 0.2563,
+      "step": 1202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9134583479397294e-05,
+      "loss": 0.292,
+      "step": 1203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9116924032438768e-05,
+      "loss": 0.3817,
+      "step": 1204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9099255608583596e-05,
+      "loss": 0.2522,
+      "step": 1205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9081578225228845e-05,
+      "loss": 0.2599,
+      "step": 1206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.906389189978038e-05,
+      "loss": 0.2691,
+      "step": 1207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9046196649652877e-05,
+      "loss": 0.2687,
+      "step": 1208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9028492492269823e-05,
+      "loss": 0.2727,
+      "step": 1209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9010779445063436e-05,
+      "loss": 0.3225,
+      "step": 1210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8993057525474718e-05,
+      "loss": 0.3267,
+      "step": 1211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.89753267509534e-05,
+      "loss": 0.2582,
+      "step": 1212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8957587138957913e-05,
+      "loss": 0.3042,
+      "step": 1213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.893983870695542e-05,
+      "loss": 0.3169,
+      "step": 1214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8922081472421754e-05,
+      "loss": 0.2859,
+      "step": 1215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8904315452841403e-05,
+      "loss": 0.3012,
+      "step": 1216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8886540665707525e-05,
+      "loss": 0.2857,
+      "step": 1217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8868757128521898e-05,
+      "loss": 0.3293,
+      "step": 1218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8850964858794932e-05,
+      "loss": 0.2915,
+      "step": 1219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8833163874045616e-05,
+      "loss": 0.3026,
+      "step": 1220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8815354191801527e-05,
+      "loss": 0.2421,
+      "step": 1221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8797535829598806e-05,
+      "loss": 0.301,
+      "step": 1222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8779708804982153e-05,
+      "loss": 0.2829,
+      "step": 1223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8761873135504776e-05,
+      "loss": 0.2765,
+      "step": 1224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8744028838728407e-05,
+      "loss": 0.2898,
+      "step": 1225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.872617593222327e-05,
+      "loss": 0.2794,
+      "step": 1226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8708314433568074e-05,
+      "loss": 0.3247,
+      "step": 1227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8690444360349977e-05,
+      "loss": 0.3186,
+      "step": 1228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8672565730164588e-05,
+      "loss": 0.2629,
+      "step": 1229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.865467856061593e-05,
+      "loss": 0.2835,
+      "step": 1230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.863678286931645e-05,
+      "loss": 0.3082,
+      "step": 1231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8618878673886973e-05,
+      "loss": 0.3227,
+      "step": 1232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.860096599195671e-05,
+      "loss": 0.2414,
+      "step": 1233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.858304484116322e-05,
+      "loss": 0.249,
+      "step": 1234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8565115239152387e-05,
+      "loss": 0.2855,
+      "step": 1235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8547177203578457e-05,
+      "loss": 0.2942,
+      "step": 1236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8529230752103932e-05,
+      "loss": 0.3294,
+      "step": 1237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8511275902399633e-05,
+      "loss": 0.2422,
+      "step": 1238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8493312672144635e-05,
+      "loss": 0.2714,
+      "step": 1239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847534107902628e-05,
+      "loss": 0.2769,
+      "step": 1240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8457361140740118e-05,
+      "loss": 0.2987,
+      "step": 1241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843937287498995e-05,
+      "loss": 0.272,
+      "step": 1242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842137629948775e-05,
+      "loss": 0.2878,
+      "step": 1243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8403371431953678e-05,
+      "loss": 0.2657,
+      "step": 1244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838535829011607e-05,
+      "loss": 0.2575,
+      "step": 1245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83673368917114e-05,
+      "loss": 0.2394,
+      "step": 1246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8349307254484282e-05,
+      "loss": 0.2914,
+      "step": 1247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8331269396187426e-05,
+      "loss": 0.2782,
+      "step": 1248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.831322333458165e-05,
+      "loss": 0.2392,
+      "step": 1249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.829516908743585e-05,
+      "loss": 0.3436,
+      "step": 1250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8277106672526966e-05,
+      "loss": 0.2793,
+      "step": 1251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.825903610763999e-05,
+      "loss": 0.2776,
+      "step": 1252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8240957410567954e-05,
+      "loss": 0.2508,
+      "step": 1253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8222870599111877e-05,
+      "loss": 0.3078,
+      "step": 1254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8204775691080768e-05,
+      "loss": 0.2564,
+      "step": 1255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8186672704291625e-05,
+      "loss": 0.2677,
+      "step": 1256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8168561656569373e-05,
+      "loss": 0.2682,
+      "step": 1257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8150442565746915e-05,
+      "loss": 0.3013,
+      "step": 1258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8132315449665024e-05,
+      "loss": 0.2498,
+      "step": 1259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8114180326172423e-05,
+      "loss": 0.294,
+      "step": 1260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8096037213125678e-05,
+      "loss": 0.2711,
+      "step": 1261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8077886128389255e-05,
+      "loss": 0.3016,
+      "step": 1262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8059727089835446e-05,
+      "loss": 0.2958,
+      "step": 1263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8041560115344396e-05,
+      "loss": 0.2542,
+      "step": 1264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8023385222804033e-05,
+      "loss": 0.2708,
+      "step": 1265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.800520243011012e-05,
+      "loss": 0.2608,
+      "step": 1266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7987011755166168e-05,
+      "loss": 0.3066,
+      "step": 1267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7968813215883458e-05,
+      "loss": 0.2344,
+      "step": 1268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7950606830181026e-05,
+      "loss": 0.283,
+      "step": 1269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7932392615985614e-05,
+      "loss": 0.2393,
+      "step": 1270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7914170591231693e-05,
+      "loss": 0.2653,
+      "step": 1271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7895940773861414e-05,
+      "loss": 0.3203,
+      "step": 1272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7877703181824588e-05,
+      "loss": 0.3147,
+      "step": 1273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.785945783307871e-05,
+      "loss": 0.2776,
+      "step": 1274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.784120474558889e-05,
+      "loss": 0.2887,
+      "step": 1275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7822943937327867e-05,
+      "loss": 0.3313,
+      "step": 1276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.780467542627598e-05,
+      "loss": 0.2802,
+      "step": 1277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7786399230421154e-05,
+      "loss": 0.2962,
+      "step": 1278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.776811536775888e-05,
+      "loss": 0.3093,
+      "step": 1279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.77498238562922e-05,
+      "loss": 0.2873,
+      "step": 1280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7731524714031677e-05,
+      "loss": 0.2947,
+      "step": 1281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7713217958995396e-05,
+      "loss": 0.2317,
+      "step": 1282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7694903609208948e-05,
+      "loss": 0.2729,
+      "step": 1283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7676581682705384e-05,
+      "loss": 0.2437,
+      "step": 1284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7658252197525225e-05,
+      "loss": 0.2968,
+      "step": 1285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7639915171716435e-05,
+      "loss": 0.2593,
+      "step": 1286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7621570623334383e-05,
+      "loss": 0.2629,
+      "step": 1287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.760321857044189e-05,
+      "loss": 0.283,
+      "step": 1288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.758485903110912e-05,
+      "loss": 0.2864,
+      "step": 1289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7566492023413633e-05,
+      "loss": 0.3129,
+      "step": 1290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7548117565440335e-05,
+      "loss": 0.2856,
+      "step": 1291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7529735675281473e-05,
+      "loss": 0.2851,
+      "step": 1292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7511346371036604e-05,
+      "loss": 0.2365,
+      "step": 1293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.749294967081259e-05,
+      "loss": 0.2598,
+      "step": 1294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7474545592723572e-05,
+      "loss": 0.2711,
+      "step": 1295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.745613415489097e-05,
+      "loss": 0.2674,
+      "step": 1296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7437715375443438e-05,
+      "loss": 0.2475,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7419289272516845e-05,
+      "loss": 0.2817,
+      "step": 1298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7400855864254304e-05,
+      "loss": 0.2556,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7382415168806092e-05,
+      "loss": 0.2632,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7363967204329674e-05,
+      "loss": 0.2765,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7345511988989683e-05,
+      "loss": 0.2789,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.732704954095786e-05,
+      "loss": 0.2839,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7308579878413096e-05,
+      "loss": 0.2784,
+      "step": 1304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.729010301954139e-05,
+      "loss": 0.2674,
+      "step": 1305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.72716189825358e-05,
+      "loss": 0.2761,
+      "step": 1306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.725312778559646e-05,
+      "loss": 0.3045,
+      "step": 1307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.723462944693057e-05,
+      "loss": 0.2613,
+      "step": 1308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7216123984752347e-05,
+      "loss": 0.266,
+      "step": 1309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.719761141728303e-05,
+      "loss": 0.302,
+      "step": 1310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7179091762750852e-05,
+      "loss": 0.271,
+      "step": 1311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.716056503939102e-05,
+      "loss": 0.2389,
+      "step": 1312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.71420312654457e-05,
+      "loss": 0.2841,
+      "step": 1313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.712349045916401e-05,
+      "loss": 0.2561,
+      "step": 1314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7104942638801986e-05,
+      "loss": 0.2672,
+      "step": 1315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7086387822622576e-05,
+      "loss": 0.2674,
+      "step": 1316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7067826028895594e-05,
+      "loss": 0.3066,
+      "step": 1317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7049257275897765e-05,
+      "loss": 0.2653,
+      "step": 1318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7030681581912634e-05,
+      "loss": 0.2966,
+      "step": 1319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7012098965230587e-05,
+      "loss": 0.3779,
+      "step": 1320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.699350944414883e-05,
+      "loss": 0.2511,
+      "step": 1321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6974913036971366e-05,
+      "loss": 0.275,
+      "step": 1322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6956309762008982e-05,
+      "loss": 0.3185,
+      "step": 1323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.693769963757922e-05,
+      "loss": 0.2672,
+      "step": 1324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6919082682006377e-05,
+      "loss": 0.2488,
+      "step": 1325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6900458913621463e-05,
+      "loss": 0.2852,
+      "step": 1326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6881828350762206e-05,
+      "loss": 0.2474,
+      "step": 1327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.686319101177302e-05,
+      "loss": 0.2536,
+      "step": 1328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.684454691500499e-05,
+      "loss": 0.2102,
+      "step": 1329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6825896078815863e-05,
+      "loss": 0.2706,
+      "step": 1330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6807238521570007e-05,
+      "loss": 0.2708,
+      "step": 1331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.678857426163843e-05,
+      "loss": 0.2778,
+      "step": 1332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6769903317398713e-05,
+      "loss": 0.2774,
+      "step": 1333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.675122570723503e-05,
+      "loss": 0.2759,
+      "step": 1334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6732541449538136e-05,
+      "loss": 0.2873,
+      "step": 1335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6713850562705307e-05,
+      "loss": 0.2472,
+      "step": 1336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6695153065140353e-05,
+      "loss": 0.3502,
+      "step": 1337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6676448975253605e-05,
+      "loss": 0.2692,
+      "step": 1338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6657738311461866e-05,
+      "loss": 0.2945,
+      "step": 1339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6639021092188424e-05,
+      "loss": 0.2954,
+      "step": 1340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6620297335863022e-05,
+      "loss": 0.2869,
+      "step": 1341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6601567060921828e-05,
+      "loss": 0.2692,
+      "step": 1342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.658283028580745e-05,
+      "loss": 0.2571,
+      "step": 1343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6564087028968882e-05,
+      "loss": 0.2687,
+      "step": 1344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6545337308861493e-05,
+      "loss": 0.322,
+      "step": 1345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6526581143947038e-05,
+      "loss": 0.2892,
+      "step": 1346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.650781855269359e-05,
+      "loss": 0.3041,
+      "step": 1347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.648904955357557e-05,
+      "loss": 0.3003,
+      "step": 1348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6470274165073716e-05,
+      "loss": 0.3305,
+      "step": 1349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.645149240567502e-05,
+      "loss": 0.3479,
+      "step": 1350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.643270429387279e-05,
+      "loss": 0.2732,
+      "step": 1351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6413909848166568e-05,
+      "loss": 0.306,
+      "step": 1352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6395109087062124e-05,
+      "loss": 0.2607,
+      "step": 1353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.637630202907147e-05,
+      "loss": 0.2938,
+      "step": 1354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6357488692712797e-05,
+      "loss": 0.2655,
+      "step": 1355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6338669096510484e-05,
+      "loss": 0.2785,
+      "step": 1356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6319843258995085e-05,
+      "loss": 0.2432,
+      "step": 1357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.630101119870329e-05,
+      "loss": 0.2921,
+      "step": 1358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6282172934177907e-05,
+      "loss": 0.2665,
+      "step": 1359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6263328483967866e-05,
+      "loss": 0.2955,
+      "step": 1360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6244477866628187e-05,
+      "loss": 0.2783,
+      "step": 1361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.622562110071996e-05,
+      "loss": 0.2402,
+      "step": 1362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6206758204810327e-05,
+      "loss": 0.2873,
+      "step": 1363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.618788919747246e-05,
+      "loss": 0.272,
+      "step": 1364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.616901409728557e-05,
+      "loss": 0.26,
+      "step": 1365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6150132922834847e-05,
+      "loss": 0.2899,
+      "step": 1366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.613124569271147e-05,
+      "loss": 0.2467,
+      "step": 1367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6112352425512578e-05,
+      "loss": 0.2653,
+      "step": 1368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6093453139841245e-05,
+      "loss": 0.2618,
+      "step": 1369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6074547854306498e-05,
+      "loss": 0.2856,
+      "step": 1370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.605563658752325e-05,
+      "loss": 0.2812,
+      "step": 1371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6036719358112302e-05,
+      "loss": 0.2895,
+      "step": 1372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.601779618470034e-05,
+      "loss": 0.2492,
+      "step": 1373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5998867085919894e-05,
+      "loss": 0.2879,
+      "step": 1374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5979932080409325e-05,
+      "loss": 0.2646,
+      "step": 1375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5960991186812825e-05,
+      "loss": 0.2689,
+      "step": 1376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5942044423780352e-05,
+      "loss": 0.2906,
+      "step": 1377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5923091809967688e-05,
+      "loss": 0.2303,
+      "step": 1378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5904133364036345e-05,
+      "loss": 0.2715,
+      "step": 1379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.588516910465358e-05,
+      "loss": 0.2601,
+      "step": 1380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5866199050492376e-05,
+      "loss": 0.2816,
+      "step": 1381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5847223220231438e-05,
+      "loss": 0.2547,
+      "step": 1382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5828241632555135e-05,
+      "loss": 0.2515,
+      "step": 1383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.580925430615352e-05,
+      "loss": 0.3006,
+      "step": 1384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.579026125972229e-05,
+      "loss": 0.2384,
+      "step": 1385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.577126251196277e-05,
+      "loss": 0.2724,
+      "step": 1386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5752258081581927e-05,
+      "loss": 0.2737,
+      "step": 1387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5733247987292274e-05,
+      "loss": 0.2475,
+      "step": 1388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.571423224781194e-05,
+      "loss": 0.2849,
+      "step": 1389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5695210881864604e-05,
+      "loss": 0.2986,
+      "step": 1390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5676183908179477e-05,
+      "loss": 0.2732,
+      "step": 1391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5657151345491293e-05,
+      "loss": 0.2771,
+      "step": 1392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5638113212540306e-05,
+      "loss": 0.2669,
+      "step": 1393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5619069528072227e-05,
+      "loss": 0.276,
+      "step": 1394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.560002031083825e-05,
+      "loss": 0.2565,
+      "step": 1395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5580965579595015e-05,
+      "loss": 0.272,
+      "step": 1396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5561905353104594e-05,
+      "loss": 0.3379,
+      "step": 1397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5542839650134466e-05,
+      "loss": 0.2532,
+      "step": 1398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.552376848945749e-05,
+      "loss": 0.2903,
+      "step": 1399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5504691889851935e-05,
+      "loss": 0.294,
+      "step": 1400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5485609870101388e-05,
+      "loss": 0.2908,
+      "step": 1401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5466522448994782e-05,
+      "loss": 0.2806,
+      "step": 1402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.544742964532638e-05,
+      "loss": 0.3172,
+      "step": 1403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5428331477895735e-05,
+      "loss": 0.2311,
+      "step": 1404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.540922796550769e-05,
+      "loss": 0.2603,
+      "step": 1405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5390119126972348e-05,
+      "loss": 0.2867,
+      "step": 1406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.537100498110504e-05,
+      "loss": 0.2641,
+      "step": 1407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5351885546726346e-05,
+      "loss": 0.2485,
+      "step": 1408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5332760842662047e-05,
+      "loss": 0.2562,
+      "step": 1409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5313630887743095e-05,
+      "loss": 0.2933,
+      "step": 1410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5294495700805644e-05,
+      "loss": 0.2516,
+      "step": 1411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5275355300690976e-05,
+      "loss": 0.2724,
+      "step": 1412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5256209706245506e-05,
+      "loss": 0.2815,
+      "step": 1413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5237058936320775e-05,
+      "loss": 0.2641,
+      "step": 1414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5217903009773425e-05,
+      "loss": 0.253,
+      "step": 1415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5198741945465147e-05,
+      "loss": 0.3037,
+      "step": 1416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5179575762262724e-05,
+      "loss": 0.2345,
+      "step": 1417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5160404479037953e-05,
+      "loss": 0.2514,
+      "step": 1418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5141228114667677e-05,
+      "loss": 0.305,
+      "step": 1419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5122046688033726e-05,
+      "loss": 0.2787,
+      "step": 1420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5102860218022903e-05,
+      "loss": 0.2643,
+      "step": 1421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.508366872352701e-05,
+      "loss": 0.2576,
+      "step": 1422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5064472223442776e-05,
+      "loss": 0.2437,
+      "step": 1423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.504527073667185e-05,
+      "loss": 0.222,
+      "step": 1424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5026064282120812e-05,
+      "loss": 0.2564,
+      "step": 1425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.500685287870112e-05,
+      "loss": 0.281,
+      "step": 1426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4987636545329105e-05,
+      "loss": 0.2771,
+      "step": 1427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4968415300925965e-05,
+      "loss": 0.2311,
+      "step": 1428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4949189164417713e-05,
+      "loss": 0.2742,
+      "step": 1429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4929958154735198e-05,
+      "loss": 0.2612,
+      "step": 1430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4910722290814057e-05,
+      "loss": 0.2596,
+      "step": 1431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4891481591594707e-05,
+      "loss": 0.2501,
+      "step": 1432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.487223607602233e-05,
+      "loss": 0.2753,
+      "step": 1433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.485298576304685e-05,
+      "loss": 0.2608,
+      "step": 1434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4833730671622908e-05,
+      "loss": 0.2348,
+      "step": 1435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.481447082070987e-05,
+      "loss": 0.221,
+      "step": 1436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.479520622927175e-05,
+      "loss": 0.2832,
+      "step": 1437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4775936916277267e-05,
+      "loss": 0.278,
+      "step": 1438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4756662900699778e-05,
+      "loss": 0.2702,
+      "step": 1439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4737384201517264e-05,
+      "loss": 0.2468,
+      "step": 1440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.471810083771232e-05,
+      "loss": 0.2819,
+      "step": 1441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4698812828272137e-05,
+      "loss": 0.2632,
+      "step": 1442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4679520192188484e-05,
+      "loss": 0.2828,
+      "step": 1443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4660222948457674e-05,
+      "loss": 0.2908,
+      "step": 1444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4640921116080564e-05,
+      "loss": 0.3365,
+      "step": 1445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.462161471406252e-05,
+      "loss": 0.2652,
+      "step": 1446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.460230376141344e-05,
+      "loss": 0.2666,
+      "step": 1447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.458298827714765e-05,
+      "loss": 0.2828,
+      "step": 1448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4563668280283984e-05,
+      "loss": 0.2644,
+      "step": 1449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4544343789845703e-05,
+      "loss": 0.2391,
+      "step": 1450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4525014824860476e-05,
+      "loss": 0.2714,
+      "step": 1451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.450568140436041e-05,
+      "loss": 0.2615,
+      "step": 1452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4486343547381975e-05,
+      "loss": 0.2388,
+      "step": 1453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4467001272966017e-05,
+      "loss": 0.2993,
+      "step": 1454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4447654600157728e-05,
+      "loss": 0.2859,
+      "step": 1455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.442830354800664e-05,
+      "loss": 0.2518,
+      "step": 1456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4408948135566587e-05,
+      "loss": 0.2947,
+      "step": 1457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.438958838189571e-05,
+      "loss": 0.2543,
+      "step": 1458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.437022430605639e-05,
+      "loss": 0.2825,
+      "step": 1459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.435085592711532e-05,
+      "loss": 0.3264,
+      "step": 1460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4331483264143385e-05,
+      "loss": 0.2458,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.43121063362157e-05,
+      "loss": 0.2534,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.429272516241158e-05,
+      "loss": 0.2707,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.427333976181452e-05,
+      "loss": 0.282,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4253950153512194e-05,
+      "loss": 0.272,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4234556356596395e-05,
+      "loss": 0.2642,
+      "step": 1466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.421515839016305e-05,
+      "loss": 0.2445,
+      "step": 1467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4195756273312178e-05,
+      "loss": 0.2734,
+      "step": 1468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4176350025147925e-05,
+      "loss": 0.266,
+      "step": 1469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4156939664778453e-05,
+      "loss": 0.2584,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.413752521131601e-05,
+      "loss": 0.3023,
+      "step": 1471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4118106683876853e-05,
+      "loss": 0.2857,
+      "step": 1472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4098684101581266e-05,
+      "loss": 0.3191,
+      "step": 1473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.407925748355352e-05,
+      "loss": 0.313,
+      "step": 1474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4059826848921855e-05,
+      "loss": 0.3248,
+      "step": 1475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.404039221681847e-05,
+      "loss": 0.254,
+      "step": 1476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4020953606379497e-05,
+      "loss": 0.2616,
+      "step": 1477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4001511036744988e-05,
+      "loss": 0.2663,
+      "step": 1478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3982064527058896e-05,
+      "loss": 0.2966,
+      "step": 1479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.396261409646905e-05,
+      "loss": 0.2434,
+      "step": 1480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3943159764127137e-05,
+      "loss": 0.288,
+      "step": 1481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.392370154918869e-05,
+      "loss": 0.2781,
+      "step": 1482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3904239470813076e-05,
+      "loss": 0.2753,
+      "step": 1483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3884773548163435e-05,
+      "loss": 0.2306,
+      "step": 1484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.386530380040672e-05,
+      "loss": 0.2922,
+      "step": 1485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3845830246713643e-05,
+      "loss": 0.2368,
+      "step": 1486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.382635290625866e-05,
+      "loss": 0.2335,
+      "step": 1487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.380687179821996e-05,
+      "loss": 0.2485,
+      "step": 1488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3787386941779444e-05,
+      "loss": 0.2097,
+      "step": 1489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.376789835612268e-05,
+      "loss": 0.2545,
+      "step": 1490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3748406060438956e-05,
+      "loss": 0.2663,
+      "step": 1491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.372891007392116e-05,
+      "loss": 0.3021,
+      "step": 1492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.370941041576585e-05,
+      "loss": 0.2534,
+      "step": 1493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.368990710517318e-05,
+      "loss": 0.2702,
+      "step": 1494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.367040016134692e-05,
+      "loss": 0.2965,
+      "step": 1495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3650889603494398e-05,
+      "loss": 0.2594,
+      "step": 1496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3631375450826514e-05,
+      "loss": 0.261,
+      "step": 1497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3611857722557684e-05,
+      "loss": 0.261,
+      "step": 1498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3592336437905882e-05,
+      "loss": 0.2505,
+      "step": 1499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3572811616092557e-05,
+      "loss": 0.2432,
+      "step": 1500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.355328327634265e-05,
+      "loss": 0.2578,
+      "step": 1501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3533751437884552e-05,
+      "loss": 0.2392,
+      "step": 1502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3514216119950128e-05,
+      "loss": 0.2787,
+      "step": 1503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.349467734177464e-05,
+      "loss": 0.2789,
+      "step": 1504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.347513512259678e-05,
+      "loss": 0.2827,
+      "step": 1505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3455589481658607e-05,
+      "loss": 0.267,
+      "step": 1506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.343604043820556e-05,
+      "loss": 0.2383,
+      "step": 1507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.341648801148643e-05,
+      "loss": 0.2546,
+      "step": 1508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3396932220753334e-05,
+      "loss": 0.264,
+      "step": 1509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3377373085261713e-05,
+      "loss": 0.2622,
+      "step": 1510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.335781062427028e-05,
+      "loss": 0.2655,
+      "step": 1511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3338244857041045e-05,
+      "loss": 0.2434,
+      "step": 1512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.331867580283926e-05,
+      "loss": 0.2579,
+      "step": 1513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3299103480933415e-05,
+      "loss": 0.2025,
+      "step": 1514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3279527910595218e-05,
+      "loss": 0.2723,
+      "step": 1515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3259949111099584e-05,
+      "loss": 0.3057,
+      "step": 1516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3240367101724596e-05,
+      "loss": 0.2633,
+      "step": 1517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.32207819017515e-05,
+      "loss": 0.2589,
+      "step": 1518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3201193530464695e-05,
+      "loss": 0.2361,
+      "step": 1519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3181602007151676e-05,
+      "loss": 0.2415,
+      "step": 1520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3162007351103077e-05,
+      "loss": 0.2405,
+      "step": 1521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3142409581612587e-05,
+      "loss": 0.2705,
+      "step": 1522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.312280871797698e-05,
+      "loss": 0.2506,
+      "step": 1523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3103204779496063e-05,
+      "loss": 0.2847,
+      "step": 1524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3083597785472675e-05,
+      "loss": 0.2563,
+      "step": 1525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.306398775521267e-05,
+      "loss": 0.269,
+      "step": 1526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.304437470802489e-05,
+      "loss": 0.289,
+      "step": 1527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.302475866322113e-05,
+      "loss": 0.2778,
+      "step": 1528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.300513964011617e-05,
+      "loss": 0.2912,
+      "step": 1529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2985517658027688e-05,
+      "loss": 0.2617,
+      "step": 1530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2965892736276296e-05,
+      "loss": 0.2814,
+      "step": 1531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2946264894185495e-05,
+      "loss": 0.2533,
+      "step": 1532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.292663415108166e-05,
+      "loss": 0.263,
+      "step": 1533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.290700052629403e-05,
+      "loss": 0.256,
+      "step": 1534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2887364039154676e-05,
+      "loss": 0.2209,
+      "step": 1535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2867724708998476e-05,
+      "loss": 0.2336,
+      "step": 1536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2848082555163127e-05,
+      "loss": 0.2404,
+      "step": 1537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.28284375969891e-05,
+      "loss": 0.2821,
+      "step": 1538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.280878985381962e-05,
+      "loss": 0.2731,
+      "step": 1539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.278913934500066e-05,
+      "loss": 0.2582,
+      "step": 1540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2769486089880917e-05,
+      "loss": 0.2564,
+      "step": 1541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.274983010781179e-05,
+      "loss": 0.2065,
+      "step": 1542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2730171418147368e-05,
+      "loss": 0.2595,
+      "step": 1543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2710510040244395e-05,
+      "loss": 0.2052,
+      "step": 1544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.269084599346227e-05,
+      "loss": 0.2571,
+      "step": 1545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2671179297163017e-05,
+      "loss": 0.2654,
+      "step": 1546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2651509970711273e-05,
+      "loss": 0.2819,
+      "step": 1547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2631838033474267e-05,
+      "loss": 0.2636,
+      "step": 1548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2612163504821793e-05,
+      "loss": 0.2653,
+      "step": 1549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2592486404126178e-05,
+      "loss": 0.2851,
+      "step": 1550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2572806750762325e-05,
+      "loss": 0.2429,
+      "step": 1551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2553124564107617e-05,
+      "loss": 0.2518,
+      "step": 1552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.253343986354194e-05,
+      "loss": 0.2849,
+      "step": 1553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2513752668447655e-05,
+      "loss": 0.2582,
+      "step": 1554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2494062998209583e-05,
+      "loss": 0.2716,
+      "step": 1555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2474370872214983e-05,
+      "loss": 0.275,
+      "step": 1556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2454676309853524e-05,
+      "loss": 0.2906,
+      "step": 1557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2434979330517278e-05,
+      "loss": 0.2509,
+      "step": 1558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.241527995360069e-05,
+      "loss": 0.2656,
+      "step": 1559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2395578198500588e-05,
+      "loss": 0.3155,
+      "step": 1560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2375874084616117e-05,
+      "loss": 0.2557,
+      "step": 1561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2356167631348758e-05,
+      "loss": 0.2741,
+      "step": 1562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2336458858102286e-05,
+      "loss": 0.2184,
+      "step": 1563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2316747784282776e-05,
+      "loss": 0.2267,
+      "step": 1564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.229703442929855e-05,
+      "loss": 0.3008,
+      "step": 1565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2277318812560184e-05,
+      "loss": 0.2489,
+      "step": 1566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2257600953480484e-05,
+      "loss": 0.2923,
+      "step": 1567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2237880871474467e-05,
+      "loss": 0.2623,
+      "step": 1568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2218158585959326e-05,
+      "loss": 0.2743,
+      "step": 1569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.219843411635443e-05,
+      "loss": 0.2898,
+      "step": 1570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2178707482081308e-05,
+      "loss": 0.2767,
+      "step": 1571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2158978702563593e-05,
+      "loss": 0.2968,
+      "step": 1572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.213924779722707e-05,
+      "loss": 0.2504,
+      "step": 1573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2119514785499584e-05,
+      "loss": 0.2641,
+      "step": 1574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2099779686811072e-05,
+      "loss": 0.2218,
+      "step": 1575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2080042520593517e-05,
+      "loss": 0.2328,
+      "step": 1576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2060303306280938e-05,
+      "loss": 0.2434,
+      "step": 1577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.204056206330938e-05,
+      "loss": 0.2322,
+      "step": 1578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.202081881111688e-05,
+      "loss": 0.2829,
+      "step": 1579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2001073569143438e-05,
+      "loss": 0.2986,
+      "step": 1580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1981326356831047e-05,
+      "loss": 0.2956,
+      "step": 1581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1961577193623607e-05,
+      "loss": 0.2441,
+      "step": 1582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1941826098966962e-05,
+      "loss": 0.2138,
+      "step": 1583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.192207309230884e-05,
+      "loss": 0.2602,
+      "step": 1584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1902318193098865e-05,
+      "loss": 0.2492,
+      "step": 1585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1882561420788515e-05,
+      "loss": 0.2465,
+      "step": 1586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.186280279483113e-05,
+      "loss": 0.2547,
+      "step": 1587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1843042334681842e-05,
+      "loss": 0.2301,
+      "step": 1588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.182328005979762e-05,
+      "loss": 0.2546,
+      "step": 1589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.180351598963721e-05,
+      "loss": 0.2782,
+      "step": 1590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1783750143661124e-05,
+      "loss": 0.2699,
+      "step": 1591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1763982541331616e-05,
+      "loss": 0.2782,
+      "step": 1592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.174421320211269e-05,
+      "loss": 0.2366,
+      "step": 1593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1724442145470034e-05,
+      "loss": 0.2882,
+      "step": 1594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.170466939087105e-05,
+      "loss": 0.2558,
+      "step": 1595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.168489495778479e-05,
+      "loss": 0.2549,
+      "step": 1596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1665118865681976e-05,
+      "loss": 0.249,
+      "step": 1597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1645341134034968e-05,
+      "loss": 0.2693,
+      "step": 1598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.162556178231771e-05,
+      "loss": 0.2682,
+      "step": 1599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1605780830005777e-05,
+      "loss": 0.2516,
+      "step": 1600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1585998296576305e-05,
+      "loss": 0.2617,
+      "step": 1601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1566214201507975e-05,
+      "loss": 0.2249,
+      "step": 1602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1546428564281028e-05,
+      "loss": 0.2516,
+      "step": 1603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.15266414043772e-05,
+      "loss": 0.2932,
+      "step": 1604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.150685274127975e-05,
+      "loss": 0.2332,
+      "step": 1605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.14870625944734e-05,
+      "loss": 0.2325,
+      "step": 1606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1467270983444342e-05,
+      "loss": 0.2678,
+      "step": 1607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.144747792768021e-05,
+      "loss": 0.3054,
+      "step": 1608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.142768344667005e-05,
+      "loss": 0.247,
+      "step": 1609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1407887559904316e-05,
+      "loss": 0.2403,
+      "step": 1610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1388090286874852e-05,
+      "loss": 0.2319,
+      "step": 1611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1368291647074874e-05,
+      "loss": 0.2313,
+      "step": 1612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1348491659998917e-05,
+      "loss": 0.2355,
+      "step": 1613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1328690345142867e-05,
+      "loss": 0.2368,
+      "step": 1614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1308887722003904e-05,
+      "loss": 0.2153,
+      "step": 1615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.128908381008051e-05,
+      "loss": 0.301,
+      "step": 1616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.126927862887243e-05,
+      "loss": 0.2516,
+      "step": 1617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.124947219788064e-05,
+      "loss": 0.2727,
+      "step": 1618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1229664536607372e-05,
+      "loss": 0.3035,
+      "step": 1619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1209855664556062e-05,
+      "loss": 0.2267,
+      "step": 1620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1190045601231333e-05,
+      "loss": 0.2301,
+      "step": 1621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1170234366138986e-05,
+      "loss": 0.2468,
+      "step": 1622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1150421978785975e-05,
+      "loss": 0.2144,
+      "step": 1623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1130608458680388e-05,
+      "loss": 0.2493,
+      "step": 1624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1110793825331427e-05,
+      "loss": 0.2451,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.109097809824939e-05,
+      "loss": 0.2595,
+      "step": 1626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.107116129694565e-05,
+      "loss": 0.2287,
+      "step": 1627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1051343440932647e-05,
+      "loss": 0.2629,
+      "step": 1628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1031524549723853e-05,
+      "loss": 0.2869,
+      "step": 1629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1011704642833756e-05,
+      "loss": 0.2475,
+      "step": 1630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0991883739777856e-05,
+      "loss": 0.2542,
+      "step": 1631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0972061860072608e-05,
+      "loss": 0.2211,
+      "step": 1632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0952239023235465e-05,
+      "loss": 0.2077,
+      "step": 1633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0932415248784795e-05,
+      "loss": 0.1808,
+      "step": 1634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0912590556239895e-05,
+      "loss": 0.1867,
+      "step": 1635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0892764965120977e-05,
+      "loss": 0.1782,
+      "step": 1636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0872938494949126e-05,
+      "loss": 0.1954,
+      "step": 1637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.085311116524629e-05,
+      "loss": 0.2106,
+      "step": 1638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0833282995535285e-05,
+      "loss": 0.1662,
+      "step": 1639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.081345400533972e-05,
+      "loss": 0.168,
+      "step": 1640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.079362421418404e-05,
+      "loss": 0.1915,
+      "step": 1641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.077379364159348e-05,
+      "loss": 0.1733,
+      "step": 1642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0753962307094012e-05,
+      "loss": 0.2078,
+      "step": 1643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.073413023021239e-05,
+      "loss": 0.1921,
+      "step": 1644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0714297430476087e-05,
+      "loss": 0.2304,
+      "step": 1645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0694463927413296e-05,
+      "loss": 0.2158,
+      "step": 1646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0674629740552892e-05,
+      "loss": 0.2171,
+      "step": 1647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0654794889424427e-05,
+      "loss": 0.1814,
+      "step": 1648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0634959393558097e-05,
+      "loss": 0.1865,
+      "step": 1649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0615123272484764e-05,
+      "loss": 0.1974,
+      "step": 1650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.059528654573587e-05,
+      "loss": 0.1978,
+      "step": 1651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.057544923284348e-05,
+      "loss": 0.2213,
+      "step": 1652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.055561135334021e-05,
+      "loss": 0.1802,
+      "step": 1653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0535772926759257e-05,
+      "loss": 0.2167,
+      "step": 1654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0515933972634352e-05,
+      "loss": 0.1903,
+      "step": 1655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.049609451049973e-05,
+      "loss": 0.1726,
+      "step": 1656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0476254559890143e-05,
+      "loss": 0.202,
+      "step": 1657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0456414140340818e-05,
+      "loss": 0.1903,
+      "step": 1658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.043657327138745e-05,
+      "loss": 0.1613,
+      "step": 1659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.041673197256616e-05,
+      "loss": 0.1609,
+      "step": 1660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0396890263413515e-05,
+      "loss": 0.1986,
+      "step": 1661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0377048163466462e-05,
+      "loss": 0.2392,
+      "step": 1662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0357205692262358e-05,
+      "loss": 0.1804,
+      "step": 1663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0337362869338905e-05,
+      "loss": 0.2044,
+      "step": 1664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0317519714234155e-05,
+      "loss": 0.2051,
+      "step": 1665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0297676246486494e-05,
+      "loss": 0.152,
+      "step": 1666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.027783248563462e-05,
+      "loss": 0.1648,
+      "step": 1667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0257988451217506e-05,
+      "loss": 0.1748,
+      "step": 1668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0238144162774405e-05,
+      "loss": 0.1998,
+      "step": 1669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0218299639844802e-05,
+      "loss": 0.1745,
+      "step": 1670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.019845490196844e-05,
+      "loss": 0.1826,
+      "step": 1671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0178609968685257e-05,
+      "loss": 0.2122,
+      "step": 1672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.015876485953539e-05,
+      "loss": 0.159,
+      "step": 1673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.013891959405914e-05,
+      "loss": 0.1999,
+      "step": 1674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.011907419179697e-05,
+      "loss": 0.1814,
+      "step": 1675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.009922867228948e-05,
+      "loss": 0.1835,
+      "step": 1676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.007938305507738e-05,
+      "loss": 0.1827,
+      "step": 1677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0059537359701467e-05,
+      "loss": 0.2056,
+      "step": 1678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0039691605702628e-05,
+      "loss": 0.2019,
+      "step": 1679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0019845812621814e-05,
+      "loss": 0.2322,
+      "step": 1680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2046,
+      "step": 1681
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.998015418737819e-05,
+      "loss": 0.1732,
+      "step": 1682
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9960308394297382e-05,
+      "loss": 0.2065,
+      "step": 1683
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.994046264029854e-05,
+      "loss": 0.1642,
+      "step": 1684
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.992061694492263e-05,
+      "loss": 0.2075,
+      "step": 1685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9900771327710527e-05,
+      "loss": 0.1748,
+      "step": 1686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.988092580820303e-05,
+      "loss": 0.1865,
+      "step": 1687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9861080405940866e-05,
+      "loss": 0.1851,
+      "step": 1688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9841235140464614e-05,
+      "loss": 0.1804,
+      "step": 1689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.982139003131474e-05,
+      "loss": 0.2073,
+      "step": 1690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9801545098031568e-05,
+      "loss": 0.1985,
+      "step": 1691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.97817003601552e-05,
+      "loss": 0.1686,
+      "step": 1692
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.976185583722561e-05,
+      "loss": 0.1916,
+      "step": 1693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.97420115487825e-05,
+      "loss": 0.1969,
+      "step": 1694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.972216751436538e-05,
+      "loss": 0.1941,
+      "step": 1695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.970232375351351e-05,
+      "loss": 0.1718,
+      "step": 1696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9682480285765852e-05,
+      "loss": 0.1833,
+      "step": 1697
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9662637130661105e-05,
+      "loss": 0.2118,
+      "step": 1698
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9642794307737652e-05,
+      "loss": 0.1931,
+      "step": 1699
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9622951836533538e-05,
+      "loss": 0.197,
+      "step": 1700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9603109736586495e-05,
+      "loss": 0.2124,
+      "step": 1701
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9583268027433847e-05,
+      "loss": 0.2094,
+      "step": 1702
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9563426728612555e-05,
+      "loss": 0.1832,
+      "step": 1703
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9543585859659186e-05,
+      "loss": 0.1956,
+      "step": 1704
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9523745440109863e-05,
+      "loss": 0.2035,
+      "step": 1705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.950390548950028e-05,
+      "loss": 0.1852,
+      "step": 1706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9484066027365655e-05,
+      "loss": 0.2035,
+      "step": 1707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9464227073240743e-05,
+      "loss": 0.213,
+      "step": 1708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9444388646659796e-05,
+      "loss": 0.1759,
+      "step": 1709
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9424550767156528e-05,
+      "loss": 0.208,
+      "step": 1710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9404713454264135e-05,
+      "loss": 0.1958,
+      "step": 1711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.938487672751524e-05,
+      "loss": 0.1749,
+      "step": 1712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9365040606441903e-05,
+      "loss": 0.1706,
+      "step": 1713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9345205110575583e-05,
+      "loss": 0.2123,
+      "step": 1714
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9325370259447114e-05,
+      "loss": 0.193,
+      "step": 1715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9305536072586707e-05,
+      "loss": 0.1858,
+      "step": 1716
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.928570256952392e-05,
+      "loss": 0.2079,
+      "step": 1717
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9265869769787618e-05,
+      "loss": 0.1791,
+      "step": 1718
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9246037692906e-05,
+      "loss": 0.2221,
+      "step": 1719
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.922620635840653e-05,
+      "loss": 0.2178,
+      "step": 1720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.920637578581596e-05,
+      "loss": 0.1854,
+      "step": 1721
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9186545994660287e-05,
+      "loss": 0.1708,
+      "step": 1722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.916671700446472e-05,
+      "loss": 0.1781,
+      "step": 1723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9146888834753716e-05,
+      "loss": 0.1851,
+      "step": 1724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.912706150505088e-05,
+      "loss": 0.1977,
+      "step": 1725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.910723503487903e-05,
+      "loss": 0.2212,
+      "step": 1726
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.908740944376011e-05,
+      "loss": 0.2025,
+      "step": 1727
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.906758475121521e-05,
+      "loss": 0.155,
+      "step": 1728
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.904776097676454e-05,
+      "loss": 0.1885,
+      "step": 1729
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.90279381399274e-05,
+      "loss": 0.1953,
+      "step": 1730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.900811626022215e-05,
+      "loss": 0.1779,
+      "step": 1731
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8988295357166247e-05,
+      "loss": 0.1989,
+      "step": 1732
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.896847545027615e-05,
+      "loss": 0.1883,
+      "step": 1733
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8948656559067356e-05,
+      "loss": 0.1815,
+      "step": 1734
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.892883870305436e-05,
+      "loss": 0.1986,
+      "step": 1735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8909021901750617e-05,
+      "loss": 0.19,
+      "step": 1736
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.888920617466858e-05,
+      "loss": 0.2057,
+      "step": 1737
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.886939154131962e-05,
+      "loss": 0.1733,
+      "step": 1738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8849578021214025e-05,
+      "loss": 0.1745,
+      "step": 1739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.882976563386102e-05,
+      "loss": 0.2065,
+      "step": 1740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8809954398768673e-05,
+      "loss": 0.2192,
+      "step": 1741
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8790144335443938e-05,
+      "loss": 0.2005,
+      "step": 1742
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8770335463392638e-05,
+      "loss": 0.1877,
+      "step": 1743
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8750527802119364e-05,
+      "loss": 0.1881,
+      "step": 1744
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.873072137112758e-05,
+      "loss": 0.2333,
+      "step": 1745
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8710916189919492e-05,
+      "loss": 0.1725,
+      "step": 1746
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8691112277996092e-05,
+      "loss": 0.1952,
+      "step": 1747
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.867130965485714e-05,
+      "loss": 0.1855,
+      "step": 1748
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8651508340001086e-05,
+      "loss": 0.2103,
+      "step": 1749
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8631708352925136e-05,
+      "loss": 0.1787,
+      "step": 1750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.861190971312515e-05,
+      "loss": 0.1882,
+      "step": 1751
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8592112440095687e-05,
+      "loss": 0.2255,
+      "step": 1752
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.857231655332996e-05,
+      "loss": 0.1613,
+      "step": 1753
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8552522072319798e-05,
+      "loss": 0.2292,
+      "step": 1754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8532729016555658e-05,
+      "loss": 0.1861,
+      "step": 1755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8512937405526603e-05,
+      "loss": 0.162,
+      "step": 1756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8493147258720253e-05,
+      "loss": 0.1938,
+      "step": 1757
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8473358595622808e-05,
+      "loss": 0.1915,
+      "step": 1758
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.845357143571898e-05,
+      "loss": 0.1816,
+      "step": 1759
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.843378579849203e-05,
+      "loss": 0.1952,
+      "step": 1760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.84140017034237e-05,
+      "loss": 0.1614,
+      "step": 1761
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8394219169994226e-05,
+      "loss": 0.1818,
+      "step": 1762
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.837443821768229e-05,
+      "loss": 0.1537,
+      "step": 1763
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.835465886596504e-05,
+      "loss": 0.2207,
+      "step": 1764
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8334881134318027e-05,
+      "loss": 0.1711,
+      "step": 1765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8315105042215218e-05,
+      "loss": 0.1904,
+      "step": 1766
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8295330609128958e-05,
+      "loss": 0.215,
+      "step": 1767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.827555785452997e-05,
+      "loss": 0.1825,
+      "step": 1768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8255786797887317e-05,
+      "loss": 0.2108,
+      "step": 1769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8236017458668388e-05,
+      "loss": 0.2129,
+      "step": 1770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.821624985633889e-05,
+      "loss": 0.1997,
+      "step": 1771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196484010362795e-05,
+      "loss": 0.1977,
+      "step": 1772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8176719940202384e-05,
+      "loss": 0.1898,
+      "step": 1773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8156957665318168e-05,
+      "loss": 0.21,
+      "step": 1774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8137197205168878e-05,
+      "loss": 0.1793,
+      "step": 1775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8117438579211484e-05,
+      "loss": 0.1738,
+      "step": 1776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8097681806901142e-05,
+      "loss": 0.2028,
+      "step": 1777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8077926907691165e-05,
+      "loss": 0.22,
+      "step": 1778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8058173901033048e-05,
+      "loss": 0.1898,
+      "step": 1779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8038422806376396e-05,
+      "loss": 0.1839,
+      "step": 1780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.801867364316896e-05,
+      "loss": 0.2137,
+      "step": 1781
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7998926430856566e-05,
+      "loss": 0.1888,
+      "step": 1782
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7979181188883126e-05,
+      "loss": 0.167,
+      "step": 1783
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7959437936690627e-05,
+      "loss": 0.165,
+      "step": 1784
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7939696693719065e-05,
+      "loss": 0.1865,
+      "step": 1785
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7919957479406486e-05,
+      "loss": 0.1913,
+      "step": 1786
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7900220313188938e-05,
+      "loss": 0.1792,
+      "step": 1787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.788048521450042e-05,
+      "loss": 0.1886,
+      "step": 1788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7860752202772934e-05,
+      "loss": 0.175,
+      "step": 1789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7841021297436414e-05,
+      "loss": 0.1827,
+      "step": 1790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.78212925179187e-05,
+      "loss": 0.1924,
+      "step": 1791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7801565883645574e-05,
+      "loss": 0.1924,
+      "step": 1792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.778184141404068e-05,
+      "loss": 0.1901,
+      "step": 1793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7762119128525533e-05,
+      "loss": 0.2091,
+      "step": 1794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.774239904651952e-05,
+      "loss": 0.1793,
+      "step": 1795
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.772268118743982e-05,
+      "loss": 0.2052,
+      "step": 1796
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.770296557070146e-05,
+      "loss": 0.1637,
+      "step": 1797
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7683252215717234e-05,
+      "loss": 0.2185,
+      "step": 1798
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7663541141897714e-05,
+      "loss": 0.239,
+      "step": 1799
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.764383236865125e-05,
+      "loss": 0.1685,
+      "step": 1800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.762412591538389e-05,
+      "loss": 0.1852,
+      "step": 1801
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7604421801499415e-05,
+      "loss": 0.1874,
+      "step": 1802
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7584720046399316e-05,
+      "loss": 0.1913,
+      "step": 1803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7565020669482732e-05,
+      "loss": 0.1764,
+      "step": 1804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7545323690146486e-05,
+      "loss": 0.1778,
+      "step": 1805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7525629127785023e-05,
+      "loss": 0.1859,
+      "step": 1806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7505937001790417e-05,
+      "loss": 0.2106,
+      "step": 1807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.748624733155235e-05,
+      "loss": 0.1773,
+      "step": 1808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7466560136458065e-05,
+      "loss": 0.1763,
+      "step": 1809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7446875435892393e-05,
+      "loss": 0.168,
+      "step": 1810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.742719324923768e-05,
+      "loss": 0.1949,
+      "step": 1811
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7407513595873826e-05,
+      "loss": 0.1563,
+      "step": 1812
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.738783649517822e-05,
+      "loss": 0.2141,
+      "step": 1813
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.736816196652574e-05,
+      "loss": 0.1698,
+      "step": 1814
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7348490029288723e-05,
+      "loss": 0.1876,
+      "step": 1815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7328820702836986e-05,
+      "loss": 0.1632,
+      "step": 1816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7309154006537737e-05,
+      "loss": 0.2092,
+      "step": 1817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7289489959755615e-05,
+      "loss": 0.2029,
+      "step": 1818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7269828581852636e-05,
+      "loss": 0.1926,
+      "step": 1819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.725016989218821e-05,
+      "loss": 0.191,
+      "step": 1820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.723051391011909e-05,
+      "loss": 0.1934,
+      "step": 1821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7210860654999345e-05,
+      "loss": 0.1704,
+      "step": 1822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.719121014618039e-05,
+      "loss": 0.173,
+      "step": 1823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7171562403010907e-05,
+      "loss": 0.183,
+      "step": 1824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7151917444836876e-05,
+      "loss": 0.1903,
+      "step": 1825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7132275291001534e-05,
+      "loss": 0.1915,
+      "step": 1826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.711263596084533e-05,
+      "loss": 0.1873,
+      "step": 1827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7092999473705972e-05,
+      "loss": 0.1526,
+      "step": 1828
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7073365848918347e-05,
+      "loss": 0.1678,
+      "step": 1829
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.705373510581451e-05,
+      "loss": 0.1832,
+      "step": 1830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7034107263723718e-05,
+      "loss": 0.1907,
+      "step": 1831
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.701448234197232e-05,
+      "loss": 0.1672,
+      "step": 1832
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6994860359883834e-05,
+      "loss": 0.1721,
+      "step": 1833
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6975241336778875e-05,
+      "loss": 0.1883,
+      "step": 1834
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6955625291975116e-05,
+      "loss": 0.1838,
+      "step": 1835
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6936012244787336e-05,
+      "loss": 0.1693,
+      "step": 1836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.691640221452733e-05,
+      "loss": 0.1807,
+      "step": 1837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.689679522050394e-05,
+      "loss": 0.1851,
+      "step": 1838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6877191282023032e-05,
+      "loss": 0.1701,
+      "step": 1839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6857590418387417e-05,
+      "loss": 0.1847,
+      "step": 1840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6837992648896927e-05,
+      "loss": 0.2096,
+      "step": 1841
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.681839799284833e-05,
+      "loss": 0.2197,
+      "step": 1842
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6798806469535312e-05,
+      "loss": 0.1806,
+      "step": 1843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6779218098248503e-05,
+      "loss": 0.1685,
+      "step": 1844
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.675963289827541e-05,
+      "loss": 0.1741,
+      "step": 1845
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6740050888900416e-05,
+      "loss": 0.1965,
+      "step": 1846
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.672047208940479e-05,
+      "loss": 0.2569,
+      "step": 1847
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6700896519066588e-05,
+      "loss": 0.1698,
+      "step": 1848
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.668132419716075e-05,
+      "loss": 0.1912,
+      "step": 1849
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6661755142958962e-05,
+      "loss": 0.188,
+      "step": 1850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6642189375729722e-05,
+      "loss": 0.1737,
+      "step": 1851
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6622626914738297e-05,
+      "loss": 0.1928,
+      "step": 1852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.660306777924667e-05,
+      "loss": 0.1485,
+      "step": 1853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6583511988513573e-05,
+      "loss": 0.182,
+      "step": 1854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6563959561794444e-05,
+      "loss": 0.1987,
+      "step": 1855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.65444105183414e-05,
+      "loss": 0.1803,
+      "step": 1856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.652486487740323e-05,
+      "loss": 0.1831,
+      "step": 1857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6505322658225362e-05,
+      "loss": 0.1767,
+      "step": 1858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6485783880049872e-05,
+      "loss": 0.2055,
+      "step": 1859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.646624856211545e-05,
+      "loss": 0.1986,
+      "step": 1860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.644671672365736e-05,
+      "loss": 0.1757,
+      "step": 1861
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6427188383907453e-05,
+      "loss": 0.1883,
+      "step": 1862
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.640766356209412e-05,
+      "loss": 0.213,
+      "step": 1863
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.638814227744232e-05,
+      "loss": 0.1776,
+      "step": 1864
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.63686245491735e-05,
+      "loss": 0.1654,
+      "step": 1865
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.634911039650561e-05,
+      "loss": 0.1654,
+      "step": 1866
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6329599838653082e-05,
+      "loss": 0.1912,
+      "step": 1867
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.631009289482682e-05,
+      "loss": 0.1849,
+      "step": 1868
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6290589584234154e-05,
+      "loss": 0.2256,
+      "step": 1869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.627108992607885e-05,
+      "loss": 0.2118,
+      "step": 1870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.625159393956105e-05,
+      "loss": 0.2334,
+      "step": 1871
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.623210164387732e-05,
+      "loss": 0.199,
+      "step": 1872
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6212613058220566e-05,
+      "loss": 0.1713,
+      "step": 1873
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6193128201780045e-05,
+      "loss": 0.1732,
+      "step": 1874
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6173647093741345e-05,
+      "loss": 0.2,
+      "step": 1875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6154169753286363e-05,
+      "loss": 0.1835,
+      "step": 1876
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6134696199593284e-05,
+      "loss": 0.1695,
+      "step": 1877
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6115226451836575e-05,
+      "loss": 0.1671,
+      "step": 1878
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.609576052918693e-05,
+      "loss": 0.1676,
+      "step": 1879
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.607629845081131e-05,
+      "loss": 0.1571,
+      "step": 1880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.605684023587287e-05,
+      "loss": 0.1928,
+      "step": 1881
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6037385903530952e-05,
+      "loss": 0.1949,
+      "step": 1882
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6017935472941114e-05,
+      "loss": 0.1713,
+      "step": 1883
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.599848896325502e-05,
+      "loss": 0.1816,
+      "step": 1884
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.597904639362051e-05,
+      "loss": 0.2214,
+      "step": 1885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.595960778318154e-05,
+      "loss": 0.179,
+      "step": 1886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.594017315107815e-05,
+      "loss": 0.1907,
+      "step": 1887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5920742516446484e-05,
+      "loss": 0.1962,
+      "step": 1888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.590131589841874e-05,
+      "loss": 0.1839,
+      "step": 1889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.588189331612315e-05,
+      "loss": 0.1611,
+      "step": 1890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5862474788684003e-05,
+      "loss": 0.1926,
+      "step": 1891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.584306033522155e-05,
+      "loss": 0.2253,
+      "step": 1892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5823649974852082e-05,
+      "loss": 0.1456,
+      "step": 1893
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5804243726687825e-05,
+      "loss": 0.1991,
+      "step": 1894
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.578484160983696e-05,
+      "loss": 0.168,
+      "step": 1895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5765443643403612e-05,
+      "loss": 0.1924,
+      "step": 1896
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.574604984648781e-05,
+      "loss": 0.2229,
+      "step": 1897
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5726660238185475e-05,
+      "loss": 0.1724,
+      "step": 1898
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5707274837588425e-05,
+      "loss": 0.2026,
+      "step": 1899
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5687893663784308e-05,
+      "loss": 0.1703,
+      "step": 1900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.566851673585662e-05,
+      "loss": 0.2164,
+      "step": 1901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5649144072884684e-05,
+      "loss": 0.1652,
+      "step": 1902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5629775693943606e-05,
+      "loss": 0.2192,
+      "step": 1903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.56104116181043e-05,
+      "loss": 0.1985,
+      "step": 1904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5591051864433416e-05,
+      "loss": 0.1973,
+      "step": 1905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.557169645199336e-05,
+      "loss": 0.1946,
+      "step": 1906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5552345399842275e-05,
+      "loss": 0.1608,
+      "step": 1907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.553299872703399e-05,
+      "loss": 0.1758,
+      "step": 1908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5513656452618032e-05,
+      "loss": 0.1846,
+      "step": 1909
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5494318595639597e-05,
+      "loss": 0.1808,
+      "step": 1910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5474985175139524e-05,
+      "loss": 0.2065,
+      "step": 1911
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5455656210154307e-05,
+      "loss": 0.1574,
+      "step": 1912
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.543633171971602e-05,
+      "loss": 0.1596,
+      "step": 1913
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5417011722852352e-05,
+      "loss": 0.1769,
+      "step": 1914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5397696238586568e-05,
+      "loss": 0.2173,
+      "step": 1915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.537838528593748e-05,
+      "loss": 0.1662,
+      "step": 1916
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5359078883919446e-05,
+      "loss": 0.1948,
+      "step": 1917
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5339777051542337e-05,
+      "loss": 0.191,
+      "step": 1918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.532047980781152e-05,
+      "loss": 0.1805,
+      "step": 1919
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5301187171727866e-05,
+      "loss": 0.1716,
+      "step": 1920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5281899162287685e-05,
+      "loss": 0.2085,
+      "step": 1921
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5262615798482746e-05,
+      "loss": 0.186,
+      "step": 1922
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5243337099300227e-05,
+      "loss": 0.1751,
+      "step": 1923
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5224063083722738e-05,
+      "loss": 0.192,
+      "step": 1924
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5204793770728262e-05,
+      "loss": 0.1704,
+      "step": 1925
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5185529179290139e-05,
+      "loss": 0.2002,
+      "step": 1926
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5166269328377092e-05,
+      "loss": 0.2114,
+      "step": 1927
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5147014236953156e-05,
+      "loss": 0.1939,
+      "step": 1928
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5127763923977674e-05,
+      "loss": 0.1851,
+      "step": 1929
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5108518408405301e-05,
+      "loss": 0.1813,
+      "step": 1930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5089277709185948e-05,
+      "loss": 0.1849,
+      "step": 1931
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5070041845264805e-05,
+      "loss": 0.1847,
+      "step": 1932
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5050810835582292e-05,
+      "loss": 0.2102,
+      "step": 1933
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5031584699074038e-05,
+      "loss": 0.174,
+      "step": 1934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5012363454670902e-05,
+      "loss": 0.156,
+      "step": 1935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4993147121298885e-05,
+      "loss": 0.1758,
+      "step": 1936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4973935717879193e-05,
+      "loss": 0.2099,
+      "step": 1937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4954729263328158e-05,
+      "loss": 0.181,
+      "step": 1938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4935527776557231e-05,
+      "loss": 0.2354,
+      "step": 1939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4916331276472994e-05,
+      "loss": 0.2251,
+      "step": 1940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4897139781977102e-05,
+      "loss": 0.1681,
+      "step": 1941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4877953311966282e-05,
+      "loss": 0.1834,
+      "step": 1942
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4858771885332333e-05,
+      "loss": 0.2,
+      "step": 1943
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4839595520962048e-05,
+      "loss": 0.1715,
+      "step": 1944
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4820424237737281e-05,
+      "loss": 0.1813,
+      "step": 1945
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.480125805453486e-05,
+      "loss": 0.2244,
+      "step": 1946
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4782096990226582e-05,
+      "loss": 0.2125,
+      "step": 1947
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4762941063679228e-05,
+      "loss": 0.1658,
+      "step": 1948
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4743790293754501e-05,
+      "loss": 0.1883,
+      "step": 1949
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4724644699309027e-05,
+      "loss": 0.1782,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4705504299194359e-05,
+      "loss": 0.1667,
+      "step": 1951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4686369112256906e-05,
+      "loss": 0.152,
+      "step": 1952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.466723915733796e-05,
+      "loss": 0.1655,
+      "step": 1953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4648114453273662e-05,
+      "loss": 0.1757,
+      "step": 1954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4628995018894965e-05,
+      "loss": 0.1644,
+      "step": 1955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4609880873027659e-05,
+      "loss": 0.1767,
+      "step": 1956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4590772034492313e-05,
+      "loss": 0.1974,
+      "step": 1957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4571668522104264e-05,
+      "loss": 0.1722,
+      "step": 1958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4552570354673625e-05,
+      "loss": 0.1976,
+      "step": 1959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4533477551005225e-05,
+      "loss": 0.1731,
+      "step": 1960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4514390129898624e-05,
+      "loss": 0.1853,
+      "step": 1961
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4495308110148071e-05,
+      "loss": 0.2045,
+      "step": 1962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4476231510542507e-05,
+      "loss": 0.1859,
+      "step": 1963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4457160349865543e-05,
+      "loss": 0.1638,
+      "step": 1964
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4438094646895411e-05,
+      "loss": 0.1956,
+      "step": 1965
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4419034420404985e-05,
+      "loss": 0.2081,
+      "step": 1966
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4399979689161756e-05,
+      "loss": 0.1708,
+      "step": 1967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4380930471927781e-05,
+      "loss": 0.1763,
+      "step": 1968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4361886787459703e-05,
+      "loss": 0.1984,
+      "step": 1969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.434284865450871e-05,
+      "loss": 0.1928,
+      "step": 1970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4323816091820523e-05,
+      "loss": 0.1903,
+      "step": 1971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4304789118135403e-05,
+      "loss": 0.1479,
+      "step": 1972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4285767752188062e-05,
+      "loss": 0.209,
+      "step": 1973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4266752012707734e-05,
+      "loss": 0.1331,
+      "step": 1974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4247741918418082e-05,
+      "loss": 0.1728,
+      "step": 1975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4228737488037227e-05,
+      "loss": 0.1562,
+      "step": 1976
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4209738740277718e-05,
+      "loss": 0.1948,
+      "step": 1977
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4190745693846484e-05,
+      "loss": 0.1859,
+      "step": 1978
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4171758367444868e-05,
+      "loss": 0.1867,
+      "step": 1979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.415277677976857e-05,
+      "loss": 0.1569,
+      "step": 1980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.413380094950763e-05,
+      "loss": 0.1897,
+      "step": 1981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4114830895346434e-05,
+      "loss": 0.1567,
+      "step": 1982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4095866635963664e-05,
+      "loss": 0.2145,
+      "step": 1983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4076908190032317e-05,
+      "loss": 0.1603,
+      "step": 1984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4057955576219653e-05,
+      "loss": 0.1677,
+      "step": 1985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4039008813187182e-05,
+      "loss": 0.1672,
+      "step": 1986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4020067919590684e-05,
+      "loss": 0.2046,
+      "step": 1987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4001132914080113e-05,
+      "loss": 0.1714,
+      "step": 1988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3982203815299664e-05,
+      "loss": 0.1896,
+      "step": 1989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3963280641887704e-05,
+      "loss": 0.1698,
+      "step": 1990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3944363412476753e-05,
+      "loss": 0.1897,
+      "step": 1991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3925452145693505e-05,
+      "loss": 0.2019,
+      "step": 1992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.390654686015876e-05,
+      "loss": 0.1802,
+      "step": 1993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.388764757448743e-05,
+      "loss": 0.1769,
+      "step": 1994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.386875430728854e-05,
+      "loss": 0.2115,
+      "step": 1995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3849867077165158e-05,
+      "loss": 0.196,
+      "step": 1996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3830985902714433e-05,
+      "loss": 0.24,
+      "step": 1997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3812110802527548e-05,
+      "loss": 0.2156,
+      "step": 1998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3793241795189682e-05,
+      "loss": 0.1565,
+      "step": 1999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3774378899280052e-05,
+      "loss": 0.1897,
+      "step": 2000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3755522133371822e-05,
+      "loss": 0.1504,
+      "step": 2001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3736671516032139e-05,
+      "loss": 0.1798,
+      "step": 2002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3717827065822102e-05,
+      "loss": 0.1753,
+      "step": 2003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.369898880129672e-05,
+      "loss": 0.1834,
+      "step": 2004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3680156741004913e-05,
+      "loss": 0.179,
+      "step": 2005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.366133090348952e-05,
+      "loss": 0.1751,
+      "step": 2006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3642511307287208e-05,
+      "loss": 0.1678,
+      "step": 2007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3623697970928538e-05,
+      "loss": 0.1767,
+      "step": 2008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.360489091293788e-05,
+      "loss": 0.1861,
+      "step": 2009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3586090151833435e-05,
+      "loss": 0.1767,
+      "step": 2010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3567295706127213e-05,
+      "loss": 0.2207,
+      "step": 2011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3548507594324982e-05,
+      "loss": 0.1705,
+      "step": 2012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3529725834926297e-05,
+      "loss": 0.1692,
+      "step": 2013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3510950446424435e-05,
+      "loss": 0.1733,
+      "step": 2014
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3492181447306414e-05,
+      "loss": 0.1682,
+      "step": 2015
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3473418856052974e-05,
+      "loss": 0.1884,
+      "step": 2016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3454662691138511e-05,
+      "loss": 0.1781,
+      "step": 2017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.343591297103112e-05,
+      "loss": 0.2057,
+      "step": 2018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3417169714192553e-05,
+      "loss": 0.1824,
+      "step": 2019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3398432939078174e-05,
+      "loss": 0.2186,
+      "step": 2020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3379702664136991e-05,
+      "loss": 0.1809,
+      "step": 2021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.336097890781158e-05,
+      "loss": 0.1563,
+      "step": 2022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3342261688538139e-05,
+      "loss": 0.185,
+      "step": 2023
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3323551024746401e-05,
+      "loss": 0.1841,
+      "step": 2024
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.330484693485965e-05,
+      "loss": 0.1919,
+      "step": 2025
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3286149437294702e-05,
+      "loss": 0.191,
+      "step": 2026
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267458550461867e-05,
+      "loss": 0.163,
+      "step": 2027
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324877429276497e-05,
+      "loss": 0.1496,
+      "step": 2028
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230096682601299e-05,
+      "loss": 0.1801,
+      "step": 2029
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211425738361577e-05,
+      "loss": 0.1582,
+      "step": 2030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3192761478429994e-05,
+      "loss": 0.1816,
+      "step": 2031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3174103921184144e-05,
+      "loss": 0.1702,
+      "step": 2032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155453084995014e-05,
+      "loss": 0.1651,
+      "step": 2033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313680898822699e-05,
+      "loss": 0.1706,
+      "step": 2034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3118171649237801e-05,
+      "loss": 0.1909,
+      "step": 2035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099541086378542e-05,
+      "loss": 0.1693,
+      "step": 2036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3080917317993631e-05,
+      "loss": 0.1787,
+      "step": 2037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3062300362420782e-05,
+      "loss": 0.1787,
+      "step": 2038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3043690237991031e-05,
+      "loss": 0.173,
+      "step": 2039
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3025086963028641e-05,
+      "loss": 0.1792,
+      "step": 2040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.300649055585118e-05,
+      "loss": 0.1805,
+      "step": 2041
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2987901034769426e-05,
+      "loss": 0.181,
+      "step": 2042
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2969318418087372e-05,
+      "loss": 0.1687,
+      "step": 2043
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2950742724102238e-05,
+      "loss": 0.1702,
+      "step": 2044
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.293217397110441e-05,
+      "loss": 0.1918,
+      "step": 2045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2913612177377433e-05,
+      "loss": 0.1681,
+      "step": 2046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.289505736119802e-05,
+      "loss": 0.1649,
+      "step": 2047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2876509540835996e-05,
+      "loss": 0.1615,
+      "step": 2048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2857968734554308e-05,
+      "loss": 0.1979,
+      "step": 2049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2839434960608992e-05,
+      "loss": 0.1548,
+      "step": 2050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2820908237249153e-05,
+      "loss": 0.1833,
+      "step": 2051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2802388582716972e-05,
+      "loss": 0.1563,
+      "step": 2052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2783876015247658e-05,
+      "loss": 0.2082,
+      "step": 2053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2765370553069434e-05,
+      "loss": 0.1771,
+      "step": 2054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2746872214403548e-05,
+      "loss": 0.1709,
+      "step": 2055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2728381017464208e-05,
+      "loss": 0.2192,
+      "step": 2056
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.270989698045861e-05,
+      "loss": 0.1489,
+      "step": 2057
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2691420121586906e-05,
+      "loss": 0.1946,
+      "step": 2058
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2672950459042144e-05,
+      "loss": 0.1675,
+      "step": 2059
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2654488011010328e-05,
+      "loss": 0.2113,
+      "step": 2060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.263603279567033e-05,
+      "loss": 0.1985,
+      "step": 2061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2617584831193913e-05,
+      "loss": 0.1885,
+      "step": 2062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2599144135745705e-05,
+      "loss": 0.2173,
+      "step": 2063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2580710727483158e-05,
+      "loss": 0.1668,
+      "step": 2064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2562284624556567e-05,
+      "loss": 0.1987,
+      "step": 2065
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2543865845109035e-05,
+      "loss": 0.1866,
+      "step": 2066
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.252545440727643e-05,
+      "loss": 0.1858,
+      "step": 2067
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.250705032918742e-05,
+      "loss": 0.1901,
+      "step": 2068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2488653628963405e-05,
+      "loss": 0.1786,
+      "step": 2069
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.247026432471853e-05,
+      "loss": 0.2036,
+      "step": 2070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.245188243455967e-05,
+      "loss": 0.1559,
+      "step": 2071
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2433507976586372e-05,
+      "loss": 0.2035,
+      "step": 2072
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2415140968890888e-05,
+      "loss": 0.2149,
+      "step": 2073
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2396781429558113e-05,
+      "loss": 0.1704,
+      "step": 2074
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2378429376665615e-05,
+      "loss": 0.1693,
+      "step": 2075
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2360084828283578e-05,
+      "loss": 0.1728,
+      "step": 2076
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2341747802474782e-05,
+      "loss": 0.1582,
+      "step": 2077
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2323418317294617e-05,
+      "loss": 0.1816,
+      "step": 2078
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2305096390791055e-05,
+      "loss": 0.1696,
+      "step": 2079
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2286782041004606e-05,
+      "loss": 0.1931,
+      "step": 2080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2268475285968333e-05,
+      "loss": 0.1983,
+      "step": 2081
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2250176143707806e-05,
+      "loss": 0.1807,
+      "step": 2082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.223188463224112e-05,
+      "loss": 0.1543,
+      "step": 2083
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2213600769578851e-05,
+      "loss": 0.1881,
+      "step": 2084
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2195324573724023e-05,
+      "loss": 0.1992,
+      "step": 2085
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2177056062672141e-05,
+      "loss": 0.1477,
+      "step": 2086
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2158795254411114e-05,
+      "loss": 0.1557,
+      "step": 2087
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2140542166921296e-05,
+      "loss": 0.1767,
+      "step": 2088
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2122296818175419e-05,
+      "loss": 0.1764,
+      "step": 2089
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2104059226138594e-05,
+      "loss": 0.1975,
+      "step": 2090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.208582940876831e-05,
+      "loss": 0.1632,
+      "step": 2091
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2067607384014391e-05,
+      "loss": 0.1895,
+      "step": 2092
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2049393169818982e-05,
+      "loss": 0.1707,
+      "step": 2093
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2031186784116549e-05,
+      "loss": 0.157,
+      "step": 2094
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2012988244833839e-05,
+      "loss": 0.1663,
+      "step": 2095
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1994797569889882e-05,
+      "loss": 0.1722,
+      "step": 2096
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.197661477719597e-05,
+      "loss": 0.1723,
+      "step": 2097
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1958439884655609e-05,
+      "loss": 0.1563,
+      "step": 2098
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1940272910164556e-05,
+      "loss": 0.1825,
+      "step": 2099
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1922113871610753e-05,
+      "loss": 0.1765,
+      "step": 2100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1903962786874322e-05,
+      "loss": 0.1624,
+      "step": 2101
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.188581967382759e-05,
+      "loss": 0.1785,
+      "step": 2102
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.186768455033498e-05,
+      "loss": 0.2008,
+      "step": 2103
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1849557434253094e-05,
+      "loss": 0.1718,
+      "step": 2104
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1831438343430632e-05,
+      "loss": 0.1879,
+      "step": 2105
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1813327295708382e-05,
+      "loss": 0.1746,
+      "step": 2106
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1795224308919237e-05,
+      "loss": 0.191,
+      "step": 2107
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.177712940088813e-05,
+      "loss": 0.1649,
+      "step": 2108
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1759042589432046e-05,
+      "loss": 0.1727,
+      "step": 2109
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1740963892360015e-05,
+      "loss": 0.164,
+      "step": 2110
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1722893327473041e-05,
+      "loss": 0.1624,
+      "step": 2111
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1704830912564162e-05,
+      "loss": 0.179,
+      "step": 2112
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1686776665418356e-05,
+      "loss": 0.1779,
+      "step": 2113
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1668730603812574e-05,
+      "loss": 0.1825,
+      "step": 2114
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1650692745515723e-05,
+      "loss": 0.1832,
+      "step": 2115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1632663108288597e-05,
+      "loss": 0.1518,
+      "step": 2116
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1614641709883937e-05,
+      "loss": 0.2005,
+      "step": 2117
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.159662856804633e-05,
+      "loss": 0.1664,
+      "step": 2118
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1578623700512259e-05,
+      "loss": 0.1658,
+      "step": 2119
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1560627125010058e-05,
+      "loss": 0.177,
+      "step": 2120
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.154263885925988e-05,
+      "loss": 0.208,
+      "step": 2121
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1524658920973728e-05,
+      "loss": 0.1744,
+      "step": 2122
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1506687327855372e-05,
+      "loss": 0.1836,
+      "step": 2123
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1488724097600372e-05,
+      "loss": 0.1848,
+      "step": 2124
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1470769247896076e-05,
+      "loss": 0.1866,
+      "step": 2125
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1452822796421548e-05,
+      "loss": 0.1884,
+      "step": 2126
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1434884760847607e-05,
+      "loss": 0.1665,
+      "step": 2127
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1416955158836793e-05,
+      "loss": 0.1773,
+      "step": 2128
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1399034008043296e-05,
+      "loss": 0.199,
+      "step": 2129
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1381121326113027e-05,
+      "loss": 0.1675,
+      "step": 2130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1363217130683556e-05,
+      "loss": 0.1643,
+      "step": 2131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1345321439384071e-05,
+      "loss": 0.1568,
+      "step": 2132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.132743426983542e-05,
+      "loss": 0.189,
+      "step": 2133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.130955563965003e-05,
+      "loss": 0.2113,
+      "step": 2134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.129168556643193e-05,
+      "loss": 0.1667,
+      "step": 2135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1273824067776733e-05,
+      "loss": 0.2009,
+      "step": 2136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1255971161271595e-05,
+      "loss": 0.1857,
+      "step": 2137
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.123812686449523e-05,
+      "loss": 0.1932,
+      "step": 2138
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1220291195017856e-05,
+      "loss": 0.1669,
+      "step": 2139
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1202464170401197e-05,
+      "loss": 0.2138,
+      "step": 2140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1184645808198482e-05,
+      "loss": 0.1563,
+      "step": 2141
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.116683612595439e-05,
+      "loss": 0.2024,
+      "step": 2142
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1149035141205066e-05,
+      "loss": 0.1712,
+      "step": 2143
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.11312428714781e-05,
+      "loss": 0.18,
+      "step": 2144
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1113459334292482e-05,
+      "loss": 0.1561,
+      "step": 2145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1095684547158609e-05,
+      "loss": 0.1953,
+      "step": 2146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1077918527578256e-05,
+      "loss": 0.1884,
+      "step": 2147
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1060161293044578e-05,
+      "loss": 0.1851,
+      "step": 2148
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1042412861042089e-05,
+      "loss": 0.154,
+      "step": 2149
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1024673249046608e-05,
+      "loss": 0.1607,
+      "step": 2150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.100694247452529e-05,
+      "loss": 0.1711,
+      "step": 2151
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0989220554936569e-05,
+      "loss": 0.1625,
+      "step": 2152
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.097150750773018e-05,
+      "loss": 0.1854,
+      "step": 2153
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0953803350347125e-05,
+      "loss": 0.1778,
+      "step": 2154
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0936108100219633e-05,
+      "loss": 0.1614,
+      "step": 2155
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0918421774771162e-05,
+      "loss": 0.1442,
+      "step": 2156
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0900744391416411e-05,
+      "loss": 0.1896,
+      "step": 2157
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0883075967561237e-05,
+      "loss": 0.1803,
+      "step": 2158
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0865416520602712e-05,
+      "loss": 0.1934,
+      "step": 2159
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.084776606792903e-05,
+      "loss": 0.1714,
+      "step": 2160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.083012462691957e-05,
+      "loss": 0.1751,
+      "step": 2161
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0812492214944807e-05,
+      "loss": 0.1714,
+      "step": 2162
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.079486884936633e-05,
+      "loss": 0.1743,
+      "step": 2163
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0777254547536837e-05,
+      "loss": 0.1928,
+      "step": 2164
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0759649326800078e-05,
+      "loss": 0.1849,
+      "step": 2165
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0742053204490892e-05,
+      "loss": 0.1651,
+      "step": 2166
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.072446619793513e-05,
+      "loss": 0.1778,
+      "step": 2167
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0706888324449673e-05,
+      "loss": 0.1711,
+      "step": 2168
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0689319601342421e-05,
+      "loss": 0.1866,
+      "step": 2169
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0671760045912264e-05,
+      "loss": 0.1794,
+      "step": 2170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0654209675449053e-05,
+      "loss": 0.1934,
+      "step": 2171
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0636668507233612e-05,
+      "loss": 0.1716,
+      "step": 2172
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.061913655853767e-05,
+      "loss": 0.1685,
+      "step": 2173
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0601613846623913e-05,
+      "loss": 0.2036,
+      "step": 2174
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0584100388745933e-05,
+      "loss": 0.1948,
+      "step": 2175
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0566596202148178e-05,
+      "loss": 0.1983,
+      "step": 2176
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0549101304066003e-05,
+      "loss": 0.1709,
+      "step": 2177
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0531615711725596e-05,
+      "loss": 0.167,
+      "step": 2178
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0514139442343976e-05,
+      "loss": 0.1586,
+      "step": 2179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0496672513129008e-05,
+      "loss": 0.1838,
+      "step": 2180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0479214941279335e-05,
+      "loss": 0.1782,
+      "step": 2181
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0461766743984405e-05,
+      "loss": 0.1676,
+      "step": 2182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0444327938424426e-05,
+      "loss": 0.1837,
+      "step": 2183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0426898541770352e-05,
+      "loss": 0.1638,
+      "step": 2184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0409478571183894e-05,
+      "loss": 0.2011,
+      "step": 2185
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0392068043817449e-05,
+      "loss": 0.1763,
+      "step": 2186
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0374666976814147e-05,
+      "loss": 0.1766,
+      "step": 2187
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.035727538730781e-05,
+      "loss": 0.191,
+      "step": 2188
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0339893292422872e-05,
+      "loss": 0.1908,
+      "step": 2189
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0322520709274473e-05,
+      "loss": 0.183,
+      "step": 2190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0305157654968374e-05,
+      "loss": 0.1938,
+      "step": 2191
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.028780414660094e-05,
+      "loss": 0.1475,
+      "step": 2192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0270460201259157e-05,
+      "loss": 0.1794,
+      "step": 2193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0253125836020577e-05,
+      "loss": 0.1676,
+      "step": 2194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0235801067953317e-05,
+      "loss": 0.1627,
+      "step": 2195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0218485914116069e-05,
+      "loss": 0.1536,
+      "step": 2196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0201180391558027e-05,
+      "loss": 0.1622,
+      "step": 2197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0183884517318934e-05,
+      "loss": 0.1705,
+      "step": 2198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0166598308429006e-05,
+      "loss": 0.1742,
+      "step": 2199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0149321781908945e-05,
+      "loss": 0.1665,
+      "step": 2200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.013205495476994e-05,
+      "loss": 0.1651,
+      "step": 2201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0114797844013602e-05,
+      "loss": 0.1835,
+      "step": 2202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0097550466631999e-05,
+      "loss": 0.1507,
+      "step": 2203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0080312839607606e-05,
+      "loss": 0.1771,
+      "step": 2204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0063084979913294e-05,
+      "loss": 0.1574,
+      "step": 2205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0045866904512319e-05,
+      "loss": 0.1816,
+      "step": 2206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.002865863035829e-05,
+      "loss": 0.1566,
+      "step": 2207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0011460174395193e-05,
+      "loss": 0.1887,
+      "step": 2208
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994271553557335e-06,
+      "loss": 0.1778,
+      "step": 2209
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.977092784769329e-06,
+      "loss": 0.1546,
+      "step": 2210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.959923884946095e-06,
+      "loss": 0.1612,
+      "step": 2211
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.942764870992827e-06,
+      "loss": 0.1942,
+      "step": 2212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.925615759804996e-06,
+      "loss": 0.1749,
+      "step": 2213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.908476568268332e-06,
+      "loss": 0.1661,
+      "step": 2214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.891347313258778e-06,
+      "loss": 0.194,
+      "step": 2215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.874228011642486e-06,
+      "loss": 0.1377,
+      "step": 2216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.85711868027584e-06,
+      "loss": 0.1723,
+      "step": 2217
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.84001933600537e-06,
+      "loss": 0.1789,
+      "step": 2218
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.822929995667814e-06,
+      "loss": 0.1781,
+      "step": 2219
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.80585067609001e-06,
+      "loss": 0.1941,
+      "step": 2220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.788781394088977e-06,
+      "loss": 0.1597,
+      "step": 2221
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.77172216647182e-06,
+      "loss": 0.1788,
+      "step": 2222
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.754673010035744e-06,
+      "loss": 0.1566,
+      "step": 2223
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.737633941568065e-06,
+      "loss": 0.1433,
+      "step": 2224
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.72060497784613e-06,
+      "loss": 0.1563,
+      "step": 2225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.703586135637366e-06,
+      "loss": 0.2051,
+      "step": 2226
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.686577431699219e-06,
+      "loss": 0.1573,
+      "step": 2227
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.669578882779145e-06,
+      "loss": 0.1505,
+      "step": 2228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.65259050561462e-06,
+      "loss": 0.1805,
+      "step": 2229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.6356123169331e-06,
+      "loss": 0.1377,
+      "step": 2230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.618644333451994e-06,
+      "loss": 0.1708,
+      "step": 2231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.601686571878694e-06,
+      "loss": 0.1542,
+      "step": 2232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.584739048910476e-06,
+      "loss": 0.192,
+      "step": 2233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.56780178123458e-06,
+      "loss": 0.1618,
+      "step": 2234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.550874785528141e-06,
+      "loss": 0.1748,
+      "step": 2235
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.533958078458157e-06,
+      "loss": 0.2219,
+      "step": 2236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.517051676681526e-06,
+      "loss": 0.1744,
+      "step": 2237
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.500155596844973e-06,
+      "loss": 0.1909,
+      "step": 2238
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.483269855585064e-06,
+      "loss": 0.1664,
+      "step": 2239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.466394469528208e-06,
+      "loss": 0.1759,
+      "step": 2240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.449529455290583e-06,
+      "loss": 0.1558,
+      "step": 2241
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.432674829478189e-06,
+      "loss": 0.1684,
+      "step": 2242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.415830608686774e-06,
+      "loss": 0.1592,
+      "step": 2243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.398996809501837e-06,
+      "loss": 0.1947,
+      "step": 2244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.382173448498644e-06,
+      "loss": 0.223,
+      "step": 2245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.365360542242148e-06,
+      "loss": 0.1584,
+      "step": 2246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.348558107287032e-06,
+      "loss": 0.2048,
+      "step": 2247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.33176616017767e-06,
+      "loss": 0.1625,
+      "step": 2248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.314984717448095e-06,
+      "loss": 0.1566,
+      "step": 2249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.298213795621997e-06,
+      "loss": 0.1719,
+      "step": 2250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.28145341121271e-06,
+      "loss": 0.1688,
+      "step": 2251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.2647035807232e-06,
+      "loss": 0.1903,
+      "step": 2252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.247964320646043e-06,
+      "loss": 0.1338,
+      "step": 2253
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.231235647463392e-06,
+      "loss": 0.1499,
+      "step": 2254
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.214517577646977e-06,
+      "loss": 0.1707,
+      "step": 2255
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.197810127658112e-06,
+      "loss": 0.1716,
+      "step": 2256
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.181113313947613e-06,
+      "loss": 0.1463,
+      "step": 2257
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.16442715295587e-06,
+      "loss": 0.1926,
+      "step": 2258
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.147751661112747e-06,
+      "loss": 0.1724,
+      "step": 2259
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.13108685483761e-06,
+      "loss": 0.1679,
+      "step": 2260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.114432750539326e-06,
+      "loss": 0.1815,
+      "step": 2261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.097789364616192e-06,
+      "loss": 0.1581,
+      "step": 2262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.08115671345598e-06,
+      "loss": 0.1563,
+      "step": 2263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.064534813435867e-06,
+      "loss": 0.2003,
+      "step": 2264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.047923680922473e-06,
+      "loss": 0.1802,
+      "step": 2265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.031323332271786e-06,
+      "loss": 0.1532,
+      "step": 2266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.014733783829186e-06,
+      "loss": 0.1901,
+      "step": 2267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.998155051929427e-06,
+      "loss": 0.1559,
+      "step": 2268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.981587152896616e-06,
+      "loss": 0.1456,
+      "step": 2269
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.965030103044174e-06,
+      "loss": 0.2046,
+      "step": 2270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.948483918674854e-06,
+      "loss": 0.1768,
+      "step": 2271
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.931948616080696e-06,
+      "loss": 0.1824,
+      "step": 2272
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.915424211543046e-06,
+      "loss": 0.1425,
+      "step": 2273
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.898910721332517e-06,
+      "loss": 0.1753,
+      "step": 2274
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.882408161708949e-06,
+      "loss": 0.1505,
+      "step": 2275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.865916548921463e-06,
+      "loss": 0.1832,
+      "step": 2276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.849435899208345e-06,
+      "loss": 0.1729,
+      "step": 2277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.832966228797133e-06,
+      "loss": 0.1813,
+      "step": 2278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.816507553904545e-06,
+      "loss": 0.1415,
+      "step": 2279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.800059890736454e-06,
+      "loss": 0.1559,
+      "step": 2280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.783623255487915e-06,
+      "loss": 0.188,
+      "step": 2281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.767197664343106e-06,
+      "loss": 0.1792,
+      "step": 2282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.750783133475328e-06,
+      "loss": 0.1875,
+      "step": 2283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.734379679047011e-06,
+      "loss": 0.1906,
+      "step": 2284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.717987317209663e-06,
+      "loss": 0.1669,
+      "step": 2285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.701606064103878e-06,
+      "loss": 0.1742,
+      "step": 2286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.685235935859309e-06,
+      "loss": 0.1956,
+      "step": 2287
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.668876948594644e-06,
+      "loss": 0.1851,
+      "step": 2288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.652529118417627e-06,
+      "loss": 0.2086,
+      "step": 2289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.636192461424988e-06,
+      "loss": 0.1852,
+      "step": 2290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.619866993702471e-06,
+      "loss": 0.1835,
+      "step": 2291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.603552731324812e-06,
+      "loss": 0.1655,
+      "step": 2292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.58724969035569e-06,
+      "loss": 0.1553,
+      "step": 2293
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.570957886847742e-06,
+      "loss": 0.1545,
+      "step": 2294
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.554677336842558e-06,
+      "loss": 0.1773,
+      "step": 2295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.538408056370622e-06,
+      "loss": 0.1637,
+      "step": 2296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.522150061451344e-06,
+      "loss": 0.1304,
+      "step": 2297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.50590336809301e-06,
+      "loss": 0.1423,
+      "step": 2298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.489667992292764e-06,
+      "loss": 0.1694,
+      "step": 2299
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.473443950036643e-06,
+      "loss": 0.1734,
+      "step": 2300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.457231257299487e-06,
+      "loss": 0.182,
+      "step": 2301
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.44102993004499e-06,
+      "loss": 0.1825,
+      "step": 2302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.424839984225635e-06,
+      "loss": 0.179,
+      "step": 2303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.408661435782701e-06,
+      "loss": 0.1456,
+      "step": 2304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.392494300646259e-06,
+      "loss": 0.195,
+      "step": 2305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.376338594735123e-06,
+      "loss": 0.1612,
+      "step": 2306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.360194333956863e-06,
+      "loss": 0.1887,
+      "step": 2307
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.344061534207795e-06,
+      "loss": 0.2005,
+      "step": 2308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.327940211372918e-06,
+      "loss": 0.1759,
+      "step": 2309
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.311830381325953e-06,
+      "loss": 0.1571,
+      "step": 2310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.295732059929289e-06,
+      "loss": 0.1579,
+      "step": 2311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.279645263033997e-06,
+      "loss": 0.153,
+      "step": 2312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.263570006479807e-06,
+      "loss": 0.1919,
+      "step": 2313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.247506306095063e-06,
+      "loss": 0.1858,
+      "step": 2314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.231454177696745e-06,
+      "loss": 0.1724,
+      "step": 2315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.21541363709043e-06,
+      "loss": 0.2048,
+      "step": 2316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.199384700070294e-06,
+      "loss": 0.1781,
+      "step": 2317
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.183367382419096e-06,
+      "loss": 0.1473,
+      "step": 2318
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.167361699908124e-06,
+      "loss": 0.1874,
+      "step": 2319
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.151367668297248e-06,
+      "loss": 0.19,
+      "step": 2320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.135385303334835e-06,
+      "loss": 0.1836,
+      "step": 2321
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.119414620757769e-06,
+      "loss": 0.1929,
+      "step": 2322
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.103455636291453e-06,
+      "loss": 0.2,
+      "step": 2323
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.087508365649743e-06,
+      "loss": 0.1668,
+      "step": 2324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.071572824534987e-06,
+      "loss": 0.1566,
+      "step": 2325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.055649028637966e-06,
+      "loss": 0.1806,
+      "step": 2326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.039736993637895e-06,
+      "loss": 0.1715,
+      "step": 2327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.023836735202424e-06,
+      "loss": 0.1961,
+      "step": 2328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.007948268987589e-06,
+      "loss": 0.181,
+      "step": 2329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.992071610637837e-06,
+      "loss": 0.1589,
+      "step": 2330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.976206775785968e-06,
+      "loss": 0.1754,
+      "step": 2331
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.960353780053145e-06,
+      "loss": 0.1511,
+      "step": 2332
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.944512639048879e-06,
+      "loss": 0.1831,
+      "step": 2333
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.928683368371013e-06,
+      "loss": 0.188,
+      "step": 2334
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.912865983605683e-06,
+      "loss": 0.1662,
+      "step": 2335
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.897060500327347e-06,
+      "loss": 0.1794,
+      "step": 2336
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.881266934098724e-06,
+      "loss": 0.1327,
+      "step": 2337
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.865485300470796e-06,
+      "loss": 0.1778,
+      "step": 2338
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.849715614982826e-06,
+      "loss": 0.1828,
+      "step": 2339
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.833957893162272e-06,
+      "loss": 0.1638,
+      "step": 2340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.818212150524848e-06,
+      "loss": 0.1469,
+      "step": 2341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.802478402574451e-06,
+      "loss": 0.1855,
+      "step": 2342
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.786756664803166e-06,
+      "loss": 0.161,
+      "step": 2343
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.771046952691273e-06,
+      "loss": 0.1601,
+      "step": 2344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.75534928170718e-06,
+      "loss": 0.1771,
+      "step": 2345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.73966366730747e-06,
+      "loss": 0.1939,
+      "step": 2346
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.723990124936853e-06,
+      "loss": 0.1707,
+      "step": 2347
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.708328670028112e-06,
+      "loss": 0.1723,
+      "step": 2348
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.692679318002175e-06,
+      "loss": 0.1869,
+      "step": 2349
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.677042084268024e-06,
+      "loss": 0.2032,
+      "step": 2350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.661416984222725e-06,
+      "loss": 0.1726,
+      "step": 2351
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.645804033251401e-06,
+      "loss": 0.1813,
+      "step": 2352
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.630203246727191e-06,
+      "loss": 0.1837,
+      "step": 2353
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.614614640011264e-06,
+      "loss": 0.1699,
+      "step": 2354
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.599038228452813e-06,
+      "loss": 0.1637,
+      "step": 2355
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.583474027388999e-06,
+      "loss": 0.1766,
+      "step": 2356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.567922052144985e-06,
+      "loss": 0.1863,
+      "step": 2357
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.552382318033871e-06,
+      "loss": 0.188,
+      "step": 2358
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.536854840356713e-06,
+      "loss": 0.1916,
+      "step": 2359
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.521339634402516e-06,
+      "loss": 0.1771,
+      "step": 2360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.505836715448169e-06,
+      "loss": 0.1584,
+      "step": 2361
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.490346098758499e-06,
+      "loss": 0.1681,
+      "step": 2362
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.474867799586181e-06,
+      "loss": 0.1724,
+      "step": 2363
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.459401833171802e-06,
+      "loss": 0.1855,
+      "step": 2364
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.443948214743777e-06,
+      "loss": 0.1625,
+      "step": 2365
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.428506959518364e-06,
+      "loss": 0.188,
+      "step": 2366
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.413078082699663e-06,
+      "loss": 0.1675,
+      "step": 2367
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.397661599479584e-06,
+      "loss": 0.1566,
+      "step": 2368
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.38225752503782e-06,
+      "loss": 0.1484,
+      "step": 2369
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.366865874541851e-06,
+      "loss": 0.1773,
+      "step": 2370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.351486663146925e-06,
+      "loss": 0.1467,
+      "step": 2371
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.336119905996046e-06,
+      "loss": 0.1423,
+      "step": 2372
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.320765618219961e-06,
+      "loss": 0.1659,
+      "step": 2373
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.305423814937122e-06,
+      "loss": 0.1489,
+      "step": 2374
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.290094511253696e-06,
+      "loss": 0.1478,
+      "step": 2375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.2747777222635396e-06,
+      "loss": 0.1617,
+      "step": 2376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.2594734630481925e-06,
+      "loss": 0.1673,
+      "step": 2377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.244181748676866e-06,
+      "loss": 0.1619,
+      "step": 2378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.228902594206391e-06,
+      "loss": 0.2048,
+      "step": 2379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.213636014681265e-06,
+      "loss": 0.1624,
+      "step": 2380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.198382025133579e-06,
+      "loss": 0.1849,
+      "step": 2381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.183140640583029e-06,
+      "loss": 0.1529,
+      "step": 2382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.167911876036917e-06,
+      "loss": 0.1607,
+      "step": 2383
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.152695746490099e-06,
+      "loss": 0.1757,
+      "step": 2384
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.13749226692501e-06,
+      "loss": 0.1748,
+      "step": 2385
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.1223014523116126e-06,
+      "loss": 0.1561,
+      "step": 2386
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.107123317607399e-06,
+      "loss": 0.1787,
+      "step": 2387
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.091957877757392e-06,
+      "loss": 0.1894,
+      "step": 2388
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.076805147694094e-06,
+      "loss": 0.1664,
+      "step": 2389
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.061665142337517e-06,
+      "loss": 0.1569,
+      "step": 2390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.046537876595121e-06,
+      "loss": 0.1356,
+      "step": 2391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.031423365361827e-06,
+      "loss": 0.1964,
+      "step": 2392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.016321623520006e-06,
+      "loss": 0.1727,
+      "step": 2393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.001232665939459e-06,
+      "loss": 0.1556,
+      "step": 2394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.986156507477379e-06,
+      "loss": 0.1817,
+      "step": 2395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.971093162978382e-06,
+      "loss": 0.1648,
+      "step": 2396
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.956042647274446e-06,
+      "loss": 0.171,
+      "step": 2397
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.9410049751849175e-06,
+      "loss": 0.1941,
+      "step": 2398
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.925980161516519e-06,
+      "loss": 0.1645,
+      "step": 2399
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.910968221063281e-06,
+      "loss": 0.1798,
+      "step": 2400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.895969168606593e-06,
+      "loss": 0.1546,
+      "step": 2401
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.880983018915126e-06,
+      "loss": 0.1425,
+      "step": 2402
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.866009786744849e-06,
+      "loss": 0.1735,
+      "step": 2403
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.851049486839036e-06,
+      "loss": 0.1654,
+      "step": 2404
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.836102133928195e-06,
+      "loss": 0.1641,
+      "step": 2405
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.821167742730108e-06,
+      "loss": 0.1561,
+      "step": 2406
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.806246327949801e-06,
+      "loss": 0.1755,
+      "step": 2407
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.791337904279487e-06,
+      "loss": 0.1456,
+      "step": 2408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.776442486398627e-06,
+      "loss": 0.1693,
+      "step": 2409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.761560088973846e-06,
+      "loss": 0.1566,
+      "step": 2410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.7466907266589685e-06,
+      "loss": 0.199,
+      "step": 2411
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.7318344140949865e-06,
+      "loss": 0.1733,
+      "step": 2412
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.716991165910023e-06,
+      "loss": 0.1689,
+      "step": 2413
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.702160996719349e-06,
+      "loss": 0.1465,
+      "step": 2414
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.687343921125353e-06,
+      "loss": 0.1667,
+      "step": 2415
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.6725399537175365e-06,
+      "loss": 0.1602,
+      "step": 2416
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.6577491090725e-06,
+      "loss": 0.1609,
+      "step": 2417
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.642971401753908e-06,
+      "loss": 0.1519,
+      "step": 2418
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.628206846312491e-06,
+      "loss": 0.1678,
+      "step": 2419
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.613455457286045e-06,
+      "loss": 0.1455,
+      "step": 2420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.59871724919938e-06,
+      "loss": 0.1883,
+      "step": 2421
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.583992236564354e-06,
+      "loss": 0.1828,
+      "step": 2422
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.569280433879801e-06,
+      "loss": 0.1627,
+      "step": 2423
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.554581855631579e-06,
+      "loss": 0.1582,
+      "step": 2424
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.539896516292501e-06,
+      "loss": 0.1704,
+      "step": 2425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.5252244303223525e-06,
+      "loss": 0.1465,
+      "step": 2426
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.510565612167876e-06,
+      "loss": 0.1679,
+      "step": 2427
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.495920076262736e-06,
+      "loss": 0.1647,
+      "step": 2428
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.481287837027537e-06,
+      "loss": 0.1923,
+      "step": 2429
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.4666689088697755e-06,
+      "loss": 0.1799,
+      "step": 2430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.45206330618384e-06,
+      "loss": 0.1697,
+      "step": 2431
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.437471043351014e-06,
+      "loss": 0.1503,
+      "step": 2432
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.422892134739436e-06,
+      "loss": 0.1619,
+      "step": 2433
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.408326594704098e-06,
+      "loss": 0.1711,
+      "step": 2434
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.393774437586822e-06,
+      "loss": 0.1617,
+      "step": 2435
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.3792356777162515e-06,
+      "loss": 0.1561,
+      "step": 2436
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.364710329407853e-06,
+      "loss": 0.1754,
+      "step": 2437
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.3501984069638855e-06,
+      "loss": 0.1906,
+      "step": 2438
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.335699924673369e-06,
+      "loss": 0.1789,
+      "step": 2439
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.3212148968121136e-06,
+      "loss": 0.1946,
+      "step": 2440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.306743337642664e-06,
+      "loss": 0.1433,
+      "step": 2441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.292285261414308e-06,
+      "loss": 0.1649,
+      "step": 2442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.277840682363067e-06,
+      "loss": 0.1363,
+      "step": 2443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.263409614711657e-06,
+      "loss": 0.155,
+      "step": 2444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.2489920726695065e-06,
+      "loss": 0.1575,
+      "step": 2445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.234588070432712e-06,
+      "loss": 0.175,
+      "step": 2446
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.2201976221840385e-06,
+      "loss": 0.1704,
+      "step": 2447
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.205820742092923e-06,
+      "loss": 0.1301,
+      "step": 2448
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.191457444315418e-06,
+      "loss": 0.1514,
+      "step": 2449
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.177107742994219e-06,
+      "loss": 0.1547,
+      "step": 2450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.1627716522586455e-06,
+      "loss": 0.1779,
+      "step": 2451
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.148449186224572e-06,
+      "loss": 0.1443,
+      "step": 2452
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.134140358994503e-06,
+      "loss": 0.1483,
+      "step": 2453
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.119845184657482e-06,
+      "loss": 0.1427,
+      "step": 2454
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.105563677289132e-06,
+      "loss": 0.1703,
+      "step": 2455
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.091295850951615e-06,
+      "loss": 0.189,
+      "step": 2456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.0770417196936105e-06,
+      "loss": 0.1501,
+      "step": 2457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.062801297550314e-06,
+      "loss": 0.163,
+      "step": 2458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.04857459854344e-06,
+      "loss": 0.1194,
+      "step": 2459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.034361636681166e-06,
+      "loss": 0.164,
+      "step": 2460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.0201624259581695e-06,
+      "loss": 0.1832,
+      "step": 2461
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.005976980355568e-06,
+      "loss": 0.1652,
+      "step": 2462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.991805313840923e-06,
+      "loss": 0.1729,
+      "step": 2463
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.977647440368257e-06,
+      "loss": 0.1358,
+      "step": 2464
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.9635033738779725e-06,
+      "loss": 0.1497,
+      "step": 2465
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.949373128296914e-06,
+      "loss": 0.1636,
+      "step": 2466
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.935256717538283e-06,
+      "loss": 0.187,
+      "step": 2467
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.921154155501689e-06,
+      "loss": 0.1594,
+      "step": 2468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.907065456073087e-06,
+      "loss": 0.1938,
+      "step": 2469
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.89299063312478e-06,
+      "loss": 0.1718,
+      "step": 2470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.878929700515421e-06,
+      "loss": 0.1625,
+      "step": 2471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.864882672089989e-06,
+      "loss": 0.1871,
+      "step": 2472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.850849561679752e-06,
+      "loss": 0.1512,
+      "step": 2473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.83683038310229e-06,
+      "loss": 0.1649,
+      "step": 2474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.822825150161451e-06,
+      "loss": 0.1611,
+      "step": 2475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.808833876647366e-06,
+      "loss": 0.1527,
+      "step": 2476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.794856576336425e-06,
+      "loss": 0.1647,
+      "step": 2477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.780893262991241e-06,
+      "loss": 0.1757,
+      "step": 2478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.766943950360659e-06,
+      "loss": 0.1666,
+      "step": 2479
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.753008652179757e-06,
+      "loss": 0.1495,
+      "step": 2480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.739087382169786e-06,
+      "loss": 0.1662,
+      "step": 2481
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.725180154038213e-06,
+      "loss": 0.159,
+      "step": 2482
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.711286981478647e-06,
+      "loss": 0.1625,
+      "step": 2483
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.6974078781708904e-06,
+      "loss": 0.1457,
+      "step": 2484
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.68354285778087e-06,
+      "loss": 0.1547,
+      "step": 2485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.669691933960648e-06,
+      "loss": 0.1491,
+      "step": 2486
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.655855120348419e-06,
+      "loss": 0.1318,
+      "step": 2487
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.642032430568467e-06,
+      "loss": 0.1692,
+      "step": 2488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.628223878231189e-06,
+      "loss": 0.1678,
+      "step": 2489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.614429476933048e-06,
+      "loss": 0.1771,
+      "step": 2490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.600649240256566e-06,
+      "loss": 0.1765,
+      "step": 2491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.586883181770337e-06,
+      "loss": 0.1439,
+      "step": 2492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.57313131502899e-06,
+      "loss": 0.1759,
+      "step": 2493
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.559393653573166e-06,
+      "loss": 0.1403,
+      "step": 2494
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.5456702109295435e-06,
+      "loss": 0.1708,
+      "step": 2495
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.531961000610766e-06,
+      "loss": 0.1672,
+      "step": 2496
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.5182660361154915e-06,
+      "loss": 0.1593,
+      "step": 2497
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.504585330928354e-06,
+      "loss": 0.1442,
+      "step": 2498
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.490918898519917e-06,
+      "loss": 0.1547,
+      "step": 2499
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.477266752346726e-06,
+      "loss": 0.1828,
+      "step": 2500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.463628905851235e-06,
+      "loss": 0.1612,
+      "step": 2501
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.450005372461822e-06,
+      "loss": 0.15,
+      "step": 2502
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4363961655927855e-06,
+      "loss": 0.1606,
+      "step": 2503
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4228012986442976e-06,
+      "loss": 0.1652,
+      "step": 2504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.409220785002431e-06,
+      "loss": 0.1863,
+      "step": 2505
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.395654638039112e-06,
+      "loss": 0.1605,
+      "step": 2506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.382102871112116e-06,
+      "loss": 0.1464,
+      "step": 2507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.368565497565079e-06,
+      "loss": 0.1697,
+      "step": 2508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.355042530727443e-06,
+      "loss": 0.1423,
+      "step": 2509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.341533983914479e-06,
+      "loss": 0.1551,
+      "step": 2510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.328039870427266e-06,
+      "loss": 0.1563,
+      "step": 2511
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.314560203552648e-06,
+      "loss": 0.1683,
+      "step": 2512
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.301094996563261e-06,
+      "loss": 0.1674,
+      "step": 2513
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.287644262717491e-06,
+      "loss": 0.1654,
+      "step": 2514
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.274208015259488e-06,
+      "loss": 0.146,
+      "step": 2515
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.260786267419138e-06,
+      "loss": 0.1828,
+      "step": 2516
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.247379032412034e-06,
+      "loss": 0.1861,
+      "step": 2517
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.233986323439488e-06,
+      "loss": 0.1491,
+      "step": 2518
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.220608153688518e-06,
+      "loss": 0.1504,
+      "step": 2519
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.207244536331808e-06,
+      "loss": 0.1815,
+      "step": 2520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.193895484527731e-06,
+      "loss": 0.1557,
+      "step": 2521
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.180561011420306e-06,
+      "loss": 0.1765,
+      "step": 2522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.167241130139196e-06,
+      "loss": 0.1303,
+      "step": 2523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.153935853799712e-06,
+      "loss": 0.1387,
+      "step": 2524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.1406451955027624e-06,
+      "loss": 0.1359,
+      "step": 2525
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.127369168334886e-06,
+      "loss": 0.1858,
+      "step": 2526
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.11410778536819e-06,
+      "loss": 0.1724,
+      "step": 2527
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.100861059660387e-06,
+      "loss": 0.1524,
+      "step": 2528
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.087629004254739e-06,
+      "loss": 0.172,
+      "step": 2529
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.074411632180063e-06,
+      "loss": 0.1381,
+      "step": 2530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.061208956450734e-06,
+      "loss": 0.1608,
+      "step": 2531
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.048020990066649e-06,
+      "loss": 0.1321,
+      "step": 2532
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.034847746013216e-06,
+      "loss": 0.173,
+      "step": 2533
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.021689237261349e-06,
+      "loss": 0.153,
+      "step": 2534
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.008545476767448e-06,
+      "loss": 0.182,
+      "step": 2535
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.995416477473405e-06,
+      "loss": 0.1462,
+      "step": 2536
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.982302252306577e-06,
+      "loss": 0.1376,
+      "step": 2537
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.969202814179754e-06,
+      "loss": 0.1592,
+      "step": 2538
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.956118175991196e-06,
+      "loss": 0.182,
+      "step": 2539
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.943048350624551e-06,
+      "loss": 0.1503,
+      "step": 2540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.929993350948914e-06,
+      "loss": 0.1626,
+      "step": 2541
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.91695318981878e-06,
+      "loss": 0.1454,
+      "step": 2542
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9039278800740155e-06,
+      "loss": 0.1718,
+      "step": 2543
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.890917434539881e-06,
+      "loss": 0.1377,
+      "step": 2544
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.877921866026991e-06,
+      "loss": 0.1644,
+      "step": 2545
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.864941187331307e-06,
+      "loss": 0.137,
+      "step": 2546
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.851975411234149e-06,
+      "loss": 0.1196,
+      "step": 2547
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.839024550502141e-06,
+      "loss": 0.1603,
+      "step": 2548
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.8260886178872365e-06,
+      "loss": 0.1762,
+      "step": 2549
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.8131676261266865e-06,
+      "loss": 0.1579,
+      "step": 2550
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.800261587943016e-06,
+      "loss": 0.1614,
+      "step": 2551
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7873705160440545e-06,
+      "loss": 0.1689,
+      "step": 2552
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.774494423122866e-06,
+      "loss": 0.1465,
+      "step": 2553
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.761633321857784e-06,
+      "loss": 0.1895,
+      "step": 2554
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.748787224912381e-06,
+      "loss": 0.1317,
+      "step": 2555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.735956144935443e-06,
+      "loss": 0.1684,
+      "step": 2556
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.723140094560976e-06,
+      "loss": 0.1682,
+      "step": 2557
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.710339086408193e-06,
+      "loss": 0.1661,
+      "step": 2558
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.697553133081485e-06,
+      "loss": 0.1647,
+      "step": 2559
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.684782247170432e-06,
+      "loss": 0.1427,
+      "step": 2560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.672026441249766e-06,
+      "loss": 0.1285,
+      "step": 2561
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.659285727879373e-06,
+      "loss": 0.1643,
+      "step": 2562
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.646560119604289e-06,
+      "loss": 0.1524,
+      "step": 2563
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.633849628954661e-06,
+      "loss": 0.1454,
+      "step": 2564
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.621154268445764e-06,
+      "loss": 0.1697,
+      "step": 2565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.608474050577971e-06,
+      "loss": 0.1714,
+      "step": 2566
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.595808987836734e-06,
+      "loss": 0.1469,
+      "step": 2567
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.583159092692604e-06,
+      "loss": 0.1504,
+      "step": 2568
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.570524377601175e-06,
+      "loss": 0.1684,
+      "step": 2569
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.55790485500311e-06,
+      "loss": 0.1861,
+      "step": 2570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.545300537324113e-06,
+      "loss": 0.1339,
+      "step": 2571
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.5327114369749055e-06,
+      "loss": 0.1463,
+      "step": 2572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.520137566351233e-06,
+      "loss": 0.1605,
+      "step": 2573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.507578937833836e-06,
+      "loss": 0.1665,
+      "step": 2574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4950355637884594e-06,
+      "loss": 0.1634,
+      "step": 2575
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.482507456565828e-06,
+      "loss": 0.1361,
+      "step": 2576
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.469994628501624e-06,
+      "loss": 0.1623,
+      "step": 2577
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.457497091916492e-06,
+      "loss": 0.154,
+      "step": 2578
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.44501485911601e-06,
+      "loss": 0.1683,
+      "step": 2579
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.432547942390703e-06,
+      "loss": 0.1508,
+      "step": 2580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.420096354016013e-06,
+      "loss": 0.1732,
+      "step": 2581
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.407660106252276e-06,
+      "loss": 0.1729,
+      "step": 2582
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.395239211344726e-06,
+      "loss": 0.162,
+      "step": 2583
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.382833681523497e-06,
+      "loss": 0.1507,
+      "step": 2584
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.37044352900357e-06,
+      "loss": 0.1514,
+      "step": 2585
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.358068765984809e-06,
+      "loss": 0.1428,
+      "step": 2586
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.3457094046518964e-06,
+      "loss": 0.1534,
+      "step": 2587
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3333654571743815e-06,
+      "loss": 0.1759,
+      "step": 2588
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.321036935706611e-06,
+      "loss": 0.1745,
+      "step": 2589
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.3087238523877484e-06,
+      "loss": 0.1608,
+      "step": 2590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2964262193417714e-06,
+      "loss": 0.1364,
+      "step": 2591
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.284144048677419e-06,
+      "loss": 0.1772,
+      "step": 2592
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.271877352488234e-06,
+      "loss": 0.1306,
+      "step": 2593
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2596261428525e-06,
+      "loss": 0.1491,
+      "step": 2594
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.247390431833253e-06,
+      "loss": 0.1667,
+      "step": 2595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.235170231478283e-06,
+      "loss": 0.1582,
+      "step": 2596
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.222965553820104e-06,
+      "loss": 0.162,
+      "step": 2597
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.21077641087593e-06,
+      "loss": 0.139,
+      "step": 2598
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.198602814647705e-06,
+      "loss": 0.1244,
+      "step": 2599
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1864447771220316e-06,
+      "loss": 0.1549,
+      "step": 2600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.17430231027022e-06,
+      "loss": 0.1476,
+      "step": 2601
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.162175426048247e-06,
+      "loss": 0.1425,
+      "step": 2602
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.150064136396727e-06,
+      "loss": 0.1644,
+      "step": 2603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.1379684532409434e-06,
+      "loss": 0.1636,
+      "step": 2604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.125888388490795e-06,
+      "loss": 0.1435,
+      "step": 2605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.1138239540408056e-06,
+      "loss": 0.167,
+      "step": 2606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.101775161770116e-06,
+      "loss": 0.1558,
+      "step": 2607
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.089742023542456e-06,
+      "loss": 0.2037,
+      "step": 2608
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.077724551206151e-06,
+      "loss": 0.1459,
+      "step": 2609
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.065722756594095e-06,
+      "loss": 0.1546,
+      "step": 2610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.053736651523741e-06,
+      "loss": 0.1599,
+      "step": 2611
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0417662477971055e-06,
+      "loss": 0.154,
+      "step": 2612
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.029811557200731e-06,
+      "loss": 0.1441,
+      "step": 2613
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.017872591505703e-06,
+      "loss": 0.1695,
+      "step": 2614
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.005949362467614e-06,
+      "loss": 0.1513,
+      "step": 2615
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.994041881826563e-06,
+      "loss": 0.1474,
+      "step": 2616
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982150161307136e-06,
+      "loss": 0.1684,
+      "step": 2617
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.970274212618421e-06,
+      "loss": 0.1265,
+      "step": 2618
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.958414047453951e-06,
+      "loss": 0.1465,
+      "step": 2619
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.946569677491739e-06,
+      "loss": 0.1837,
+      "step": 2620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.934741114394234e-06,
+      "loss": 0.1826,
+      "step": 2621
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.922928369808314e-06,
+      "loss": 0.1317,
+      "step": 2622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.911131455365307e-06,
+      "loss": 0.1584,
+      "step": 2623
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.899350382680924e-06,
+      "loss": 0.1357,
+      "step": 2624
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.887585163355305e-06,
+      "loss": 0.1593,
+      "step": 2625
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.875835808972957e-06,
+      "loss": 0.1625,
+      "step": 2626
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.864102331102772e-06,
+      "loss": 0.1863,
+      "step": 2627
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.852384741298025e-06,
+      "loss": 0.1515,
+      "step": 2628
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.840683051096321e-06,
+      "loss": 0.1608,
+      "step": 2629
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.828997272019632e-06,
+      "loss": 0.1896,
+      "step": 2630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.817327415574257e-06,
+      "loss": 0.1423,
+      "step": 2631
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.805673493250808e-06,
+      "loss": 0.1439,
+      "step": 2632
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.794035516524217e-06,
+      "loss": 0.1296,
+      "step": 2633
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.7824134968537074e-06,
+      "loss": 0.1528,
+      "step": 2634
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.770807445682798e-06,
+      "loss": 0.1116,
+      "step": 2635
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.75921737443929e-06,
+      "loss": 0.1887,
+      "step": 2636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7476432945352347e-06,
+      "loss": 0.1658,
+      "step": 2637
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.736085217366945e-06,
+      "loss": 0.19,
+      "step": 2638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7245431543149723e-06,
+      "loss": 0.1849,
+      "step": 2639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.713017116744111e-06,
+      "loss": 0.1453,
+      "step": 2640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7015071160033712e-06,
+      "loss": 0.1492,
+      "step": 2641
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.690013163425965e-06,
+      "loss": 0.1585,
+      "step": 2642
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6785352703293152e-06,
+      "loss": 0.1594,
+      "step": 2643
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6670734480150215e-06,
+      "loss": 0.1671,
+      "step": 2644
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6556277077688586e-06,
+      "loss": 0.1427,
+      "step": 2645
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.644198060860782e-06,
+      "loss": 0.1584,
+      "step": 2646
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.632784518544876e-06,
+      "loss": 0.1681,
+      "step": 2647
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6213870920593964e-06,
+      "loss": 0.1337,
+      "step": 2648
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6100057926267095e-06,
+      "loss": 0.1524,
+      "step": 2649
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.5986406314533005e-06,
+      "loss": 0.1809,
+      "step": 2650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.5872916197297868e-06,
+      "loss": 0.1651,
+      "step": 2651
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.5759587686308584e-06,
+      "loss": 0.1516,
+      "step": 2652
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5646420893153133e-06,
+      "loss": 0.1695,
+      "step": 2653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5533415929260116e-06,
+      "loss": 0.1522,
+      "step": 2654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.542057290589882e-06,
+      "loss": 0.142,
+      "step": 2655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.530789193417916e-06,
+      "loss": 0.1884,
+      "step": 2656
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5195373125051456e-06,
+      "loss": 0.1623,
+      "step": 2657
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5083016589306264e-06,
+      "loss": 0.1575,
+      "step": 2658
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4970822437574503e-06,
+      "loss": 0.1497,
+      "step": 2659
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.485879078032708e-06,
+      "loss": 0.1832,
+      "step": 2660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4746921727874907e-06,
+      "loss": 0.1522,
+      "step": 2661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4635215390368958e-06,
+      "loss": 0.1517,
+      "step": 2662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4523671877799723e-06,
+      "loss": 0.1678,
+      "step": 2663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4412291299997634e-06,
+      "loss": 0.1969,
+      "step": 2664
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.430107376663252e-06,
+      "loss": 0.1552,
+      "step": 2665
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.419001938721365e-06,
+      "loss": 0.1584,
+      "step": 2666
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4079128271089836e-06,
+      "loss": 0.1437,
+      "step": 2667
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3968400527448895e-06,
+      "loss": 0.1591,
+      "step": 2668
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3857836265317956e-06,
+      "loss": 0.1796,
+      "step": 2669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3747435593563216e-06,
+      "loss": 0.1532,
+      "step": 2670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3637198620889477e-06,
+      "loss": 0.135,
+      "step": 2671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.352712545584076e-06,
+      "loss": 0.1683,
+      "step": 2672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3417216206799474e-06,
+      "loss": 0.1472,
+      "step": 2673
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3307470981986835e-06,
+      "loss": 0.1317,
+      "step": 2674
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.319788988946253e-06,
+      "loss": 0.1444,
+      "step": 2675
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.308847303712455e-06,
+      "loss": 0.159,
+      "step": 2676
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.297922053270919e-06,
+      "loss": 0.1468,
+      "step": 2677
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2870132483790894e-06,
+      "loss": 0.1516,
+      "step": 2678
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.276120899778228e-06,
+      "loss": 0.1577,
+      "step": 2679
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2652450181933905e-06,
+      "loss": 0.1454,
+      "step": 2680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2543856143334106e-06,
+      "loss": 0.1622,
+      "step": 2681
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.243542698890898e-06,
+      "loss": 0.1413,
+      "step": 2682
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.232716282542241e-06,
+      "loss": 0.1477,
+      "step": 2683
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.221906375947561e-06,
+      "loss": 0.149,
+      "step": 2684
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2111129897507465e-06,
+      "loss": 0.1475,
+      "step": 2685
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.200336134579396e-06,
+      "loss": 0.1549,
+      "step": 2686
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1895758210448523e-06,
+      "loss": 0.1728,
+      "step": 2687
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.178832059742156e-06,
+      "loss": 0.1634,
+      "step": 2688
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1681048612500496e-06,
+      "loss": 0.1454,
+      "step": 2689
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1573942361309817e-06,
+      "loss": 0.165,
+      "step": 2690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1467001949310607e-06,
+      "loss": 0.1356,
+      "step": 2691
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.13602274818009e-06,
+      "loss": 0.1774,
+      "step": 2692
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1253619063915154e-06,
+      "loss": 0.1775,
+      "step": 2693
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.114717680062431e-06,
+      "loss": 0.1353,
+      "step": 2694
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.104090079673585e-06,
+      "loss": 0.1356,
+      "step": 2695
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.093479115689351e-06,
+      "loss": 0.175,
+      "step": 2696
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0828847985577147e-06,
+      "loss": 0.1615,
+      "step": 2697
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0723071387102753e-06,
+      "loss": 0.1754,
+      "step": 2698
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0617461465622234e-06,
+      "loss": 0.1672,
+      "step": 2699
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0512018325123495e-06,
+      "loss": 0.1589,
+      "step": 2700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.040674206943022e-06,
+      "loss": 0.1516,
+      "step": 2701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.030163280220162e-06,
+      "loss": 0.1573,
+      "step": 2702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.019669062693269e-06,
+      "loss": 0.164,
+      "step": 2703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.0091915646953728e-06,
+      "loss": 0.1498,
+      "step": 2704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9987307965430435e-06,
+      "loss": 0.1747,
+      "step": 2705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.988286768536388e-06,
+      "loss": 0.1516,
+      "step": 2706
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9778594909590184e-06,
+      "loss": 0.1605,
+      "step": 2707
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9674489740780665e-06,
+      "loss": 0.1588,
+      "step": 2708
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.95705522814415e-06,
+      "loss": 0.1572,
+      "step": 2709
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9466782633913717e-06,
+      "loss": 0.1622,
+      "step": 2710
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9363180900373223e-06,
+      "loss": 0.1615,
+      "step": 2711
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9259747182830467e-06,
+      "loss": 0.164,
+      "step": 2712
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.915648158313056e-06,
+      "loss": 0.1826,
+      "step": 2713
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9053384202953138e-06,
+      "loss": 0.1509,
+      "step": 2714
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.8950455143811893e-06,
+      "loss": 0.1497,
+      "step": 2715
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.8847694507055156e-06,
+      "loss": 0.1518,
+      "step": 2716
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.8745102393865144e-06,
+      "loss": 0.1568,
+      "step": 2717
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.8642678905258313e-06,
+      "loss": 0.1756,
+      "step": 2718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8540424142085065e-06,
+      "loss": 0.1601,
+      "step": 2719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8438338205029593e-06,
+      "loss": 0.16,
+      "step": 2720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8336421194609798e-06,
+      "loss": 0.1999,
+      "step": 2721
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.823467321117748e-06,
+      "loss": 0.141,
+      "step": 2722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8133094354917757e-06,
+      "loss": 0.1589,
+      "step": 2723
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.803168472584943e-06,
+      "loss": 0.1251,
+      "step": 2724
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.793044442382453e-06,
+      "loss": 0.1765,
+      "step": 2725
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.782937354852833e-06,
+      "loss": 0.1509,
+      "step": 2726
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7728472199479474e-06,
+      "loss": 0.1673,
+      "step": 2727
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.762774047602945e-06,
+      "loss": 0.1655,
+      "step": 2728
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7527178477362927e-06,
+      "loss": 0.1787,
+      "step": 2729
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.742678630249729e-06,
+      "loss": 0.1503,
+      "step": 2730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7326564050282865e-06,
+      "loss": 0.1524,
+      "step": 2731
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7226511819402545e-06,
+      "loss": 0.1381,
+      "step": 2732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.712662970837181e-06,
+      "loss": 0.1426,
+      "step": 2733
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.70269178155387e-06,
+      "loss": 0.2075,
+      "step": 2734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6927376239083703e-06,
+      "loss": 0.1621,
+      "step": 2735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6828005077019457e-06,
+      "loss": 0.1581,
+      "step": 2736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.67288044271909e-06,
+      "loss": 0.1592,
+      "step": 2737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6629774387274987e-06,
+      "loss": 0.1402,
+      "step": 2738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6530915054780803e-06,
+      "loss": 0.1777,
+      "step": 2739
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6432226527049355e-06,
+      "loss": 0.1738,
+      "step": 2740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.633370890125333e-06,
+      "loss": 0.1532,
+      "step": 2741
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6235362274397225e-06,
+      "loss": 0.1533,
+      "step": 2742
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.613718674331713e-06,
+      "loss": 0.1353,
+      "step": 2743
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.603918240468073e-06,
+      "loss": 0.1646,
+      "step": 2744
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5941349354987157e-06,
+      "loss": 0.1609,
+      "step": 2745
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.584368769056671e-06,
+      "loss": 0.1777,
+      "step": 2746
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5746197507581205e-06,
+      "loss": 0.161,
+      "step": 2747
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.564887890202339e-06,
+      "loss": 0.1551,
+      "step": 2748
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.555173196971712e-06,
+      "loss": 0.1363,
+      "step": 2749
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5454756806317306e-06,
+      "loss": 0.168,
+      "step": 2750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.535795350730961e-06,
+      "loss": 0.1584,
+      "step": 2751
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5261322168010583e-06,
+      "loss": 0.1393,
+      "step": 2752
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5164862883567366e-06,
+      "loss": 0.1488,
+      "step": 2753
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5068575748957667e-06,
+      "loss": 0.143,
+      "step": 2754
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4972460858989834e-06,
+      "loss": 0.1366,
+      "step": 2755
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.487651830830244e-06,
+      "loss": 0.14,
+      "step": 2756
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.478074819136449e-06,
+      "loss": 0.1721,
+      "step": 2757
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.468515060247525e-06,
+      "loss": 0.1521,
+      "step": 2758
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.458972563576385e-06,
+      "loss": 0.1479,
+      "step": 2759
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.449447338518969e-06,
+      "loss": 0.1699,
+      "step": 2760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.439939394454207e-06,
+      "loss": 0.132,
+      "step": 2761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4304487407440048e-06,
+      "loss": 0.1474,
+      "step": 2762
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.420975386733253e-06,
+      "loss": 0.1515,
+      "step": 2763
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.411519341749804e-06,
+      "loss": 0.2013,
+      "step": 2764
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4020806151044607e-06,
+      "loss": 0.1812,
+      "step": 2765
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.392659216090987e-06,
+      "loss": 0.166,
+      "step": 2766
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.3832551539860725e-06,
+      "loss": 0.1521,
+      "step": 2767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3738684380493447e-06,
+      "loss": 0.1623,
+      "step": 2768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3644990775233502e-06,
+      "loss": 0.1626,
+      "step": 2769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3551470816335397e-06,
+      "loss": 0.1531,
+      "step": 2770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3458124595882746e-06,
+      "loss": 0.1747,
+      "step": 2771
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.336495220578805e-06,
+      "loss": 0.1501,
+      "step": 2772
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.327195373779263e-06,
+      "loss": 0.1501,
+      "step": 2773
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3179129283466704e-06,
+      "loss": 0.1379,
+      "step": 2774
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.308647893420888e-06,
+      "loss": 0.1746,
+      "step": 2775
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2994002781246573e-06,
+      "loss": 0.1374,
+      "step": 2776
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2901700915635528e-06,
+      "loss": 0.1526,
+      "step": 2777
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.280957342825998e-06,
+      "loss": 0.1393,
+      "step": 2778
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.271762040983245e-06,
+      "loss": 0.1688,
+      "step": 2779
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2625841950893612e-06,
+      "loss": 0.1426,
+      "step": 2780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2534238141812283e-06,
+      "loss": 0.1546,
+      "step": 2781
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2442809072785355e-06,
+      "loss": 0.1315,
+      "step": 2782
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2351554833837574e-06,
+      "loss": 0.1618,
+      "step": 2783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.226047551482169e-06,
+      "loss": 0.149,
+      "step": 2784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2169571205418093e-06,
+      "loss": 0.168,
+      "step": 2785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2078841995134816e-06,
+      "loss": 0.146,
+      "step": 2786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1988287973307653e-06,
+      "loss": 0.1731,
+      "step": 2787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1897909229099735e-06,
+      "loss": 0.1592,
+      "step": 2788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1807705851501716e-06,
+      "loss": 0.1483,
+      "step": 2789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1717677929331484e-06,
+      "loss": 0.1633,
+      "step": 2790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1627825551234307e-06,
+      "loss": 0.1571,
+      "step": 2791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1538148805682437e-06,
+      "loss": 0.1748,
+      "step": 2792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1448647780975264e-06,
+      "loss": 0.1474,
+      "step": 2793
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.135932256523918e-06,
+      "loss": 0.13,
+      "step": 2794
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1270173246427505e-06,
+      "loss": 0.1584,
+      "step": 2795
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.118119991232026e-06,
+      "loss": 0.1728,
+      "step": 2796
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1092402650524215e-06,
+      "loss": 0.1358,
+      "step": 2797
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1003781548472777e-06,
+      "loss": 0.1516,
+      "step": 2798
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0915336693425938e-06,
+      "loss": 0.2026,
+      "step": 2799
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0827068172470177e-06,
+      "loss": 0.1364,
+      "step": 2800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0738976072518225e-06,
+      "loss": 0.1436,
+      "step": 2801
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0651060480309182e-06,
+      "loss": 0.176,
+      "step": 2802
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0563321482408316e-06,
+      "loss": 0.1554,
+      "step": 2803
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0475759165207076e-06,
+      "loss": 0.1527,
+      "step": 2804
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.038837361492292e-06,
+      "loss": 0.1645,
+      "step": 2805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.030116491759917e-06,
+      "loss": 0.1561,
+      "step": 2806
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.021413315910514e-06,
+      "loss": 0.1805,
+      "step": 2807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0127278425135864e-06,
+      "loss": 0.14,
+      "step": 2808
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.004060080121202e-06,
+      "loss": 0.1703,
+      "step": 2809
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.995410037268e-06,
+      "loss": 0.1465,
+      "step": 2810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9867777224711583e-06,
+      "loss": 0.1404,
+      "step": 2811
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9781631442304185e-06,
+      "loss": 0.1548,
+      "step": 2812
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9695663110280415e-06,
+      "loss": 0.1521,
+      "step": 2813
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.960987231328817e-06,
+      "loss": 0.1541,
+      "step": 2814
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9524259135800672e-06,
+      "loss": 0.1606,
+      "step": 2815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.943882366211607e-06,
+      "loss": 0.1593,
+      "step": 2816
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.935356597635767e-06,
+      "loss": 0.1448,
+      "step": 2817
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9268486162473765e-06,
+      "loss": 0.1097,
+      "step": 2818
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9183584304237256e-06,
+      "loss": 0.1519,
+      "step": 2819
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9098860485246072e-06,
+      "loss": 0.1696,
+      "step": 2820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9014314788922795e-06,
+      "loss": 0.1639,
+      "step": 2821
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8929947298514495e-06,
+      "loss": 0.1471,
+      "step": 2822
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8845758097092926e-06,
+      "loss": 0.1536,
+      "step": 2823
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.876174726755422e-06,
+      "loss": 0.1624,
+      "step": 2824
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8677914892618787e-06,
+      "loss": 0.1781,
+      "step": 2825
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8594261054831553e-06,
+      "loss": 0.1574,
+      "step": 2826
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8510785836561385e-06,
+      "loss": 0.1342,
+      "step": 2827
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.842748932000149e-06,
+      "loss": 0.1834,
+      "step": 2828
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8344371587168997e-06,
+      "loss": 0.1689,
+      "step": 2829
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8261432719904993e-06,
+      "loss": 0.1379,
+      "step": 2830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8178672799874553e-06,
+      "loss": 0.1321,
+      "step": 2831
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8096091908566426e-06,
+      "loss": 0.1668,
+      "step": 2832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8013690127293126e-06,
+      "loss": 0.1511,
+      "step": 2833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7931467537190904e-06,
+      "loss": 0.134,
+      "step": 2834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7849424219219425e-06,
+      "loss": 0.1544,
+      "step": 2835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7767560254161931e-06,
+      "loss": 0.1415,
+      "step": 2836
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7685875722624967e-06,
+      "loss": 0.1732,
+      "step": 2837
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7604370705038488e-06,
+      "loss": 0.1442,
+      "step": 2838
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.752304528165576e-06,
+      "loss": 0.1357,
+      "step": 2839
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7441899532553042e-06,
+      "loss": 0.146,
+      "step": 2840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7360933537629799e-06,
+      "loss": 0.1563,
+      "step": 2841
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.728014737660837e-06,
+      "loss": 0.1831,
+      "step": 2842
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7199541129034214e-06,
+      "loss": 0.1482,
+      "step": 2843
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7119114874275555e-06,
+      "loss": 0.1404,
+      "step": 2844
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7038868691523336e-06,
+      "loss": 0.1846,
+      "step": 2845
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6958802659791195e-06,
+      "loss": 0.1636,
+      "step": 2846
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6878916857915495e-06,
+      "loss": 0.1552,
+      "step": 2847
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6799211364555024e-06,
+      "loss": 0.1501,
+      "step": 2848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6719686258191114e-06,
+      "loss": 0.1538,
+      "step": 2849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6640341617127376e-06,
+      "loss": 0.1446,
+      "step": 2850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6561177519489867e-06,
+      "loss": 0.1614,
+      "step": 2851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6482194043226751e-06,
+      "loss": 0.1528,
+      "step": 2852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.640339126610837e-06,
+      "loss": 0.1769,
+      "step": 2853
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6324769265727236e-06,
+      "loss": 0.1278,
+      "step": 2854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6246328119497734e-06,
+      "loss": 0.1463,
+      "step": 2855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6168067904656259e-06,
+      "loss": 0.1692,
+      "step": 2856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6089988698260994e-06,
+      "loss": 0.1651,
+      "step": 2857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.601209057719193e-06,
+      "loss": 0.16,
+      "step": 2858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5934373618150733e-06,
+      "loss": 0.1473,
+      "step": 2859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.585683789766077e-06,
+      "loss": 0.182,
+      "step": 2860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5779483492066793e-06,
+      "loss": 0.1772,
+      "step": 2861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5702310477535231e-06,
+      "loss": 0.1485,
+      "step": 2862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5625318930053656e-06,
+      "loss": 0.1583,
+      "step": 2863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5548508925431161e-06,
+      "loss": 0.1436,
+      "step": 2864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5471880539298067e-06,
+      "loss": 0.1644,
+      "step": 2865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5395433847105735e-06,
+      "loss": 0.1724,
+      "step": 2866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5319168924126793e-06,
+      "loss": 0.1567,
+      "step": 2867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5243085845454752e-06,
+      "loss": 0.1381,
+      "step": 2868
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5167184686004111e-06,
+      "loss": 0.1587,
+      "step": 2869
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5091465520510307e-06,
+      "loss": 0.1742,
+      "step": 2870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5015928423529502e-06,
+      "loss": 0.136,
+      "step": 2871
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4940573469438625e-06,
+      "loss": 0.2021,
+      "step": 2872
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4865400732435276e-06,
+      "loss": 0.1692,
+      "step": 2873
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4790410286537516e-06,
+      "loss": 0.1492,
+      "step": 2874
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4715602205584123e-06,
+      "loss": 0.1468,
+      "step": 2875
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4640976563234089e-06,
+      "loss": 0.1656,
+      "step": 2876
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4566533432966945e-06,
+      "loss": 0.1498,
+      "step": 2877
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.449227288808248e-06,
+      "loss": 0.1629,
+      "step": 2878
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4418195001700585e-06,
+      "loss": 0.1629,
+      "step": 2879
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4344299846761444e-06,
+      "loss": 0.1488,
+      "step": 2880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4270587496025168e-06,
+      "loss": 0.1505,
+      "step": 2881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4197058022072008e-06,
+      "loss": 0.1563,
+      "step": 2882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4123711497302117e-06,
+      "loss": 0.1646,
+      "step": 2883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4050547993935482e-06,
+      "loss": 0.184,
+      "step": 2884
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3977567584011809e-06,
+      "loss": 0.1363,
+      "step": 2885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3904770339390706e-06,
+      "loss": 0.1484,
+      "step": 2886
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.383215633175121e-06,
+      "loss": 0.1642,
+      "step": 2887
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3759725632592157e-06,
+      "loss": 0.1863,
+      "step": 2888
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3687478313231718e-06,
+      "loss": 0.1097,
+      "step": 2889
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3615414444807539e-06,
+      "loss": 0.1446,
+      "step": 2890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.354353409827669e-06,
+      "loss": 0.163,
+      "step": 2891
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3471837344415484e-06,
+      "loss": 0.1809,
+      "step": 2892
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3400324253819518e-06,
+      "loss": 0.1561,
+      "step": 2893
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3328994896903468e-06,
+      "loss": 0.1403,
+      "step": 2894
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3257849343901174e-06,
+      "loss": 0.1384,
+      "step": 2895
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3186887664865445e-06,
+      "loss": 0.1794,
+      "step": 2896
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3116109929668008e-06,
+      "loss": 0.145,
+      "step": 2897
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3045516207999564e-06,
+      "loss": 0.1611,
+      "step": 2898
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.29751065693696e-06,
+      "loss": 0.1752,
+      "step": 2899
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2904881083106324e-06,
+      "loss": 0.1501,
+      "step": 2900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2834839818356604e-06,
+      "loss": 0.1139,
+      "step": 2901
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2764982844085893e-06,
+      "loss": 0.1326,
+      "step": 2902
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2695310229078283e-06,
+      "loss": 0.178,
+      "step": 2903
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.262582204193632e-06,
+      "loss": 0.1774,
+      "step": 2904
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2556518351080827e-06,
+      "loss": 0.1734,
+      "step": 2905
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2487399224751152e-06,
+      "loss": 0.1515,
+      "step": 2906
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2418464731004742e-06,
+      "loss": 0.1496,
+      "step": 2907
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2349714937717328e-06,
+      "loss": 0.1576,
+      "step": 2908
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2281149912582803e-06,
+      "loss": 0.183,
+      "step": 2909
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2212769723113072e-06,
+      "loss": 0.1784,
+      "step": 2910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.21445744366381e-06,
+      "loss": 0.1349,
+      "step": 2911
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2076564120305734e-06,
+      "loss": 0.1526,
+      "step": 2912
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.200873884108169e-06,
+      "loss": 0.1628,
+      "step": 2913
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1941098665749595e-06,
+      "loss": 0.1651,
+      "step": 2914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1873643660910617e-06,
+      "loss": 0.164,
+      "step": 2915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1806373892983847e-06,
+      "loss": 0.1544,
+      "step": 2916
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.173928942820577e-06,
+      "loss": 0.1283,
+      "step": 2917
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1672390332630478e-06,
+      "loss": 0.1399,
+      "step": 2918
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1605676672129617e-06,
+      "loss": 0.1627,
+      "step": 2919
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1539148512392174e-06,
+      "loss": 0.1404,
+      "step": 2920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.147280591892448e-06,
+      "loss": 0.1887,
+      "step": 2921
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1406648957050192e-06,
+      "loss": 0.1317,
+      "step": 2922
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1340677691910162e-06,
+      "loss": 0.1792,
+      "step": 2923
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1274892188462362e-06,
+      "loss": 0.1473,
+      "step": 2924
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.120929251148193e-06,
+      "loss": 0.1512,
+      "step": 2925
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1143878725560974e-06,
+      "loss": 0.1758,
+      "step": 2926
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.10786508951086e-06,
+      "loss": 0.1343,
+      "step": 2927
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1013609084350806e-06,
+      "loss": 0.1569,
+      "step": 2928
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.0948753357330366e-06,
+      "loss": 0.114,
+      "step": 2929
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.0884083777906972e-06,
+      "loss": 0.1449,
+      "step": 2930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0819600409756848e-06,
+      "loss": 0.1425,
+      "step": 2931
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0755303316373022e-06,
+      "loss": 0.1677,
+      "step": 2932
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0691192561065079e-06,
+      "loss": 0.1441,
+      "step": 2933
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0627268206958963e-06,
+      "loss": 0.1457,
+      "step": 2934
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0563530316997328e-06,
+      "loss": 0.1497,
+      "step": 2935
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0499978953939038e-06,
+      "loss": 0.1342,
+      "step": 2936
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0436614180359416e-06,
+      "loss": 0.1759,
+      "step": 2937
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0373436058649999e-06,
+      "loss": 0.1631,
+      "step": 2938
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.031044465101856e-06,
+      "loss": 0.1479,
+      "step": 2939
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0247640019489015e-06,
+      "loss": 0.146,
+      "step": 2940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.018502222590132e-06,
+      "loss": 0.1355,
+      "step": 2941
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0122591331911558e-06,
+      "loss": 0.1353,
+      "step": 2942
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0060347398991798e-06,
+      "loss": 0.1669,
+      "step": 2943
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.998290488429908e-07,
+      "loss": 0.1619,
+      "step": 2944
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.936420661329626e-07,
+      "loss": 0.1571,
+      "step": 2945
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.874737978610604e-07,
+      "loss": 0.1466,
+      "step": 2946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.813242501008057e-07,
+      "loss": 0.1693,
+      "step": 2947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.751934289073018e-07,
+      "loss": 0.1348,
+      "step": 2948
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.690813403171995e-07,
+      "loss": 0.1363,
+      "step": 2949
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.629879903487184e-07,
+      "loss": 0.1354,
+      "step": 2950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.569133850016187e-07,
+      "loss": 0.1527,
+      "step": 2951
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.508575302571988e-07,
+      "loss": 0.127,
+      "step": 2952
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.448204320783083e-07,
+      "loss": 0.1848,
+      "step": 2953
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.388020964093169e-07,
+      "loss": 0.1642,
+      "step": 2954
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.328025291761311e-07,
+      "loss": 0.1435,
+      "step": 2955
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.268217362861631e-07,
+      "loss": 0.1524,
+      "step": 2956
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.208597236283468e-07,
+      "loss": 0.167,
+      "step": 2957
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.149164970731283e-07,
+      "loss": 0.174,
+      "step": 2958
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.08992062472449e-07,
+      "loss": 0.1625,
+      "step": 2959
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.030864256597538e-07,
+      "loss": 0.1317,
+      "step": 2960
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.971995924499687e-07,
+      "loss": 0.151,
+      "step": 2961
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.913315686395107e-07,
+      "loss": 0.1639,
+      "step": 2962
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.854823600062801e-07,
+      "loss": 0.1564,
+      "step": 2963
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.796519723096475e-07,
+      "loss": 0.1401,
+      "step": 2964
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.738404112904475e-07,
+      "loss": 0.1423,
+      "step": 2965
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.680476826709849e-07,
+      "loss": 0.1495,
+      "step": 2966
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.622737921550151e-07,
+      "loss": 0.1603,
+      "step": 2967
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.565187454277457e-07,
+      "loss": 0.1486,
+      "step": 2968
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.507825481558352e-07,
+      "loss": 0.1442,
+      "step": 2969
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.450652059873743e-07,
+      "loss": 0.1613,
+      "step": 2970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.393667245518977e-07,
+      "loss": 0.163,
+      "step": 2971
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.336871094603593e-07,
+      "loss": 0.1574,
+      "step": 2972
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.280263663051369e-07,
+      "loss": 0.1679,
+      "step": 2973
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.223845006600384e-07,
+      "loss": 0.1615,
+      "step": 2974
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.167615180802691e-07,
+      "loss": 0.1507,
+      "step": 2975
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.111574241024534e-07,
+      "loss": 0.1156,
+      "step": 2976
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.055722242446063e-07,
+      "loss": 0.1733,
+      "step": 2977
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.000059240061465e-07,
+      "loss": 0.1399,
+      "step": 2978
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.94458528867883e-07,
+      "loss": 0.1593,
+      "step": 2979
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.889300442920022e-07,
+      "loss": 0.1831,
+      "step": 2980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.834204757220831e-07,
+      "loss": 0.1389,
+      "step": 2981
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.779298285830706e-07,
+      "loss": 0.1433,
+      "step": 2982
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.724581082812798e-07,
+      "loss": 0.1377,
+      "step": 2983
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.670053202043882e-07,
+      "loss": 0.171,
+      "step": 2984
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.615714697214383e-07,
+      "loss": 0.1497,
+      "step": 2985
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.561565621828193e-07,
+      "loss": 0.1254,
+      "step": 2986
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.50760602920273e-07,
+      "loss": 0.1541,
+      "step": 2987
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.453835972468803e-07,
+      "loss": 0.1613,
+      "step": 2988
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.400255504570597e-07,
+      "loss": 0.1493,
+      "step": 2989
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.346864678265664e-07,
+      "loss": 0.1607,
+      "step": 2990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.293663546124752e-07,
+      "loss": 0.1764,
+      "step": 2991
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.240652160531936e-07,
+      "loss": 0.1598,
+      "step": 2992
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.187830573684374e-07,
+      "loss": 0.1475,
+      "step": 2993
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.135198837592305e-07,
+      "loss": 0.1561,
+      "step": 2994
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.082757004079188e-07,
+      "loss": 0.1652,
+      "step": 2995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.030505124781295e-07,
+      "loss": 0.144,
+      "step": 2996
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.978443251148048e-07,
+      "loss": 0.1419,
+      "step": 2997
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.926571434441686e-07,
+      "loss": 0.1244,
+      "step": 2998
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.87488972573731e-07,
+      "loss": 0.1477,
+      "step": 2999
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823398175922857e-07,
+      "loss": 0.1649,
+      "step": 3000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.772096835698949e-07,
+      "loss": 0.1836,
+      "step": 3001
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.720985755579068e-07,
+      "loss": 0.161,
+      "step": 3002
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.670064985889269e-07,
+      "loss": 0.1772,
+      "step": 3003
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.6193345767682e-07,
+      "loss": 0.1705,
+      "step": 3004
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.568794578167104e-07,
+      "loss": 0.1473,
+      "step": 3005
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.518445039849708e-07,
+      "loss": 0.147,
+      "step": 3006
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.468286011392245e-07,
+      "loss": 0.1675,
+      "step": 3007
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.418317542183406e-07,
+      "loss": 0.1637,
+      "step": 3008
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.368539681424101e-07,
+      "loss": 0.1653,
+      "step": 3009
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.318952478127704e-07,
+      "loss": 0.1776,
+      "step": 3010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.269555981119802e-07,
+      "loss": 0.1685,
+      "step": 3011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.220350239038175e-07,
+      "loss": 0.1609,
+      "step": 3012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.17133530033287e-07,
+      "loss": 0.1412,
+      "step": 3013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.122511213265925e-07,
+      "loss": 0.165,
+      "step": 3014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.073878025911594e-07,
+      "loss": 0.1416,
+      "step": 3015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.025435786156109e-07,
+      "loss": 0.1673,
+      "step": 3016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.977184541697623e-07,
+      "loss": 0.1422,
+      "step": 3017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.929124340046355e-07,
+      "loss": 0.1499,
+      "step": 3018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.881255228524275e-07,
+      "loss": 0.1278,
+      "step": 3019
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.833577254265343e-07,
+      "loss": 0.1548,
+      "step": 3020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.786090464215232e-07,
+      "loss": 0.133,
+      "step": 3021
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.738794905131318e-07,
+      "loss": 0.1604,
+      "step": 3022
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.69169062358279e-07,
+      "loss": 0.1195,
+      "step": 3023
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.644777665950462e-07,
+      "loss": 0.1627,
+      "step": 3024
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.598056078426717e-07,
+      "loss": 0.1647,
+      "step": 3025
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.551525907015598e-07,
+      "loss": 0.1505,
+      "step": 3026
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.505187197532569e-07,
+      "loss": 0.1648,
+      "step": 3027
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.45903999560462e-07,
+      "loss": 0.154,
+      "step": 3028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.413084346670228e-07,
+      "loss": 0.1705,
+      "step": 3029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.367320295979128e-07,
+      "loss": 0.1177,
+      "step": 3030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.321747888592543e-07,
+      "loss": 0.1583,
+      "step": 3031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.276367169382912e-07,
+      "loss": 0.1645,
+      "step": 3032
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.231178183033914e-07,
+      "loss": 0.1293,
+      "step": 3033
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.186180974040534e-07,
+      "loss": 0.1295,
+      "step": 3034
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.141375586708797e-07,
+      "loss": 0.1686,
+      "step": 3035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.096762065155992e-07,
+      "loss": 0.1603,
+      "step": 3036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.0523404533104e-07,
+      "loss": 0.166,
+      "step": 3037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.008110794911325e-07,
+      "loss": 0.1629,
+      "step": 3038
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.964073133509128e-07,
+      "loss": 0.1482,
+      "step": 3039
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.920227512465059e-07,
+      "loss": 0.1395,
+      "step": 3040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.87657397495136e-07,
+      "loss": 0.1424,
+      "step": 3041
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.83311256395107e-07,
+      "loss": 0.137,
+      "step": 3042
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.789843322258093e-07,
+      "loss": 0.1474,
+      "step": 3043
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.7467662924770564e-07,
+      "loss": 0.1448,
+      "step": 3044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.703881517023368e-07,
+      "loss": 0.1208,
+      "step": 3045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6611890381231596e-07,
+      "loss": 0.1289,
+      "step": 3046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.618688897813206e-07,
+      "loss": 0.1649,
+      "step": 3047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.576381137940877e-07,
+      "loss": 0.161,
+      "step": 3048
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.534265800164073e-07,
+      "loss": 0.1525,
+      "step": 3049
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.492342925951354e-07,
+      "loss": 0.1501,
+      "step": 3050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.450612556581635e-07,
+      "loss": 0.1431,
+      "step": 3051
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.4090747331444027e-07,
+      "loss": 0.1365,
+      "step": 3052
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.36772949653943e-07,
+      "loss": 0.1508,
+      "step": 3053
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3265768874770186e-07,
+      "loss": 0.152,
+      "step": 3054
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2856169464776445e-07,
+      "loss": 0.1714,
+      "step": 3055
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.244849713872157e-07,
+      "loss": 0.1508,
+      "step": 3056
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2042752298016463e-07,
+      "loss": 0.1475,
+      "step": 3057
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.163893534217422e-07,
+      "loss": 0.1445,
+      "step": 3058
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1237046668809455e-07,
+      "loss": 0.1455,
+      "step": 3059
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.083708667363828e-07,
+      "loss": 0.1511,
+      "step": 3060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.043905575047746e-07,
+      "loss": 0.1505,
+      "step": 3061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0042954291244605e-07,
+      "loss": 0.1339,
+      "step": 3062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.964878268595751e-07,
+      "loss": 0.1586,
+      "step": 3063
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.925654132273393e-07,
+      "loss": 0.1638,
+      "step": 3064
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.886623058779027e-07,
+      "loss": 0.1587,
+      "step": 3065
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8477850865442647e-07,
+      "loss": 0.1449,
+      "step": 3066
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.809140253810584e-07,
+      "loss": 0.1296,
+      "step": 3067
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.770688598629235e-07,
+      "loss": 0.1515,
+      "step": 3068
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.732430158861311e-07,
+      "loss": 0.16,
+      "step": 3069
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6943649721776556e-07,
+      "loss": 0.1434,
+      "step": 3070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.656493076058798e-07,
+      "loss": 0.1301,
+      "step": 3071
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.61881450779491e-07,
+      "loss": 0.1325,
+      "step": 3072
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5813293044859366e-07,
+      "loss": 0.1445,
+      "step": 3073
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5440375030412643e-07,
+      "loss": 0.1486,
+      "step": 3074
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.506939140179988e-07,
+      "loss": 0.1868,
+      "step": 3075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.4700342524306654e-07,
+      "loss": 0.1563,
+      "step": 3076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.4333228761312954e-07,
+      "loss": 0.1386,
+      "step": 3077
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.3968050474294965e-07,
+      "loss": 0.1416,
+      "step": 3078
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.360480802282129e-07,
+      "loss": 0.161,
+      "step": 3079
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.324350176455604e-07,
+      "loss": 0.1453,
+      "step": 3080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.288413205525598e-07,
+      "loss": 0.1643,
+      "step": 3081
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.252669924877072e-07,
+      "loss": 0.1328,
+      "step": 3082
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.217120369704385e-07,
+      "loss": 0.1447,
+      "step": 3083
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1817645750110704e-07,
+      "loss": 0.1293,
+      "step": 3084
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1466025756098586e-07,
+      "loss": 0.1515,
+      "step": 3085
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1116344061227435e-07,
+      "loss": 0.127,
+      "step": 3086
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.0768601009808054e-07,
+      "loss": 0.1871,
+      "step": 3087
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.042279694424188e-07,
+      "loss": 0.1624,
+      "step": 3088
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.0078932205022557e-07,
+      "loss": 0.1549,
+      "step": 3089
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9737007130732797e-07,
+      "loss": 0.1276,
+      "step": 3090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.939702205804662e-07,
+      "loss": 0.1852,
+      "step": 3091
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.905897732172669e-07,
+      "loss": 0.1591,
+      "step": 3092
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.872287325462586e-07,
+      "loss": 0.1464,
+      "step": 3093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8388710187686074e-07,
+      "loss": 0.1799,
+      "step": 3094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8056488449937913e-07,
+      "loss": 0.1531,
+      "step": 3095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.772620836850015e-07,
+      "loss": 0.1407,
+      "step": 3096
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.739787026858087e-07,
+      "loss": 0.1668,
+      "step": 3097
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.707147447347458e-07,
+      "loss": 0.1659,
+      "step": 3098
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.674702130456419e-07,
+      "loss": 0.1539,
+      "step": 3099
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.642451108131927e-07,
+      "loss": 0.1531,
+      "step": 3100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6103944121296907e-07,
+      "loss": 0.1316,
+      "step": 3101
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5785320740140397e-07,
+      "loss": 0.1614,
+      "step": 3102
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.546864125157944e-07,
+      "loss": 0.1345,
+      "step": 3103
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5153905967429505e-07,
+      "loss": 0.1266,
+      "step": 3104
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.484111519759158e-07,
+      "loss": 0.1653,
+      "step": 3105
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.453026925005242e-07,
+      "loss": 0.1619,
+      "step": 3106
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.422136843088407e-07,
+      "loss": 0.162,
+      "step": 3107
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3914413044242803e-07,
+      "loss": 0.1635,
+      "step": 3108
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3609403392368835e-07,
+      "loss": 0.1386,
+      "step": 3109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3306339775587938e-07,
+      "loss": 0.1588,
+      "step": 3110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.30052224923083e-07,
+      "loss": 0.1337,
+      "step": 3111
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.270605183902297e-07,
+      "loss": 0.1608,
+      "step": 3112
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.240882811030698e-07,
+      "loss": 0.1581,
+      "step": 3113
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.211355159881956e-07,
+      "loss": 0.1784,
+      "step": 3114
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.1820222595301922e-07,
+      "loss": 0.1361,
+      "step": 3115
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.1528841388577248e-07,
+      "loss": 0.1519,
+      "step": 3116
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.1239408265552263e-07,
+      "loss": 0.1642,
+      "step": 3117
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.095192351121389e-07,
+      "loss": 0.1473,
+      "step": 3118
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0666387408631917e-07,
+      "loss": 0.1503,
+      "step": 3119
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.038280023895678e-07,
+      "loss": 0.1471,
+      "step": 3120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0101162281419563e-07,
+      "loss": 0.1669,
+      "step": 3121
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9821473813333104e-07,
+      "loss": 0.1645,
+      "step": 3122
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.954373511009e-07,
+      "loss": 0.153,
+      "step": 3123
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9267946445163055e-07,
+      "loss": 0.1412,
+      "step": 3124
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8994108090105268e-07,
+      "loss": 0.1712,
+      "step": 3125
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8722220314548733e-07,
+      "loss": 0.182,
+      "step": 3126
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8452283386205306e-07,
+      "loss": 0.1412,
+      "step": 3127
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8184297570866372e-07,
+      "loss": 0.1634,
+      "step": 3128
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7918263132401303e-07,
+      "loss": 0.1541,
+      "step": 3129
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7654180332758564e-07,
+      "loss": 0.1463,
+      "step": 3130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7392049431965264e-07,
+      "loss": 0.1417,
+      "step": 3131
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7131870688125829e-07,
+      "loss": 0.1586,
+      "step": 3132
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6873644357422892e-07,
+      "loss": 0.1617,
+      "step": 3133
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.661737069411684e-07,
+      "loss": 0.1755,
+      "step": 3134
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6363049950545384e-07,
+      "loss": 0.1624,
+      "step": 3135
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6110682377122655e-07,
+      "loss": 0.1756,
+      "step": 3136
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5860268222340324e-07,
+      "loss": 0.1694,
+      "step": 3137
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5611807732766716e-07,
+      "loss": 0.1189,
+      "step": 3138
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5365301153045686e-07,
+      "loss": 0.1593,
+      "step": 3139
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5120748725898194e-07,
+      "loss": 0.1458,
+      "step": 3140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4878150692120285e-07,
+      "loss": 0.1922,
+      "step": 3141
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4637507290583774e-07,
+      "loss": 0.1402,
+      "step": 3142
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4398818758236455e-07,
+      "loss": 0.1289,
+      "step": 3143
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4162085330100329e-07,
+      "loss": 0.1489,
+      "step": 3144
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3927307239273157e-07,
+      "loss": 0.1341,
+      "step": 3145
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3694484716927137e-07,
+      "loss": 0.1533,
+      "step": 3146
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3463617992308664e-07,
+      "loss": 0.1581,
+      "step": 3147
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3234707292738346e-07,
+      "loss": 0.1724,
+      "step": 3148
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3007752843611444e-07,
+      "loss": 0.1383,
+      "step": 3149
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2782754868396306e-07,
+      "loss": 0.1404,
+      "step": 3150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2559713588635502e-07,
+      "loss": 0.173,
+      "step": 3151
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2338629223944244e-07,
+      "loss": 0.1676,
+      "step": 3152
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2119501992011284e-07,
+      "loss": 0.1349,
+      "step": 3153
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.190233210859848e-07,
+      "loss": 0.1384,
+      "step": 3154
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1687119787540113e-07,
+      "loss": 0.1285,
+      "step": 3155
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1473865240742899e-07,
+      "loss": 0.1259,
+      "step": 3156
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1262568678185981e-07,
+      "loss": 0.1489,
+      "step": 3157
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1053230307920937e-07,
+      "loss": 0.1452,
+      "step": 3158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0845850336070663e-07,
+      "loss": 0.163,
+      "step": 3159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0640428966830041e-07,
+      "loss": 0.1491,
+      "step": 3160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0436966402465054e-07,
+      "loss": 0.1439,
+      "step": 3161
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0235462843313671e-07,
+      "loss": 0.1607,
+      "step": 3162
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0035918487784513e-07,
+      "loss": 0.1377,
+      "step": 3163
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.838333532357081e-08,
+      "loss": 0.1371,
+      "step": 3164
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.642708171581305e-08,
+      "loss": 0.1556,
+      "step": 3165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.449042598078218e-08,
+      "loss": 0.1552,
+      "step": 3166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.257337002538835e-08,
+      "loss": 0.1398,
+      "step": 3167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.067591573724388e-08,
+      "loss": 0.1713,
+      "step": 3168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.879806498465871e-08,
+      "loss": 0.1304,
+      "step": 3169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.693981961664044e-08,
+      "loss": 0.1651,
+      "step": 3170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.510118146289658e-08,
+      "loss": 0.1511,
+      "step": 3171
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.328215233382341e-08,
+      "loss": 0.1369,
+      "step": 3172
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.148273402051265e-08,
+      "loss": 0.147,
+      "step": 3173
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.97029282947448e-08,
+      "loss": 0.1564,
+      "step": 3174
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.794273690899135e-08,
+      "loss": 0.1664,
+      "step": 3175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.62021615964037e-08,
+      "loss": 0.1667,
+      "step": 3176
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.44812040708287e-08,
+      "loss": 0.1718,
+      "step": 3177
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.277986602678866e-08,
+      "loss": 0.1577,
+      "step": 3178
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.109814913949242e-08,
+      "loss": 0.1523,
+      "step": 3179
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.943605506482653e-08,
+      "loss": 0.1733,
+      "step": 3180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.779358543935522e-08,
+      "loss": 0.1659,
+      "step": 3181
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.617074188032258e-08,
+      "loss": 0.1665,
+      "step": 3182
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.456752598564819e-08,
+      "loss": 0.1519,
+      "step": 3183
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.298393933392266e-08,
+      "loss": 0.1547,
+      "step": 3184
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.141998348441202e-08,
+      "loss": 0.1679,
+      "step": 3185
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.987565997704891e-08,
+      "loss": 0.145,
+      "step": 3186
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8350970332436974e-08,
+      "loss": 0.1347,
+      "step": 3187
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6845916051850855e-08,
+      "loss": 0.1335,
+      "step": 3188
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5360498617227364e-08,
+      "loss": 0.1464,
+      "step": 3189
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.389471949116987e-08,
+      "loss": 0.1447,
+      "step": 3190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.244858011694165e-08,
+      "loss": 0.1695,
+      "step": 3191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.10220819184748e-08,
+      "loss": 0.1479,
+      "step": 3192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.9615226300354644e-08,
+      "loss": 0.1445,
+      "step": 3193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.822801464782867e-08,
+      "loss": 0.1667,
+      "step": 3194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6860448326806475e-08,
+      "loss": 0.1566,
+      "step": 3195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.551252868384648e-08,
+      "loss": 0.1589,
+      "step": 3196
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.4184257046164804e-08,
+      "loss": 0.1462,
+      "step": 3197
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2875634721635254e-08,
+      "loss": 0.1305,
+      "step": 3198
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.1586662998776006e-08,
+      "loss": 0.1497,
+      "step": 3199
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.031734314676739e-08,
+      "loss": 0.1536,
+      "step": 3200
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3264,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 200,
+  "total_flos": 6.07149750651177e+21,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}