diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18253 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 3038,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-07,
+      "loss": 3.3606,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 3.1084,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.521739130434783e-07,
+      "loss": 2.9241,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 2.7249,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0869565217391306e-06,
+      "loss": 3.0549,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 3.4679,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.521739130434783e-06,
+      "loss": 2.6214,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 3.1068,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.956521739130435e-06,
+      "loss": 2.6991,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 2.1599,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.391304347826087e-06,
+      "loss": 2.3008,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 2.0458,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8260869565217393e-06,
+      "loss": 2.1521,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 1.8656,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2608695652173914e-06,
+      "loss": 1.9967,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.6188,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6956521739130436e-06,
+      "loss": 1.6884,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 1.652,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.130434782608696e-06,
+      "loss": 1.6944,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 1.7657,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.565217391304348e-06,
+      "loss": 1.4748,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 1.5941,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.5832,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 1.5005,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4347826086956525e-06,
+      "loss": 1.6937,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 1.747,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8695652173913055e-06,
+      "loss": 1.7165,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 1.572,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.304347826086958e-06,
+      "loss": 1.3604,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 1.3899,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.739130434782609e-06,
+      "loss": 1.3702,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 1.5494,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.173913043478261e-06,
+      "loss": 1.474,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 1.4282,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.608695652173914e-06,
+      "loss": 1.4724,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 1.45,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.043478260869566e-06,
+      "loss": 1.3026,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 1.5396,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.478260869565218e-06,
+      "loss": 1.4549,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 1.5142,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.91304347826087e-06,
+      "loss": 1.5001,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 1.4477,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.347826086956523e-06,
+      "loss": 1.2861,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 1.3809,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.782608695652175e-06,
+      "loss": 1.6367,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4074,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0217391304347829e-05,
+      "loss": 1.4236,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 1.6102,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0652173913043479e-05,
+      "loss": 1.4612,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 1.3934,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1086956521739131e-05,
+      "loss": 1.4294,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 1.5092,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1521739130434783e-05,
+      "loss": 1.2208,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1739130434782611e-05,
+      "loss": 1.2975,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1956521739130435e-05,
+      "loss": 1.4067,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 1.3488,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2391304347826088e-05,
+      "loss": 1.3825,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2608695652173915e-05,
+      "loss": 1.3548,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.282608695652174e-05,
+      "loss": 1.4098,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 1.5195,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3260869565217392e-05,
+      "loss": 1.4409,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 1.5297,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3695652173913046e-05,
+      "loss": 1.5921,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 1.2743,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4130434782608698e-05,
+      "loss": 1.3369,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 1.3681,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.456521739130435e-05,
+      "loss": 1.4715,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 1.4458,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.3767,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 1.4051,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5434782608695654e-05,
+      "loss": 1.5299,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 1.4234,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5869565217391306e-05,
+      "loss": 1.4651,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6086956521739132e-05,
+      "loss": 1.6008,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.630434782608696e-05,
+      "loss": 1.4812,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 1.3406,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.673913043478261e-05,
+      "loss": 1.4521,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6956521739130437e-05,
+      "loss": 1.5115,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7173913043478263e-05,
+      "loss": 1.3607,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 1.5419,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7608695652173915e-05,
+      "loss": 1.4382,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 1.4131,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8043478260869567e-05,
+      "loss": 1.2361,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 1.3589,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.847826086956522e-05,
+      "loss": 1.511,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 1.4226,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.891304347826087e-05,
+      "loss": 1.4287,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 1.4446,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9347826086956523e-05,
+      "loss": 1.3839,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 1.4631,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9782608695652176e-05,
+      "loss": 1.4006,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 1.4534,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999999431403446e-05,
+      "loss": 1.5352,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999977256144297e-05,
+      "loss": 1.4683,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999948826348907e-05,
+      "loss": 1.5199,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999909024680632e-05,
+      "loss": 1.398,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999857851184724e-05,
+      "loss": 1.5483,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999795305919377e-05,
+      "loss": 1.4268,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999721388955722e-05,
+      "loss": 1.2935,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999636100377818e-05,
+      "loss": 1.5527,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999539440282648e-05,
+      "loss": 1.2354,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999943140878014e-05,
+      "loss": 1.3866,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999931200599314e-05,
+      "loss": 1.4202,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999181232057437e-05,
+      "loss": 1.3781,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999039087121746e-05,
+      "loss": 1.1009,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998885571347712e-05,
+      "loss": 1.3709,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998720684909913e-05,
+      "loss": 1.4178,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999854442799585e-05,
+      "loss": 1.2833,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998356800805973e-05,
+      "loss": 1.3406,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998157803553638e-05,
+      "loss": 1.4118,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997947436465157e-05,
+      "loss": 1.4097,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997725699779747e-05,
+      "loss": 1.3415,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999749259374957e-05,
+      "loss": 1.4323,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999724811863971e-05,
+      "loss": 1.4016,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999699227472818e-05,
+      "loss": 1.4401,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996725062305934e-05,
+      "loss": 1.4507,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999644648167684e-05,
+      "loss": 1.387,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996156533157692e-05,
+      "loss": 1.2825,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999585521707822e-05,
+      "loss": 1.2307,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999554253378108e-05,
+      "loss": 1.301,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995218483621858e-05,
+      "loss": 1.302,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994883066969055e-05,
+      "loss": 1.342,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994536284204106e-05,
+      "loss": 1.5534,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994178135721372e-05,
+      "loss": 1.2546,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999380862192813e-05,
+      "loss": 1.3045,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99934277432446e-05,
+      "loss": 1.3729,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993035500103904e-05,
+      "loss": 1.4059,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992631892952108e-05,
+      "loss": 1.3092,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999221692224818e-05,
+      "loss": 1.2541,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991790588464034e-05,
+      "loss": 1.5616,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991352892084483e-05,
+      "loss": 1.3117,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999090383360728e-05,
+      "loss": 1.4864,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990443413543086e-05,
+      "loss": 1.1912,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998997163241549e-05,
+      "loss": 1.3293,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989488490761e-05,
+      "loss": 1.3906,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988993989129035e-05,
+      "loss": 1.2405,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988488128081943e-05,
+      "loss": 1.4372,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998797090819499e-05,
+      "loss": 1.2302,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987442330056352e-05,
+      "loss": 1.3168,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986902394267117e-05,
+      "loss": 1.3076,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986351101441308e-05,
+      "loss": 1.1699,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985788452205847e-05,
+      "loss": 1.3322,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985214447200566e-05,
+      "loss": 1.3678,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984629087078234e-05,
+      "loss": 1.2653,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998403237250451e-05,
+      "loss": 1.3188,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983424304157976e-05,
+      "loss": 1.4451,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998280488273012e-05,
+      "loss": 1.3479,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982174108925346e-05,
+      "loss": 1.4162,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998153198346097e-05,
+      "loss": 1.4135,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980878507067207e-05,
+      "loss": 1.3905,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980213680487186e-05,
+      "loss": 1.4447,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979537504476945e-05,
+      "loss": 1.4285,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997884997980543e-05,
+      "loss": 1.4105,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978151107254477e-05,
+      "loss": 1.2898,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977440887618855e-05,
+      "loss": 1.3787,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976719321706213e-05,
+      "loss": 1.419,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997598641033711e-05,
+      "loss": 1.4784,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997524215434501e-05,
+      "loss": 1.4202,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974486554576274e-05,
+      "loss": 1.2834,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973719611890164e-05,
+      "loss": 1.4078,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972941327158848e-05,
+      "loss": 1.2659,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997215170126738e-05,
+      "loss": 1.3769,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971350735113718e-05,
+      "loss": 1.3037,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970538429608716e-05,
+      "loss": 1.3615,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996971478567612e-05,
+      "loss": 1.4036,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996887980425258e-05,
+      "loss": 1.2713,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996803348628762e-05,
+      "loss": 1.3764,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967175832743677e-05,
+      "loss": 1.4328,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966306844596063e-05,
+      "loss": 1.4352,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965426522832985e-05,
+      "loss": 1.2599,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964534868455544e-05,
+      "loss": 1.3638,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963631882477716e-05,
+      "loss": 1.3307,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962717565926376e-05,
+      "loss": 1.36,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961791919841275e-05,
+      "loss": 1.3391,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960854945275055e-05,
+      "loss": 1.4955,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995990664329323e-05,
+      "loss": 1.4003,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958947014974216e-05,
+      "loss": 1.2154,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957976061409277e-05,
+      "loss": 1.3286,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956993783702594e-05,
+      "loss": 1.4478,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956000182971192e-05,
+      "loss": 1.2906,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954995260344993e-05,
+      "loss": 1.1524,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995397901696679e-05,
+      "loss": 1.3923,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995295145399224e-05,
+      "loss": 1.3211,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951912572589892e-05,
+      "loss": 1.3806,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950862373941145e-05,
+      "loss": 1.3799,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994980085924028e-05,
+      "loss": 1.3846,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994872802969445e-05,
+      "loss": 1.4632,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994764388652366e-05,
+      "loss": 1.3105,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946548430960796e-05,
+      "loss": 1.3987,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9945441664251605e-05,
+      "loss": 1.4065,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944323587654685e-05,
+      "loss": 1.4344,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943194202441514e-05,
+      "loss": 1.3795,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942053509896417e-05,
+      "loss": 1.552,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940901511316585e-05,
+      "loss": 1.3405,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993973820801206e-05,
+      "loss": 1.3728,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993856360130574e-05,
+      "loss": 1.4416,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937377692533382e-05,
+      "loss": 1.3926,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99361804830436e-05,
+      "loss": 1.2651,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934971974197843e-05,
+      "loss": 1.2976,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993375216737042e-05,
+      "loss": 1.4121,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932521063948494e-05,
+      "loss": 1.276,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931278665332057e-05,
+      "loss": 1.3322,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930024972933966e-05,
+      "loss": 1.3596,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928759988179908e-05,
+      "loss": 1.4512,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992748371250841e-05,
+      "loss": 1.4053,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992619614737085e-05,
+      "loss": 1.2646,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924897294231434e-05,
+      "loss": 1.5216,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923587154567214e-05,
+      "loss": 1.4538,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922265729868065e-05,
+      "loss": 1.4267,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992093302163671e-05,
+      "loss": 1.5343,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919589031388688e-05,
+      "loss": 1.3869,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991823376065238e-05,
+      "loss": 1.3431,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991686721096899e-05,
+      "loss": 1.2332,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991548938389255e-05,
+      "loss": 1.4032,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991410028098991e-05,
+      "loss": 1.3987,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912699903840754e-05,
+      "loss": 1.5462,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991128825403758e-05,
+      "loss": 1.3819,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909865333185704e-05,
+      "loss": 1.3906,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908431142903263e-05,
+      "loss": 1.4463,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906985684821208e-05,
+      "loss": 1.3995,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9905528960583305e-05,
+      "loss": 1.1931,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990406097184613e-05,
+      "loss": 1.328,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902581720279074e-05,
+      "loss": 1.3747,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901091207564326e-05,
+      "loss": 1.356,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899589435396888e-05,
+      "loss": 1.5297,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898076405484565e-05,
+      "loss": 1.2664,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896552119547964e-05,
+      "loss": 1.3391,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895016579320495e-05,
+      "loss": 1.5459,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989346978654836e-05,
+      "loss": 1.3014,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9891911742990567e-05,
+      "loss": 1.4365,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890342450418907e-05,
+      "loss": 1.5066,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888761910617967e-05,
+      "loss": 1.3875,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9887170125385135e-05,
+      "loss": 1.3043,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885567096530566e-05,
+      "loss": 1.3391,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883952825877225e-05,
+      "loss": 1.51,
+      "step": 235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988232731526084e-05,
+      "loss": 1.4932,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880690566529933e-05,
+      "loss": 1.3678,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987904258154581e-05,
+      "loss": 1.4132,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987738336218254e-05,
+      "loss": 1.4784,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9875712910326982e-05,
+      "loss": 1.2517,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987403122787876e-05,
+      "loss": 1.3915,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872338316750264e-05,
+      "loss": 1.3817,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870634178866676e-05,
+      "loss": 1.3482,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986891881616592e-05,
+      "loss": 1.4669,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867192230598692e-05,
+      "loss": 1.521,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986545442412846e-05,
+      "loss": 1.3481,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863705398731444e-05,
+      "loss": 1.2345,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986194515639662e-05,
+      "loss": 1.39,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986017369912573e-05,
+      "loss": 1.4151,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9858391028933258e-05,
+      "loss": 1.3328,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9856597147846447e-05,
+      "loss": 1.3425,
+      "step": 251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985479205790528e-05,
+      "loss": 1.485,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98529757611625e-05,
+      "loss": 1.2616,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851148259683586e-05,
+      "loss": 1.1877,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984930955554676e-05,
+      "loss": 1.489,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984745965084298e-05,
+      "loss": 1.4196,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845598547675945e-05,
+      "loss": 1.3344,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843726248162096e-05,
+      "loss": 1.3713,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984184275443059e-05,
+      "loss": 1.2307,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983994806862333e-05,
+      "loss": 1.4117,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9838042192894935e-05,
+      "loss": 1.4123,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836125129412756e-05,
+      "loss": 1.1802,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834196880356863e-05,
+      "loss": 1.4824,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983225744792005e-05,
+      "loss": 1.3801,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9830306834307828e-05,
+      "loss": 1.4142,
+      "step": 265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9828345041738413e-05,
+      "loss": 1.2993,
+      "step": 266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826372072442747e-05,
+      "loss": 1.24,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824387928664478e-05,
+      "loss": 1.3416,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9822392612659964e-05,
+      "loss": 1.4188,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9820386126698256e-05,
+      "loss": 1.401,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981836847306112e-05,
+      "loss": 1.262,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981633965404302e-05,
+      "loss": 1.1779,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9814299671951113e-05,
+      "loss": 1.3676,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9812248529105254e-05,
+      "loss": 1.3851,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810186227837986e-05,
+      "loss": 1.4388,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808112770494546e-05,
+      "loss": 1.1772,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806028159432854e-05,
+      "loss": 1.2482,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9803932397023512e-05,
+      "loss": 1.3222,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801825485649816e-05,
+      "loss": 1.3487,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799707427707723e-05,
+      "loss": 1.1701,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9797578225605875e-05,
+      "loss": 1.3803,
+      "step": 281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979543788176559e-05,
+      "loss": 1.2782,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979328639862085e-05,
+      "loss": 1.4364,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9791123778618303e-05,
+      "loss": 1.3674,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978895002421727e-05,
+      "loss": 1.3909,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978676513788973e-05,
+      "loss": 1.3825,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978456912212032e-05,
+      "loss": 1.2967,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9782361979406334e-05,
+      "loss": 1.2333,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780143712257715e-05,
+      "loss": 1.3207,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9777914323197063e-05,
+      "loss": 1.322,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977567381475963e-05,
+      "loss": 1.4313,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97734221894933e-05,
+      "loss": 1.4255,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9771159449958612e-05,
+      "loss": 1.3717,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9768885598728734e-05,
+      "loss": 1.2229,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976660063838947e-05,
+      "loss": 1.4424,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764304571539266e-05,
+      "loss": 1.3519,
+      "step": 296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9761997400789194e-05,
+      "loss": 1.3743,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759679128762946e-05,
+      "loss": 1.3588,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9757349758096855e-05,
+      "loss": 1.3905,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975500929143986e-05,
+      "loss": 1.4055,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752657731453523e-05,
+      "loss": 1.3259,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750295080812025e-05,
+      "loss": 1.342,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974792134220215e-05,
+      "loss": 1.4263,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745536518323304e-05,
+      "loss": 1.393,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743140611887487e-05,
+      "loss": 1.3359,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974073362561931e-05,
+      "loss": 1.4202,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738315562255983e-05,
+      "loss": 1.3547,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9735886424547306e-05,
+      "loss": 1.4709,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973344621525568e-05,
+      "loss": 1.2986,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97309949371561e-05,
+      "loss": 1.2082,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728532593036132e-05,
+      "loss": 1.3314,
+      "step": 311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972605918569594e-05,
+      "loss": 1.387,
+      "step": 312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972357471794827e-05,
+      "loss": 1.1319,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972107919261844e-05,
+      "loss": 1.2984,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9718572612544345e-05,
+      "loss": 1.4288,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9716054980576446e-05,
+      "loss": 1.213,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971352629957778e-05,
+      "loss": 1.3471,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710986572423947e-05,
+      "loss": 1.2475,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9708435802003105e-05,
+      "loss": 1.4886,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9705873991215973e-05,
+      "loss": 1.3517,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970330114297583e-05,
+      "loss": 1.4266,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970071726020849e-05,
+      "loss": 1.3975,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969812234585233e-05,
+      "loss": 1.2978,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695516402858277e-05,
+      "loss": 1.2647,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9692899434189786e-05,
+      "loss": 1.3324,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969027144282285e-05,
+      "loss": 1.475,
+      "step": 326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9687632431746014e-05,
+      "loss": 1.2854,
+      "step": 327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9684982403960332e-05,
+      "loss": 1.3711,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968232136247941e-05,
+      "loss": 1.3683,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9679649310329353e-05,
+      "loss": 1.3178,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9676966250548807e-05,
+      "loss": 1.3448,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967427218618893e-05,
+      "loss": 1.3387,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967156712031339e-05,
+      "loss": 1.3689,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966885105599837e-05,
+      "loss": 1.3894,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9666123996332564e-05,
+      "loss": 1.3845,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966338594441716e-05,
+      "loss": 1.3974,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660636903365854e-05,
+      "loss": 1.4133,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9657876876304835e-05,
+      "loss": 1.2042,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965510586637279e-05,
+      "loss": 1.3591,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965232387672089e-05,
+      "loss": 1.4954,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9649530910512796e-05,
+      "loss": 1.3161,
+      "step": 341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9646726970924645e-05,
+      "loss": 1.374,
+      "step": 342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643912061145065e-05,
+      "loss": 1.4118,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9641086184375148e-05,
+      "loss": 1.3582,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9638249343828463e-05,
+      "loss": 1.2793,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635401542731043e-05,
+      "loss": 1.4125,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963254278432139e-05,
+      "loss": 1.3889,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9629673071850463e-05,
+      "loss": 1.2159,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962679240858168e-05,
+      "loss": 1.2363,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623900797790914e-05,
+      "loss": 1.2471,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9620998242766482e-05,
+      "loss": 1.3256,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9618084746809147e-05,
+      "loss": 1.2357,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9615160313232122e-05,
+      "loss": 1.4569,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961222494536105e-05,
+      "loss": 1.4797,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960927864653401e-05,
+      "loss": 1.2745,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9606321420101514e-05,
+      "loss": 1.3715,
+      "step": 356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96033532694265e-05,
+      "loss": 1.2943,
+      "step": 357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960037419788433e-05,
+      "loss": 1.338,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9597384208862776e-05,
+      "loss": 1.246,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9594383305762045e-05,
+      "loss": 1.3237,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591371491994735e-05,
+      "loss": 1.3279,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958834877098586e-05,
+      "loss": 1.3185,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9585315146172843e-05,
+      "loss": 1.3534,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9582270621005494e-05,
+      "loss": 1.1566,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9579215198946028e-05,
+      "loss": 1.2134,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9576148883469054e-05,
+      "loss": 1.3722,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9573071678061565e-05,
+      "loss": 1.3042,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956998358622293e-05,
+      "loss": 1.4297,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566884611464916e-05,
+      "loss": 1.3346,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9563774757311646e-05,
+      "loss": 1.2136,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956065402729963e-05,
+      "loss": 1.4095,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9557522424977745e-05,
+      "loss": 1.2655,
+      "step": 372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9554379953907217e-05,
+      "loss": 1.3623,
+      "step": 373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955122661766165e-05,
+      "loss": 1.3615,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954806241982699e-05,
+      "loss": 1.4546,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954488736400155e-05,
+      "loss": 1.5023,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9541701453795976e-05,
+      "loss": 1.3962,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9538504692833263e-05,
+      "loss": 1.313,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9535297084748747e-05,
+      "loss": 1.2892,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532078633190097e-05,
+      "loss": 1.3584,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528849341817316e-05,
+      "loss": 1.4247,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525609214302727e-05,
+      "loss": 1.4444,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9522358254330987e-05,
+      "loss": 1.3144,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951909646559906e-05,
+      "loss": 1.4022,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951582385181623e-05,
+      "loss": 1.3433,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512540416704095e-05,
+      "loss": 1.4016,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950924616399655e-05,
+      "loss": 1.3799,
+      "step": 387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9505941097439805e-05,
+      "loss": 1.4395,
+      "step": 388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9502625220792345e-05,
+      "loss": 1.303,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9499298537824972e-05,
+      "loss": 1.3902,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949596105232077e-05,
+      "loss": 1.3047,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9492612768075094e-05,
+      "loss": 1.1657,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489253688895594e-05,
+      "loss": 1.4182,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485883818602194e-05,
+      "loss": 1.205,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948250316102709e-05,
+      "loss": 1.3438,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9479111720014737e-05,
+      "loss": 1.3099,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947570949942186e-05,
+      "loss": 1.346,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472296503117437e-05,
+      "loss": 1.4133,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9468872734982705e-05,
+      "loss": 1.5194,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465438198911158e-05,
+      "loss": 1.3365,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946199289880852e-05,
+      "loss": 1.3004,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458536838592764e-05,
+      "loss": 1.4811,
+      "step": 402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9455070022194094e-05,
+      "loss": 1.3259,
+      "step": 403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451592453554956e-05,
+      "loss": 1.2946,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944810413663001e-05,
+      "loss": 1.2141,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9444605075386156e-05,
+      "loss": 1.4021,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944109527380249e-05,
+      "loss": 1.455,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9437574735870346e-05,
+      "loss": 1.4715,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9434043465593247e-05,
+      "loss": 1.4539,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9430501466986932e-05,
+      "loss": 1.4801,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9426948744079338e-05,
+      "loss": 1.2148,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94233853009106e-05,
+      "loss": 1.2596,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9419811141533032e-05,
+      "loss": 1.2145,
+      "step": 413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416226270011152e-05,
+      "loss": 1.2662,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9412630690421648e-05,
+      "loss": 1.302,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940902440685339e-05,
+      "loss": 1.2591,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405407423407414e-05,
+      "loss": 1.4317,
+      "step": 417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9401779744196934e-05,
+      "loss": 1.3283,
+      "step": 418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9398141373347317e-05,
+      "loss": 1.2165,
+      "step": 419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93944923149961e-05,
+      "loss": 1.3613,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9390832573292964e-05,
+      "loss": 1.3744,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938716215239974e-05,
+      "loss": 1.1967,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9383481056490402e-05,
+      "loss": 1.2715,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9379789289751073e-05,
+      "loss": 1.3597,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376086856380007e-05,
+      "loss": 1.4245,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937237376058758e-05,
+      "loss": 1.3464,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93686500065963e-05,
+      "loss": 1.4296,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9364915598640796e-05,
+      "loss": 1.2627,
+      "step": 428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936117054096781e-05,
+      "loss": 1.2727,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9357414837836192e-05,
+      "loss": 1.2721,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935364849351691e-05,
+      "loss": 1.3247,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934987151229302e-05,
+      "loss": 1.5809,
+      "step": 432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934608389845968e-05,
+      "loss": 1.3678,
+      "step": 433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9342285656324135e-05,
+      "loss": 1.4506,
+      "step": 434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9338476790205726e-05,
+      "loss": 1.2935,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933465730443586e-05,
+      "loss": 1.1895,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9330827203358044e-05,
+      "loss": 1.248,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932698649132783e-05,
+      "loss": 1.3905,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932313517271285e-05,
+      "loss": 1.3918,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9319273251892805e-05,
+      "loss": 1.3228,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931540073325944e-05,
+      "loss": 1.4809,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9311517621216555e-05,
+      "loss": 1.4098,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930762392018e-05,
+      "loss": 1.2479,
+      "step": 443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9303719634577668e-05,
+      "loss": 1.4779,
+      "step": 444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929980476884948e-05,
+      "loss": 1.2986,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.92958793274474e-05,
+      "loss": 1.3807,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9291943314835406e-05,
+      "loss": 1.535,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928799673548951e-05,
+      "loss": 1.1961,
+      "step": 448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9284039593897734e-05,
+      "loss": 1.327,
+      "step": 449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928007189456011e-05,
+      "loss": 1.5256,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276093641988684e-05,
+      "loss": 1.5381,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9272104840707487e-05,
+      "loss": 1.323,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9268105495252568e-05,
+      "loss": 1.3281,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9264095610171944e-05,
+      "loss": 1.3403,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926007519002564e-05,
+      "loss": 1.3414,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925604423938564e-05,
+      "loss": 1.2793,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9252002762835917e-05,
+      "loss": 1.3011,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924795076497241e-05,
+      "loss": 1.2783,
+      "step": 458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924388825040303e-05,
+      "loss": 1.2557,
+      "step": 459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9239815223747632e-05,
+      "loss": 1.3768,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9235731689638035e-05,
+      "loss": 1.2036,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9231637652718004e-05,
+      "loss": 1.2012,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9227533117643256e-05,
+      "loss": 1.4446,
+      "step": 463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922341808908144e-05,
+      "loss": 1.3174,
+      "step": 464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921929257171213e-05,
+      "loss": 1.3249,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9215156570226844e-05,
+      "loss": 1.255,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921101008932901e-05,
+      "loss": 1.2195,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920685313373398e-05,
+      "loss": 1.3089,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920268570816901e-05,
+      "loss": 1.2939,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9198507817373274e-05,
+      "loss": 1.4276,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919431946609784e-05,
+      "loss": 1.3933,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9190120659105663e-05,
+      "loss": 1.3057,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9185911401171615e-05,
+      "loss": 1.3188,
+      "step": 473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918169169708242e-05,
+      "loss": 1.2174,
+      "step": 474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.91774615516367e-05,
+      "loss": 1.4041,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9173220969644947e-05,
+      "loss": 1.1018,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916896995592953e-05,
+      "loss": 1.3083,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916470851532466e-05,
+      "loss": 1.1981,
+      "step": 478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9160436652676425e-05,
+      "loss": 1.4576,
+      "step": 479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9156154372842757e-05,
+      "loss": 1.1978,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9151861680693432e-05,
+      "loss": 1.193,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914755858111008e-05,
+      "loss": 1.2737,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914324507898614e-05,
+      "loss": 1.3568,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9138921179226908e-05,
+      "loss": 1.4216,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9134586886749492e-05,
+      "loss": 1.2753,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9130242206482817e-05,
+      "loss": 1.3887,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912588714336762e-05,
+      "loss": 1.3849,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912152170235646e-05,
+      "loss": 1.3427,
+      "step": 488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9117145888413678e-05,
+      "loss": 1.3543,
+      "step": 489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911275970651542e-05,
+      "loss": 1.3034,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910836316164962e-05,
+      "loss": 1.4744,
+      "step": 491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9103956258816008e-05,
+      "loss": 1.3314,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.909953900302607e-05,
+      "loss": 1.4215,
+      "step": 493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.909511139930309e-05,
+      "loss": 1.2962,
+      "step": 494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9090673452682108e-05,
+      "loss": 1.3132,
+      "step": 495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9086225168209917e-05,
+      "loss": 1.2975,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081766550945084e-05,
+      "loss": 1.3207,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907729760595791e-05,
+      "loss": 1.3353,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9072818338330455e-05,
+      "loss": 1.2535,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9068328753156514e-05,
+      "loss": 1.0754,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9063828855541605e-05,
+      "loss": 1.3487,
+      "step": 501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9059318650602986e-05,
+      "loss": 1.4404,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905479814346962e-05,
+      "loss": 1.2397,
+      "step": 503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9050267339282214e-05,
+      "loss": 1.378,
+      "step": 504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9045726243193155e-05,
+      "loss": 1.2465,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904117486036655e-05,
+      "loss": 1.3214,
+      "step": 506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9036613195978198e-05,
+      "loss": 1.3161,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9032041255215596e-05,
+      "loss": 1.2936,
+      "step": 508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9027459043277923e-05,
+      "loss": 1.436,
+      "step": 509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902286656537603e-05,
+      "loss": 1.2758,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9018263826732467e-05,
+      "loss": 1.3259,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9013650832581424e-05,
+      "loss": 1.3638,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9009027588168773e-05,
+      "loss": 1.357,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900439409875203e-05,
+      "loss": 1.3922,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899975036960037e-05,
+      "loss": 1.2587,
+      "step": 515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8995096405994614e-05,
+      "loss": 1.4294,
+      "step": 516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899043221322721e-05,
+      "loss": 1.244,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898575779660225e-05,
+      "loss": 1.1213,
+      "step": 518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8981073161435447e-05,
+      "loss": 1.4039,
+      "step": 519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897637831305414e-05,
+      "loss": 1.3606,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8971673256797275e-05,
+      "loss": 1.4622,
+      "step": 521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8966957998015403e-05,
+      "loss": 1.3161,
+      "step": 522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896223254207069e-05,
+      "loss": 1.3393,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89574968943369e-05,
+      "loss": 1.3398,
+      "step": 524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895275106019937e-05,
+      "loss": 1.462,
+      "step": 525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8947995045055023e-05,
+      "loss": 1.3711,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894322885431238e-05,
+      "loss": 1.4027,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8938452493391515e-05,
+      "loss": 1.4405,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8933665967724072e-05,
+      "loss": 1.3696,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892886928275325e-05,
+      "loss": 1.3695,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892406244393382e-05,
+      "loss": 1.2519,
+      "step": 531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891924545673207e-05,
+      "loss": 1.438,
+      "step": 532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891441832662586e-05,
+      "loss": 1.2753,
+      "step": 533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8909581059104557e-05,
+      "loss": 1.3365,
+      "step": 534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8904733659669072e-05,
+      "loss": 1.3209,
+      "step": 535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8899876133831835e-05,
+      "loss": 1.1781,
+      "step": 536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8895008487116793e-05,
+      "loss": 1.3314,
+      "step": 537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8890130725059396e-05,
+      "loss": 1.2742,
+      "step": 538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8885242853206606e-05,
+      "loss": 1.3207,
+      "step": 539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.888034487711687e-05,
+      "loss": 1.2713,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8875436802360142e-05,
+      "loss": 1.4202,
+      "step": 541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8870518634517843e-05,
+      "loss": 1.3311,
+      "step": 542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8865590379182885e-05,
+      "loss": 1.3785,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8860652041959643e-05,
+      "loss": 1.3968,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8855703628463957e-05,
+      "loss": 1.3965,
+      "step": 545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8850745144323137e-05,
+      "loss": 1.2262,
+      "step": 546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884577659517593e-05,
+      "loss": 1.358,
+      "step": 547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884079798667254e-05,
+      "loss": 1.286,
+      "step": 548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88358093244746e-05,
+      "loss": 1.3762,
+      "step": 549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883081061425519e-05,
+      "loss": 1.3745,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882580186169881e-05,
+      "loss": 1.2543,
+      "step": 551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882078307250137e-05,
+      "loss": 1.3128,
+      "step": 552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8815754252370205e-05,
+      "loss": 1.4083,
+      "step": 553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881071540702406e-05,
+      "loss": 1.4121,
+      "step": 554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8805666542193073e-05,
+      "loss": 1.2899,
+      "step": 555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8800607663618774e-05,
+      "loss": 1.4122,
+      "step": 556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879553877705409e-05,
+      "loss": 1.2897,
+      "step": 557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8790459888263323e-05,
+      "loss": 1.3716,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8785371003022145e-05,
+      "loss": 1.2673,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8780272127117606e-05,
+      "loss": 1.2832,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8775163266348116e-05,
+      "loss": 1.3121,
+      "step": 561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877004442652343e-05,
+      "loss": 1.2356,
+      "step": 562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8764915613464657e-05,
+      "loss": 1.2424,
+      "step": 563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8759776833004256e-05,
+      "loss": 1.4182,
+      "step": 564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8754628090986e-05,
+      "loss": 1.141,
+      "step": 565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8749469393265016e-05,
+      "loss": 1.3906,
+      "step": 566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874430074570773e-05,
+      "loss": 1.2956,
+      "step": 567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87391221541919e-05,
+      "loss": 1.3609,
+      "step": 568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873393362460658e-05,
+      "loss": 1.3083,
+      "step": 569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8728735162852125e-05,
+      "loss": 1.4665,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.87235267748402e-05,
+      "loss": 1.3554,
+      "step": 571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8718308466493745e-05,
+      "loss": 1.3665,
+      "step": 572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871308024374698e-05,
+      "loss": 1.4273,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8707842112545406e-05,
+      "loss": 1.344,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870259407884579e-05,
+      "loss": 1.4037,
+      "step": 575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8697336148616162e-05,
+      "loss": 1.3336,
+      "step": 576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.86920683278358e-05,
+      "loss": 1.3447,
+      "step": 577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868679062249524e-05,
+      "loss": 1.3287,
+      "step": 578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8681503038596248e-05,
+      "loss": 1.2865,
+      "step": 579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8676205582151822e-05,
+      "loss": 1.1761,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8670898259186202e-05,
+      "loss": 1.2776,
+      "step": 581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8665581075734836e-05,
+      "loss": 1.2189,
+      "step": 582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 1.3302,
+      "step": 583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865491715157273e-05,
+      "loss": 1.3992,
+      "step": 584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864957042298893e-05,
+      "loss": 1.3607,
+      "step": 585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864421385817325e-05,
+      "loss": 1.3804,
+      "step": 586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8638847463217145e-05,
+      "loss": 1.2228,
+      "step": 587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8633471244223232e-05,
+      "loss": 1.2283,
+      "step": 588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8628085207305325e-05,
+      "loss": 1.3609,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8622689358588373e-05,
+      "loss": 1.3056,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8617283704208507e-05,
+      "loss": 1.3889,
+      "step": 591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8611868250312994e-05,
+      "loss": 1.389,
+      "step": 592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860644300306026e-05,
+      "loss": 1.3262,
+      "step": 593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8601007968619847e-05,
+      "loss": 1.3581,
+      "step": 594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8595563153172448e-05,
+      "loss": 1.3574,
+      "step": 595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8590108562909862e-05,
+      "loss": 1.1755,
+      "step": 596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858464420403502e-05,
+      "loss": 1.4054,
+      "step": 597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857917008276195e-05,
+      "loss": 1.3015,
+      "step": 598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857368620531578e-05,
+      "loss": 1.3561,
+      "step": 599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8568192577932746e-05,
+      "loss": 1.4332,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8562689206860155e-05,
+      "loss": 1.3722,
+      "step": 601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8557176098356406e-05,
+      "loss": 1.2927,
+      "step": 602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8551653258690972e-05,
+      "loss": 1.4575,
+      "step": 603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854612069414438e-05,
+      "loss": 1.4379,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8540578411008233e-05,
+      "loss": 1.3694,
+      "step": 605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8535026415585168e-05,
+      "loss": 1.2722,
+      "step": 606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8529464714188888e-05,
+      "loss": 1.192,
+      "step": 607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852389331314411e-05,
+      "loss": 1.2234,
+      "step": 608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85183122187866e-05,
+      "loss": 1.3328,
+      "step": 609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8512721437463136e-05,
+      "loss": 1.2733,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8507120975531517e-05,
+      "loss": 1.3759,
+      "step": 611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850151083936055e-05,
+      "loss": 1.2952,
+      "step": 612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849589103533004e-05,
+      "loss": 1.3329,
+      "step": 613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.84902615698308e-05,
+      "loss": 1.4101,
+      "step": 614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8484622449264608e-05,
+      "loss": 1.3062,
+      "step": 615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8478973680044237e-05,
+      "loss": 1.3987,
+      "step": 616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8473315268593432e-05,
+      "loss": 1.3151,
+      "step": 617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846764722134689e-05,
+      "loss": 1.3956,
+      "step": 618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8461969544750286e-05,
+      "loss": 1.392,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845628224526023e-05,
+      "loss": 1.5057,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8450585329344277e-05,
+      "loss": 1.2634,
+      "step": 621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8444878803480925e-05,
+      "loss": 1.3566,
+      "step": 622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8439162674159595e-05,
+      "loss": 1.3247,
+      "step": 623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8433436947880626e-05,
+      "loss": 1.3977,
+      "step": 624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842770163115528e-05,
+      "loss": 1.3899,
+      "step": 625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842195673050572e-05,
+      "loss": 1.308,
+      "step": 626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8416202252465e-05,
+      "loss": 1.4328,
+      "step": 627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841043820357708e-05,
+      "loss": 1.3289,
+      "step": 628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8404664590396796e-05,
+      "loss": 1.3466,
+      "step": 629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8398881419489854e-05,
+      "loss": 1.3224,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8393088697432847e-05,
+      "loss": 1.3337,
+      "step": 631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.838728643081321e-05,
+      "loss": 1.3094,
+      "step": 632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8381474626229245e-05,
+      "loss": 1.3409,
+      "step": 633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8375653290290093e-05,
+      "loss": 1.4036,
+      "step": 634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836982242961574e-05,
+      "loss": 1.266,
+      "step": 635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8363982050836996e-05,
+      "loss": 1.2164,
+      "step": 636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835813216059551e-05,
+      "loss": 1.3615,
+      "step": 637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8352272765543723e-05,
+      "loss": 1.3961,
+      "step": 638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8346403872344907e-05,
+      "loss": 1.3247,
+      "step": 639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8340525487673128e-05,
+      "loss": 1.4376,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8334637618213238e-05,
+      "loss": 1.3172,
+      "step": 641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8328740270660887e-05,
+      "loss": 1.3366,
+      "step": 642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83228334517225e-05,
+      "loss": 1.2715,
+      "step": 643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831691716811526e-05,
+      "loss": 1.387,
+      "step": 644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8310991426567137e-05,
+      "loss": 1.2501,
+      "step": 645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8305056233816833e-05,
+      "loss": 1.1845,
+      "step": 646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8299111596613815e-05,
+      "loss": 1.4116,
+      "step": 647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8293157521718284e-05,
+      "loss": 1.2341,
+      "step": 648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8287194015901167e-05,
+      "loss": 1.2673,
+      "step": 649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8281221085944128e-05,
+      "loss": 1.3124,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827523873863954e-05,
+      "loss": 1.3819,
+      "step": 651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8269246980790482e-05,
+      "loss": 1.2699,
+      "step": 652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8263245819210745e-05,
+      "loss": 1.4479,
+      "step": 653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.825723526072481e-05,
+      "loss": 1.2992,
+      "step": 654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8251215312167833e-05,
+      "loss": 1.2057,
+      "step": 655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8245185980385673e-05,
+      "loss": 1.2626,
+      "step": 656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823914727223483e-05,
+      "loss": 1.2323,
+      "step": 657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823309919458249e-05,
+      "loss": 1.2677,
+      "step": 658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8227041754306484e-05,
+      "loss": 1.3755,
+      "step": 659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822097495829529e-05,
+      "loss": 1.1518,
+      "step": 660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.821489881344803e-05,
+      "loss": 1.3471,
+      "step": 661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8208813326674444e-05,
+      "loss": 1.2778,
+      "step": 662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8202718504894917e-05,
+      "loss": 1.3122,
+      "step": 663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8196614355040436e-05,
+      "loss": 1.4677,
+      "step": 664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819050088405259e-05,
+      "loss": 1.284,
+      "step": 665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818437809888359e-05,
+      "loss": 1.4112,
+      "step": 666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817824600649621e-05,
+      "loss": 1.3281,
+      "step": 667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8172104613863838e-05,
+      "loss": 1.329,
+      "step": 668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816595392797041e-05,
+      "loss": 1.1344,
+      "step": 669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8159793955810453e-05,
+      "loss": 1.4005,
+      "step": 670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815362470438904e-05,
+      "loss": 1.3454,
+      "step": 671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8147446180721802e-05,
+      "loss": 1.3562,
+      "step": 672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8141258391834914e-05,
+      "loss": 1.2067,
+      "step": 673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8135061344765088e-05,
+      "loss": 1.2327,
+      "step": 674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812885504655956e-05,
+      "loss": 1.1144,
+      "step": 675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8122639504276096e-05,
+      "loss": 1.2406,
+      "step": 676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811641472498296e-05,
+      "loss": 1.3028,
+      "step": 677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8110180715758934e-05,
+      "loss": 1.3957,
+      "step": 678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8103937483693288e-05,
+      "loss": 1.3159,
+      "step": 679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809768503588578e-05,
+      "loss": 1.4038,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8091423379446658e-05,
+      "loss": 1.3685,
+      "step": 681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8085152521496627e-05,
+      "loss": 1.3142,
+      "step": 682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8078872469166864e-05,
+      "loss": 1.3661,
+      "step": 683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8072583229599004e-05,
+      "loss": 1.3521,
+      "step": 684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8066284809945128e-05,
+      "loss": 1.4548,
+      "step": 685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8059977217367753e-05,
+      "loss": 1.2938,
+      "step": 686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8053660459039833e-05,
+      "loss": 1.2422,
+      "step": 687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8047334542144738e-05,
+      "loss": 1.3967,
+      "step": 688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8040999473876258e-05,
+      "loss": 1.428,
+      "step": 689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8034655261438588e-05,
+      "loss": 1.2242,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8028301912046325e-05,
+      "loss": 1.3014,
+      "step": 691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8021939432924456e-05,
+      "loss": 1.3546,
+      "step": 692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8015567831308344e-05,
+      "loss": 1.3246,
+      "step": 693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800918711444373e-05,
+      "loss": 1.3759,
+      "step": 694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8002797289586726e-05,
+      "loss": 1.3185,
+      "step": 695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7996398364003797e-05,
+      "loss": 1.3874,
+      "step": 696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7989990344971747e-05,
+      "loss": 1.277,
+      "step": 697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7983573239777746e-05,
+      "loss": 1.3898,
+      "step": 698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797714705571927e-05,
+      "loss": 1.3666,
+      "step": 699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7970711800104138e-05,
+      "loss": 1.3797,
+      "step": 700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7964267480250476e-05,
+      "loss": 1.3042,
+      "step": 701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7957814103486716e-05,
+      "loss": 1.2159,
+      "step": 702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7951351677151604e-05,
+      "loss": 1.3311,
+      "step": 703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7944880208594156e-05,
+      "loss": 1.3292,
+      "step": 704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7938399705173687e-05,
+      "loss": 1.443,
+      "step": 705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7931910174259777e-05,
+      "loss": 1.148,
+      "step": 706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792541162323228e-05,
+      "loss": 1.3056,
+      "step": 707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7918904059481303e-05,
+      "loss": 1.1624,
+      "step": 708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.79123874904072e-05,
+      "loss": 1.2849,
+      "step": 709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7905861923420568e-05,
+      "loss": 1.1711,
+      "step": 710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7899327365942242e-05,
+      "loss": 1.3396,
+      "step": 711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7892783825403272e-05,
+      "loss": 1.3184,
+      "step": 712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.788623130924493e-05,
+      "loss": 1.3957,
+      "step": 713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7879669824918685e-05,
+      "loss": 1.3386,
+      "step": 714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7873099379886217e-05,
+      "loss": 1.2501,
+      "step": 715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7866519981619396e-05,
+      "loss": 1.4069,
+      "step": 716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7859931637600256e-05,
+      "loss": 1.4704,
+      "step": 717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7853334355321025e-05,
+      "loss": 1.2778,
+      "step": 718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784672814228409e-05,
+      "loss": 1.3288,
+      "step": 719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784011300600198e-05,
+      "loss": 1.4501,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783348895399739e-05,
+      "loss": 1.3109,
+      "step": 721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7826855993803148e-05,
+      "loss": 1.2988,
+      "step": 722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7820214132962205e-05,
+      "loss": 1.4882,
+      "step": 723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7813563379027644e-05,
+      "loss": 1.5883,
+      "step": 724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7806903739562652e-05,
+      "loss": 1.2206,
+      "step": 725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7800235222140525e-05,
+      "loss": 1.6747,
+      "step": 726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7793557834344662e-05,
+      "loss": 1.2666,
+      "step": 727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7786871583768536e-05,
+      "loss": 1.2639,
+      "step": 728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7780176478015706e-05,
+      "loss": 1.2828,
+      "step": 729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7773472524699803e-05,
+      "loss": 1.2552,
+      "step": 730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7766759731444512e-05,
+      "loss": 1.2183,
+      "step": 731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776003810588358e-05,
+      "loss": 1.3489,
+      "step": 732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7753307655660792e-05,
+      "loss": 1.3426,
+      "step": 733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7746568388429966e-05,
+      "loss": 1.3664,
+      "step": 734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7739820311854957e-05,
+      "loss": 1.2674,
+      "step": 735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7733063433609623e-05,
+      "loss": 1.4536,
+      "step": 736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7726297761377847e-05,
+      "loss": 1.2702,
+      "step": 737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7719523302853498e-05,
+      "loss": 1.3485,
+      "step": 738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771274006574045e-05,
+      "loss": 1.4078,
+      "step": 739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7705948057752547e-05,
+      "loss": 1.1793,
+      "step": 740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769914728661362e-05,
+      "loss": 1.3122,
+      "step": 741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7692337760057454e-05,
+      "loss": 1.3285,
+      "step": 742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.76855194858278e-05,
+      "loss": 1.4051,
+      "step": 743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767869247167835e-05,
+      "loss": 1.3203,
+      "step": 744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7671856725372737e-05,
+      "loss": 1.3042,
+      "step": 745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7665012254684522e-05,
+      "loss": 1.2165,
+      "step": 746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.76581590673972e-05,
+      "loss": 1.3246,
+      "step": 747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7651297171304157e-05,
+      "loss": 1.3853,
+      "step": 748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7644426574208698e-05,
+      "loss": 1.3546,
+      "step": 749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763754728392402e-05,
+      "loss": 1.3478,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7630659308273204e-05,
+      "loss": 1.2798,
+      "step": 751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7623762655089208e-05,
+      "loss": 1.2209,
+      "step": 752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761685733221486e-05,
+      "loss": 1.3514,
+      "step": 753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7609943347502846e-05,
+      "loss": 1.3803,
+      "step": 754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7603020708815695e-05,
+      "loss": 1.3253,
+      "step": 755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7596089424025794e-05,
+      "loss": 1.2287,
+      "step": 756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7589149501015345e-05,
+      "loss": 1.5018,
+      "step": 757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758220094767638e-05,
+      "loss": 1.1544,
+      "step": 758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757524377191075e-05,
+      "loss": 1.2132,
+      "step": 759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7568277981630107e-05,
+      "loss": 1.3695,
+      "step": 760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7561303584755903e-05,
+      "loss": 1.2031,
+      "step": 761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7554320589219368e-05,
+      "loss": 1.285,
+      "step": 762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754732900296151e-05,
+      "loss": 1.2214,
+      "step": 763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754032883393313e-05,
+      "loss": 1.3069,
+      "step": 764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7533320090094763e-05,
+      "loss": 1.3448,
+      "step": 765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7526302779416703e-05,
+      "loss": 1.287,
+      "step": 766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7519276909878988e-05,
+      "loss": 1.3828,
+      "step": 767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751224248947139e-05,
+      "loss": 1.4209,
+      "step": 768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7505199526193398e-05,
+      "loss": 1.2545,
+      "step": 769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749814802805423e-05,
+      "loss": 1.3509,
+      "step": 770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7491088003072798e-05,
+      "loss": 1.2408,
+      "step": 771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748401945927771e-05,
+      "loss": 1.3241,
+      "step": 772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7476942404707267e-05,
+      "loss": 1.2471,
+      "step": 773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746985684740945e-05,
+      "loss": 1.356,
+      "step": 774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.74627627954419e-05,
+      "loss": 1.297,
+      "step": 775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.745566025687193e-05,
+      "loss": 1.3418,
+      "step": 776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.74485492397765e-05,
+      "loss": 1.3253,
+      "step": 777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.74414297522422e-05,
+      "loss": 1.3031,
+      "step": 778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7434301802365264e-05,
+      "loss": 1.3136,
+      "step": 779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7427165398251557e-05,
+      "loss": 1.2255,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7420020548016538e-05,
+      "loss": 1.2092,
+      "step": 781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7412867259785286e-05,
+      "loss": 1.3789,
+      "step": 782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7405705541692476e-05,
+      "loss": 1.2396,
+      "step": 783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7398535401882353e-05,
+      "loss": 1.2194,
+      "step": 784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739135684850876e-05,
+      "loss": 1.3393,
+      "step": 785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7384169889735094e-05,
+      "loss": 1.3454,
+      "step": 786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7376974533734317e-05,
+      "loss": 1.3465,
+      "step": 787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7369770788688937e-05,
+      "loss": 1.3755,
+      "step": 788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7362558662791003e-05,
+      "loss": 1.2685,
+      "step": 789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7355338164242098e-05,
+      "loss": 1.2445,
+      "step": 790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734810930125332e-05,
+      "loss": 1.4126,
+      "step": 791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734087208204528e-05,
+      "loss": 1.2169,
+      "step": 792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.73336265148481e-05,
+      "loss": 1.3358,
+      "step": 793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7326372607901388e-05,
+      "loss": 1.364,
+      "step": 794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7319110369454232e-05,
+      "loss": 1.309,
+      "step": 795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7311839807765204e-05,
+      "loss": 1.2683,
+      "step": 796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7304560931102334e-05,
+      "loss": 1.3465,
+      "step": 797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.729727374774311e-05,
+      "loss": 1.2255,
+      "step": 798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7289978265974466e-05,
+      "loss": 1.2879,
+      "step": 799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728267449409278e-05,
+      "loss": 1.28,
+      "step": 800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7275362440403844e-05,
+      "loss": 1.226,
+      "step": 801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7268042113222877e-05,
+      "loss": 1.3465,
+      "step": 802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726071352087451e-05,
+      "loss": 1.3905,
+      "step": 803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725337667169276e-05,
+      "loss": 1.3734,
+      "step": 804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7246031574021047e-05,
+      "loss": 1.3877,
+      "step": 805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.723867823621216e-05,
+      "loss": 1.4236,
+      "step": 806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7231316666628268e-05,
+      "loss": 1.4508,
+      "step": 807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7223946873640897e-05,
+      "loss": 1.3736,
+      "step": 808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7216568865630924e-05,
+      "loss": 1.3627,
+      "step": 809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720918265098857e-05,
+      "loss": 1.3579,
+      "step": 810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720178823811338e-05,
+      "loss": 1.3208,
+      "step": 811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7194385635414242e-05,
+      "loss": 1.3973,
+      "step": 812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718697485130934e-05,
+      "loss": 1.2564,
+      "step": 813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717955589422616e-05,
+      "loss": 1.1836,
+      "step": 814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7172128772601497e-05,
+      "loss": 1.3372,
+      "step": 815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7164693494881418e-05,
+      "loss": 1.3002,
+      "step": 816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.715725006952127e-05,
+      "loss": 1.3185,
+      "step": 817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7149798504985665e-05,
+      "loss": 1.2068,
+      "step": 818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7142338809748475e-05,
+      "loss": 1.3255,
+      "step": 819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7134870992292808e-05,
+      "loss": 1.3848,
+      "step": 820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712739506111102e-05,
+      "loss": 1.2077,
+      "step": 821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7119911024704684e-05,
+      "loss": 1.346,
+      "step": 822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7112418891584598e-05,
+      "loss": 1.2968,
+      "step": 823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7104918670270763e-05,
+      "loss": 1.3476,
+      "step": 824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7097410369292376e-05,
+      "loss": 1.2234,
+      "step": 825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708989399718783e-05,
+      "loss": 1.2995,
+      "step": 826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708236956250469e-05,
+      "loss": 1.454,
+      "step": 827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.707483707379969e-05,
+      "loss": 1.1749,
+      "step": 828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706729653963872e-05,
+      "loss": 1.3245,
+      "step": 829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7059747968596836e-05,
+      "loss": 1.3509,
+      "step": 830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705219136925821e-05,
+      "loss": 1.3059,
+      "step": 831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.704462675021616e-05,
+      "loss": 1.2868,
+      "step": 832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7037054120073115e-05,
+      "loss": 1.3464,
+      "step": 833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7029473487440617e-05,
+      "loss": 1.3552,
+      "step": 834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7021884860939314e-05,
+      "loss": 1.2263,
+      "step": 835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7014288249198937e-05,
+      "loss": 1.3225,
+      "step": 836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.70066836608583e-05,
+      "loss": 1.3354,
+      "step": 837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699907110456529e-05,
+      "loss": 1.2825,
+      "step": 838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6991450588976853e-05,
+      "loss": 1.3966,
+      "step": 839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698382212275899e-05,
+      "loss": 1.2653,
+      "step": 840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697618571458673e-05,
+      "loss": 1.3246,
+      "step": 841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6968541373144157e-05,
+      "loss": 1.4638,
+      "step": 842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696088910712435e-05,
+      "loss": 1.34,
+      "step": 843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695322892522943e-05,
+      "loss": 1.3196,
+      "step": 844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6945560836170485e-05,
+      "loss": 1.316,
+      "step": 845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6937884848667625e-05,
+      "loss": 1.2138,
+      "step": 846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6930200971449923e-05,
+      "loss": 1.2587,
+      "step": 847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692250921325544e-05,
+      "loss": 1.2763,
+      "step": 848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691480958283118e-05,
+      "loss": 1.3115,
+      "step": 849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6907102088933118e-05,
+      "loss": 1.6282,
+      "step": 850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6899386740326164e-05,
+      "loss": 1.3779,
+      "step": 851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6891663545784146e-05,
+      "loss": 1.4226,
+      "step": 852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6883932514089842e-05,
+      "loss": 1.2005,
+      "step": 853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6876193654034922e-05,
+      "loss": 1.3838,
+      "step": 854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6868446974419962e-05,
+      "loss": 1.1105,
+      "step": 855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686069248405444e-05,
+      "loss": 1.4163,
+      "step": 856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68529301917567e-05,
+      "loss": 1.3145,
+      "step": 857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6845160106353975e-05,
+      "loss": 1.2457,
+      "step": 858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6837382236682347e-05,
+      "loss": 1.2608,
+      "step": 859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682959659158676e-05,
+      "loss": 1.1787,
+      "step": 860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6821803179920996e-05,
+      "loss": 1.259,
+      "step": 861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681400201054766e-05,
+      "loss": 1.3097,
+      "step": 862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.68061930923382e-05,
+      "loss": 1.199,
+      "step": 863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6798376434172862e-05,
+      "loss": 1.3635,
+      "step": 864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6790552044940692e-05,
+      "loss": 1.347,
+      "step": 865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678271993353953e-05,
+      "loss": 1.4459,
+      "step": 866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6774880108876005e-05,
+      "loss": 1.3616,
+      "step": 867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6767032579865503e-05,
+      "loss": 1.096,
+      "step": 868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6759177355432186e-05,
+      "loss": 1.4208,
+      "step": 869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6751314444508967e-05,
+      "loss": 1.2412,
+      "step": 870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674344385603748e-05,
+      "loss": 1.3455,
+      "step": 871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6735565598968114e-05,
+      "loss": 1.2871,
+      "step": 872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672767968225997e-05,
+      "loss": 1.2198,
+      "step": 873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6719786114880846e-05,
+      "loss": 1.2766,
+      "step": 874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6711884905807264e-05,
+      "loss": 1.2012,
+      "step": 875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670397606402442e-05,
+      "loss": 1.4071,
+      "step": 876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.66960595985262e-05,
+      "loss": 1.429,
+      "step": 877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6688135518315144e-05,
+      "loss": 1.3494,
+      "step": 878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668020383240247e-05,
+      "loss": 1.2628,
+      "step": 879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6672264549808033e-05,
+      "loss": 1.3206,
+      "step": 880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6664317679560333e-05,
+      "loss": 1.3291,
+      "step": 881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665636323069649e-05,
+      "loss": 1.4145,
+      "step": 882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6648401212262253e-05,
+      "loss": 1.3661,
+      "step": 883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6640431633311975e-05,
+      "loss": 1.3263,
+      "step": 884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6632454502908605e-05,
+      "loss": 1.3406,
+      "step": 885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662446983012368e-05,
+      "loss": 1.3074,
+      "step": 886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6616477624037317e-05,
+      "loss": 1.3181,
+      "step": 887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6608477893738197e-05,
+      "loss": 1.3829,
+      "step": 888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6600470648323554e-05,
+      "loss": 1.284,
+      "step": 889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6592455896899177e-05,
+      "loss": 1.259,
+      "step": 890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658443364857939e-05,
+      "loss": 1.2656,
+      "step": 891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6576403912487026e-05,
+      "loss": 1.2248,
+      "step": 892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656836669775346e-05,
+      "loss": 1.4117,
+      "step": 893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6560322013518542e-05,
+      "loss": 1.3967,
+      "step": 894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6552269868930646e-05,
+      "loss": 1.3535,
+      "step": 895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6544210273146608e-05,
+      "loss": 1.3346,
+      "step": 896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6536143235331743e-05,
+      "loss": 1.4138,
+      "step": 897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6528068764659838e-05,
+      "loss": 1.3445,
+      "step": 898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651998687031312e-05,
+      "loss": 1.2632,
+      "step": 899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6511897561482263e-05,
+      "loss": 1.1623,
+      "step": 900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6503800847366373e-05,
+      "loss": 1.265,
+      "step": 901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6495696737172982e-05,
+      "loss": 1.2231,
+      "step": 902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6487585240118025e-05,
+      "loss": 1.3099,
+      "step": 903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647946636542584e-05,
+      "loss": 1.3066,
+      "step": 904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6471340122329158e-05,
+      "loss": 1.2451,
+      "step": 905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6463206520069083e-05,
+      "loss": 1.2643,
+      "step": 906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6455065567895098e-05,
+      "loss": 1.3528,
+      "step": 907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6446917275065032e-05,
+      "loss": 1.3333,
+      "step": 908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.643876165084507e-05,
+      "loss": 1.2977,
+      "step": 909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6430598704509723e-05,
+      "loss": 1.3413,
+      "step": 910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642242844534185e-05,
+      "loss": 1.2886,
+      "step": 911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6414250882632607e-05,
+      "loss": 1.236,
+      "step": 912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6406066025681462e-05,
+      "loss": 1.5691,
+      "step": 913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6397873883796183e-05,
+      "loss": 1.2697,
+      "step": 914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6389674466292812e-05,
+      "loss": 1.3026,
+      "step": 915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.638146778249567e-05,
+      "loss": 1.1728,
+      "step": 916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637325384173734e-05,
+      "loss": 1.3402,
+      "step": 917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6365032653358665e-05,
+      "loss": 1.2947,
+      "step": 918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.635680422670872e-05,
+      "loss": 1.3534,
+      "step": 919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6348568571144816e-05,
+      "loss": 1.3345,
+      "step": 920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634032569603248e-05,
+      "loss": 1.3136,
+      "step": 921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633207561074546e-05,
+      "loss": 1.2049,
+      "step": 922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6323818324665693e-05,
+      "loss": 1.2287,
+      "step": 923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.63155538471833e-05,
+      "loss": 1.457,
+      "step": 924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6307282187696602e-05,
+      "loss": 1.236,
+      "step": 925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629900335561206e-05,
+      "loss": 1.2837,
+      "step": 926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6290717360344312e-05,
+      "loss": 1.3094,
+      "step": 927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6282424211316134e-05,
+      "loss": 1.2838,
+      "step": 928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6274123917958438e-05,
+      "loss": 1.2095,
+      "step": 929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626581648971026e-05,
+      "loss": 1.2071,
+      "step": 930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625750193601875e-05,
+      "loss": 1.2638,
+      "step": 931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624918026633916e-05,
+      "loss": 1.2514,
+      "step": 932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6240851490134834e-05,
+      "loss": 1.1429,
+      "step": 933
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6232515616877204e-05,
+      "loss": 1.3468,
+      "step": 934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6224172656045767e-05,
+      "loss": 1.3237,
+      "step": 935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6215822617128075e-05,
+      "loss": 1.2862,
+      "step": 936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.620746550961974e-05,
+      "loss": 1.2863,
+      "step": 937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6199101343024405e-05,
+      "loss": 1.4152,
+      "step": 938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619073012685374e-05,
+      "loss": 1.4615,
+      "step": 939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618235187062744e-05,
+      "loss": 1.2388,
+      "step": 940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6173966583873195e-05,
+      "loss": 1.2309,
+      "step": 941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6165574276126702e-05,
+      "loss": 1.1803,
+      "step": 942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6157174956931627e-05,
+      "loss": 1.3408,
+      "step": 943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6148768635839623e-05,
+      "loss": 1.3339,
+      "step": 944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6140355322410304e-05,
+      "loss": 1.3751,
+      "step": 945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6131935026211224e-05,
+      "loss": 1.3877,
+      "step": 946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.612350775681789e-05,
+      "loss": 1.207,
+      "step": 947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.611507352381373e-05,
+      "loss": 1.4494,
+      "step": 948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6106632336790105e-05,
+      "loss": 1.3727,
+      "step": 949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609818420534627e-05,
+      "loss": 1.3039,
+      "step": 950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608972913908938e-05,
+      "loss": 1.1756,
+      "step": 951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608126714763448e-05,
+      "loss": 1.2914,
+      "step": 952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6072798240604487e-05,
+      "loss": 1.3273,
+      "step": 953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6064322427630186e-05,
+      "loss": 1.322,
+      "step": 954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6055839718350207e-05,
+      "loss": 1.1453,
+      "step": 955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6047350122411038e-05,
+      "loss": 1.3025,
+      "step": 956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6038853649466983e-05,
+      "loss": 1.1693,
+      "step": 957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6030350309180173e-05,
+      "loss": 1.2767,
+      "step": 958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6021840111220548e-05,
+      "loss": 1.4647,
+      "step": 959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6013323065265848e-05,
+      "loss": 1.2958,
+      "step": 960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60047991810016e-05,
+      "loss": 1.3274,
+      "step": 961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.59962684681211e-05,
+      "loss": 1.325,
+      "step": 962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5987730936325424e-05,
+      "loss": 1.3704,
+      "step": 963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.597918659532339e-05,
+      "loss": 1.3307,
+      "step": 964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5970635454831565e-05,
+      "loss": 1.2188,
+      "step": 965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5962077524574248e-05,
+      "loss": 1.2714,
+      "step": 966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.595351281428345e-05,
+      "loss": 1.2796,
+      "step": 967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5944941333698912e-05,
+      "loss": 1.2497,
+      "step": 968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5936363092568058e-05,
+      "loss": 1.244,
+      "step": 969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5927778100646007e-05,
+      "loss": 1.212,
+      "step": 970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5919186367695546e-05,
+      "loss": 1.3979,
+      "step": 971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5910587903487142e-05,
+      "loss": 1.3326,
+      "step": 972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5901982717798903e-05,
+      "loss": 1.3511,
+      "step": 973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5893370820416594e-05,
+      "loss": 1.3456,
+      "step": 974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5884752221133597e-05,
+      "loss": 1.3462,
+      "step": 975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587612692975093e-05,
+      "loss": 1.2286,
+      "step": 976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5867494956077215e-05,
+      "loss": 1.3136,
+      "step": 977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585885630992867e-05,
+      "loss": 1.3715,
+      "step": 978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.58502110011291e-05,
+      "loss": 1.2604,
+      "step": 979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5841559039509898e-05,
+      "loss": 1.1808,
+      "step": 980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5832900434910014e-05,
+      "loss": 1.2897,
+      "step": 981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5824235197175954e-05,
+      "loss": 1.2339,
+      "step": 982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5815563336161756e-05,
+      "loss": 1.2399,
+      "step": 983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580688486172902e-05,
+      "loss": 1.3925,
+      "step": 984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5798199783746828e-05,
+      "loss": 1.3116,
+      "step": 985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5789508112091804e-05,
+      "loss": 1.4549,
+      "step": 986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578080985664805e-05,
+      "loss": 1.2233,
+      "step": 987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5772105027307168e-05,
+      "loss": 1.1387,
+      "step": 988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5763393633968223e-05,
+      "loss": 1.2636,
+      "step": 989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5754675686537756e-05,
+      "loss": 1.155,
+      "step": 990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5745951194929756e-05,
+      "loss": 1.098,
+      "step": 991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5737220169065656e-05,
+      "loss": 1.1932,
+      "step": 992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5728482618874314e-05,
+      "loss": 1.1976,
+      "step": 993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5719738554292017e-05,
+      "loss": 1.2005,
+      "step": 994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.571098798526245e-05,
+      "loss": 1.5259,
+      "step": 995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5702230921736705e-05,
+      "loss": 1.2184,
+      "step": 996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569346737367325e-05,
+      "loss": 1.3392,
+      "step": 997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5684697351037935e-05,
+      "loss": 1.3473,
+      "step": 998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5675920863803967e-05,
+      "loss": 1.2805,
+      "step": 999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.566713792195191e-05,
+      "loss": 1.1129,
+      "step": 1000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565834853546966e-05,
+      "loss": 1.2706,
+      "step": 1001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5649552714352448e-05,
+      "loss": 1.323,
+      "step": 1002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5640750468602822e-05,
+      "loss": 1.3073,
+      "step": 1003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563194180823064e-05,
+      "loss": 1.2237,
+      "step": 1004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562312674325304e-05,
+      "loss": 1.3814,
+      "step": 1005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5614305283694462e-05,
+      "loss": 1.3482,
+      "step": 1006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56054774395866e-05,
+      "loss": 1.2636,
+      "step": 1007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559664322096843e-05,
+      "loss": 1.2905,
+      "step": 1008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5587802637886156e-05,
+      "loss": 1.2747,
+      "step": 1009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.557895570039323e-05,
+      "loss": 1.2749,
+      "step": 1010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5570102418550327e-05,
+      "loss": 1.2037,
+      "step": 1011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5561242802425337e-05,
+      "loss": 1.3082,
+      "step": 1012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5552376862093357e-05,
+      "loss": 1.1666,
+      "step": 1013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554350460763667e-05,
+      "loss": 1.1587,
+      "step": 1014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5534626049144753e-05,
+      "loss": 1.3303,
+      "step": 1015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552574119671423e-05,
+      "loss": 1.254,
+      "step": 1016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.55168500604489e-05,
+      "loss": 1.2916,
+      "step": 1017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5507952650459697e-05,
+      "loss": 1.2834,
+      "step": 1018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.54990489768647e-05,
+      "loss": 1.2852,
+      "step": 1019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5490139049789097e-05,
+      "loss": 1.37,
+      "step": 1020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5481222879365205e-05,
+      "loss": 1.2925,
+      "step": 1021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5472300475732427e-05,
+      "loss": 1.3541,
+      "step": 1022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546337184903726e-05,
+      "loss": 1.3664,
+      "step": 1023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5454437009433277e-05,
+      "loss": 1.2974,
+      "step": 1024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5445495967081114e-05,
+      "loss": 1.3451,
+      "step": 1025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543654873214846e-05,
+      "loss": 1.4349,
+      "step": 1026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542759531481006e-05,
+      "loss": 1.2617,
+      "step": 1027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5418635725247666e-05,
+      "loss": 1.2663,
+      "step": 1028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.540966997365007e-05,
+      "loss": 1.2458,
+      "step": 1029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.540069807021306e-05,
+      "loss": 1.3168,
+      "step": 1030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.539172002513942e-05,
+      "loss": 1.286,
+      "step": 1031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.538273584863893e-05,
+      "loss": 1.233,
+      "step": 1032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.537374555092832e-05,
+      "loss": 1.3463,
+      "step": 1033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5364749142231302e-05,
+      "loss": 1.3144,
+      "step": 1034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5355746632778533e-05,
+      "loss": 1.2864,
+      "step": 1035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5346738032807603e-05,
+      "loss": 1.352,
+      "step": 1036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5337723352563028e-05,
+      "loss": 1.2151,
+      "step": 1037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5328702602296238e-05,
+      "loss": 1.2613,
+      "step": 1038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.531967579226557e-05,
+      "loss": 1.1542,
+      "step": 1039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5310642932736253e-05,
+      "loss": 1.2407,
+      "step": 1040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5301604033980387e-05,
+      "loss": 1.2442,
+      "step": 1041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529255910627695e-05,
+      "loss": 1.203,
+      "step": 1042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5283508159911776e-05,
+      "loss": 1.3566,
+      "step": 1043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5274451205177526e-05,
+      "loss": 1.2837,
+      "step": 1044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5265388252373712e-05,
+      "loss": 1.3364,
+      "step": 1045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.525631931180667e-05,
+      "loss": 1.2052,
+      "step": 1046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.524724439378953e-05,
+      "loss": 1.3054,
+      "step": 1047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5238163508642219e-05,
+      "loss": 1.224,
+      "step": 1048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5229076666691471e-05,
+      "loss": 1.4566,
+      "step": 1049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5219983878270767e-05,
+      "loss": 1.3532,
+      "step": 1050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5210885153720374e-05,
+      "loss": 1.26,
+      "step": 1051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.520178050338729e-05,
+      "loss": 1.1311,
+      "step": 1052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.519266993762527e-05,
+      "loss": 1.3461,
+      "step": 1053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5183553466794778e-05,
+      "loss": 1.2327,
+      "step": 1054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5174431101263004e-05,
+      "loss": 1.4256,
+      "step": 1055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5165302851403843e-05,
+      "loss": 1.3932,
+      "step": 1056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5156168727597873e-05,
+      "loss": 1.288,
+      "step": 1057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5147028740232361e-05,
+      "loss": 1.3192,
+      "step": 1058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5137882899701234e-05,
+      "loss": 1.27,
+      "step": 1059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5128731216405084e-05,
+      "loss": 1.2837,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5119573700751133e-05,
+      "loss": 1.4091,
+      "step": 1061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5110410363153256e-05,
+      "loss": 1.2189,
+      "step": 1062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5101241214031928e-05,
+      "loss": 1.3311,
+      "step": 1063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5092066263814245e-05,
+      "loss": 1.2795,
+      "step": 1064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5082885522933901e-05,
+      "loss": 1.4888,
+      "step": 1065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5073699001831168e-05,
+      "loss": 1.1529,
+      "step": 1066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.506450671095289e-05,
+      "loss": 1.2273,
+      "step": 1067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5055308660752486e-05,
+      "loss": 1.4096,
+      "step": 1068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5046104861689911e-05,
+      "loss": 1.2175,
+      "step": 1069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5036895324231662e-05,
+      "loss": 1.2721,
+      "step": 1070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5027680058850757e-05,
+      "loss": 1.2757,
+      "step": 1071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.501845907602674e-05,
+      "loss": 1.306,
+      "step": 1072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5009232386245644e-05,
+      "loss": 1.3438,
+      "step": 1073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.2436,
+      "step": 1074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4990761927788815e-05,
+      "loss": 1.333,
+      "step": 1075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4981518180117558e-05,
+      "loss": 1.402,
+      "step": 1076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.497226876749815e-05,
+      "loss": 1.3782,
+      "step": 1077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4963013700448969e-05,
+      "loss": 1.3064,
+      "step": 1078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4953752989494815e-05,
+      "loss": 1.2067,
+      "step": 1079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4944486645166892e-05,
+      "loss": 1.3255,
+      "step": 1080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4935214678002834e-05,
+      "loss": 1.3169,
+      "step": 1081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4925937098546652e-05,
+      "loss": 1.2642,
+      "step": 1082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4916653917348751e-05,
+      "loss": 1.1457,
+      "step": 1083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4907365144965896e-05,
+      "loss": 1.297,
+      "step": 1084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4898070791961215e-05,
+      "loss": 1.2597,
+      "step": 1085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4888770868904185e-05,
+      "loss": 1.3395,
+      "step": 1086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4879465386370613e-05,
+      "loss": 1.2254,
+      "step": 1087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.487015435494263e-05,
+      "loss": 1.1784,
+      "step": 1088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4860837785208678e-05,
+      "loss": 1.2387,
+      "step": 1089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4851515687763492e-05,
+      "loss": 1.2022,
+      "step": 1090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.48421880732081e-05,
+      "loss": 1.3362,
+      "step": 1091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.48328549521498e-05,
+      "loss": 1.2714,
+      "step": 1092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4823516335202152e-05,
+      "loss": 1.2764,
+      "step": 1093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4814172232984968e-05,
+      "loss": 1.1313,
+      "step": 1094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4804822656124302e-05,
+      "loss": 1.3175,
+      "step": 1095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.479546761525242e-05,
+      "loss": 1.2485,
+      "step": 1096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4786107121007813e-05,
+      "loss": 1.3056,
+      "step": 1097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4776741184035171e-05,
+      "loss": 1.1867,
+      "step": 1098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.476736981498537e-05,
+      "loss": 1.266,
+      "step": 1099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.475799302451547e-05,
+      "loss": 1.227,
+      "step": 1100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4748610823288693e-05,
+      "loss": 1.3661,
+      "step": 1101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4739223221974415e-05,
+      "loss": 1.3033,
+      "step": 1102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4729830231248143e-05,
+      "loss": 1.2722,
+      "step": 1103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4720431861791531e-05,
+      "loss": 1.2511,
+      "step": 1104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4711028124292332e-05,
+      "loss": 1.2971,
+      "step": 1105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4701619029444419e-05,
+      "loss": 1.348,
+      "step": 1106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4692204587947745e-05,
+      "loss": 1.3339,
+      "step": 1107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.468278481050835e-05,
+      "loss": 1.1522,
+      "step": 1108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4673359707838338e-05,
+      "loss": 1.1803,
+      "step": 1109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4663929290655872e-05,
+      "loss": 1.1292,
+      "step": 1110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4654493569685157e-05,
+      "loss": 1.3618,
+      "step": 1111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.464505255565643e-05,
+      "loss": 1.369,
+      "step": 1112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4635606259305948e-05,
+      "loss": 1.2768,
+      "step": 1113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4626154691375974e-05,
+      "loss": 1.2776,
+      "step": 1114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4616697862614766e-05,
+      "loss": 1.2922,
+      "step": 1115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4607235783776563e-05,
+      "loss": 1.3472,
+      "step": 1116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4597768465621575e-05,
+      "loss": 1.2253,
+      "step": 1117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4588295918915978e-05,
+      "loss": 1.303,
+      "step": 1118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4578818154431878e-05,
+      "loss": 1.2062,
+      "step": 1119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4569335182947325e-05,
+      "loss": 1.3157,
+      "step": 1120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4559847015246294e-05,
+      "loss": 1.3139,
+      "step": 1121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4550353662118659e-05,
+      "loss": 1.423,
+      "step": 1122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4540855134360198e-05,
+      "loss": 1.2684,
+      "step": 1123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.453135144277257e-05,
+      "loss": 1.4146,
+      "step": 1124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4521842598163309e-05,
+      "loss": 1.3044,
+      "step": 1125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4512328611345808e-05,
+      "loss": 1.2642,
+      "step": 1126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4502809493139305e-05,
+      "loss": 1.3398,
+      "step": 1127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4493285254368878e-05,
+      "loss": 1.1626,
+      "step": 1128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4483755905865423e-05,
+      "loss": 1.2641,
+      "step": 1129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4474221458465651e-05,
+      "loss": 1.3286,
+      "step": 1130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4464681923012072e-05,
+      "loss": 1.4002,
+      "step": 1131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4455137310352978e-05,
+      "loss": 1.1133,
+      "step": 1132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4445587631342433e-05,
+      "loss": 1.3146,
+      "step": 1133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4436032896840271e-05,
+      "loss": 1.2476,
+      "step": 1134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4426473117712071e-05,
+      "loss": 1.3524,
+      "step": 1135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4416908304829142e-05,
+      "loss": 1.2622,
+      "step": 1136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4407338469068532e-05,
+      "loss": 1.2244,
+      "step": 1137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4397763621312985e-05,
+      "loss": 1.3472,
+      "step": 1138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4388183772450957e-05,
+      "loss": 1.1446,
+      "step": 1139
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.437859893337658e-05,
+      "loss": 1.2744,
+      "step": 1140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4369009114989671e-05,
+      "loss": 1.2349,
+      "step": 1141
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4359414328195704e-05,
+      "loss": 1.2209,
+      "step": 1142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4349814583905807e-05,
+      "loss": 1.3382,
+      "step": 1143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434020989303674e-05,
+      "loss": 1.2584,
+      "step": 1144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4330600266510895e-05,
+      "loss": 1.3664,
+      "step": 1145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4320985715256269e-05,
+      "loss": 1.162,
+      "step": 1146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4311366250206464e-05,
+      "loss": 1.3295,
+      "step": 1147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4301741882300673e-05,
+      "loss": 1.3086,
+      "step": 1148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4292112622483653e-05,
+      "loss": 1.278,
+      "step": 1149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4282478481705742e-05,
+      "loss": 1.2891,
+      "step": 1150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4272839470922812e-05,
+      "loss": 1.2409,
+      "step": 1151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.426319560109628e-05,
+      "loss": 1.2973,
+      "step": 1152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4253546883193093e-05,
+      "loss": 1.2486,
+      "step": 1153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42438933281857e-05,
+      "loss": 1.2693,
+      "step": 1154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4234234947052063e-05,
+      "loss": 1.2249,
+      "step": 1155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4224571750775625e-05,
+      "loss": 1.3395,
+      "step": 1156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4214903750345298e-05,
+      "loss": 1.449,
+      "step": 1157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4205230956755479e-05,
+      "loss": 1.1161,
+      "step": 1158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.419555338100599e-05,
+      "loss": 1.3085,
+      "step": 1159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4185871034102117e-05,
+      "loss": 1.351,
+      "step": 1160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4176183927054545e-05,
+      "loss": 1.3234,
+      "step": 1161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4166492070879389e-05,
+      "loss": 1.2037,
+      "step": 1162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4156795476598166e-05,
+      "loss": 1.3678,
+      "step": 1163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4147094155237772e-05,
+      "loss": 1.3293,
+      "step": 1164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4137388117830486e-05,
+      "loss": 1.2505,
+      "step": 1165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4127677375413943e-05,
+      "loss": 1.1536,
+      "step": 1166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4117961939031135e-05,
+      "loss": 1.2298,
+      "step": 1167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4108241819730384e-05,
+      "loss": 1.2324,
+      "step": 1168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409851702856535e-05,
+      "loss": 1.3091,
+      "step": 1169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4088787576594996e-05,
+      "loss": 1.2983,
+      "step": 1170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4079053474883584e-05,
+      "loss": 1.1055,
+      "step": 1171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4069314734500674e-05,
+      "loss": 1.2983,
+      "step": 1172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.405957136652109e-05,
+      "loss": 1.4468,
+      "step": 1173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4049823382024922e-05,
+      "loss": 1.3416,
+      "step": 1174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4040070792097515e-05,
+      "loss": 1.3257,
+      "step": 1175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4030313607829444e-05,
+      "loss": 1.3353,
+      "step": 1176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4020551840316515e-05,
+      "loss": 1.1875,
+      "step": 1177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4010785500659735e-05,
+      "loss": 1.3478,
+      "step": 1178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4001014599965327e-05,
+      "loss": 1.3157,
+      "step": 1179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3991239149344686e-05,
+      "loss": 1.1173,
+      "step": 1180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3981459159914392e-05,
+      "loss": 1.2161,
+      "step": 1181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3971674642796178e-05,
+      "loss": 1.2952,
+      "step": 1182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.396188560911693e-05,
+      "loss": 1.2907,
+      "step": 1183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3952092070008669e-05,
+      "loss": 1.4054,
+      "step": 1184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3942294036608544e-05,
+      "loss": 1.2853,
+      "step": 1185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3932491520058807e-05,
+      "loss": 1.2316,
+      "step": 1186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.392268453150681e-05,
+      "loss": 1.3066,
+      "step": 1187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3912873082105002e-05,
+      "loss": 1.2377,
+      "step": 1188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3903057183010887e-05,
+      "loss": 1.3614,
+      "step": 1189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3893236845387041e-05,
+      "loss": 1.3397,
+      "step": 1190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3883412080401083e-05,
+      "loss": 1.1668,
+      "step": 1191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3873582899225672e-05,
+      "loss": 1.1924,
+      "step": 1192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3863749313038484e-05,
+      "loss": 1.3731,
+      "step": 1193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3853911333022199e-05,
+      "loss": 1.1934,
+      "step": 1194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3844068970364506e-05,
+      "loss": 1.1635,
+      "step": 1195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.383422223625807e-05,
+      "loss": 1.3358,
+      "step": 1196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3824371141900533e-05,
+      "loss": 1.2897,
+      "step": 1197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3814515698494484e-05,
+      "loss": 1.3208,
+      "step": 1198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3804655917247471e-05,
+      "loss": 1.2866,
+      "step": 1199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3794791809371971e-05,
+      "loss": 1.0992,
+      "step": 1200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3784923386085373e-05,
+      "loss": 1.1884,
+      "step": 1201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3775050658609988e-05,
+      "loss": 1.2635,
+      "step": 1202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3765173638173004e-05,
+      "loss": 1.3466,
+      "step": 1203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3755292336006512e-05,
+      "loss": 1.2894,
+      "step": 1204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3745406763347446e-05,
+      "loss": 1.237,
+      "step": 1205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3735516931437625e-05,
+      "loss": 1.1691,
+      "step": 1206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3725622851523692e-05,
+      "loss": 1.367,
+      "step": 1207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3715724534857127e-05,
+      "loss": 1.1355,
+      "step": 1208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3705821992694228e-05,
+      "loss": 1.4048,
+      "step": 1209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3695915236296095e-05,
+      "loss": 1.225,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3686004276928627e-05,
+      "loss": 1.2157,
+      "step": 1211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3676089125862495e-05,
+      "loss": 1.2612,
+      "step": 1212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3666169794373141e-05,
+      "loss": 1.2672,
+      "step": 1213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3656246293740767e-05,
+      "loss": 1.1888,
+      "step": 1214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.36463186352503e-05,
+      "loss": 1.3704,
+      "step": 1215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3636386830191409e-05,
+      "loss": 1.2579,
+      "step": 1216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3626450889858476e-05,
+      "loss": 1.3641,
+      "step": 1217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3616510825550582e-05,
+      "loss": 1.2377,
+      "step": 1218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.36065666485715e-05,
+      "loss": 1.3744,
+      "step": 1219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3596618370229679e-05,
+      "loss": 1.3327,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3586666001838231e-05,
+      "loss": 1.2221,
+      "step": 1221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3576709554714923e-05,
+      "loss": 1.27,
+      "step": 1222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.356674904018216e-05,
+      "loss": 1.3184,
+      "step": 1223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3556784469566968e-05,
+      "loss": 1.3058,
+      "step": 1224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3546815854200987e-05,
+      "loss": 1.297,
+      "step": 1225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.353684320542046e-05,
+      "loss": 1.3612,
+      "step": 1226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3526866534566213e-05,
+      "loss": 1.2144,
+      "step": 1227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3516885852983645e-05,
+      "loss": 1.3782,
+      "step": 1228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3506901172022724e-05,
+      "loss": 1.3937,
+      "step": 1229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3496912503037959e-05,
+      "loss": 1.3548,
+      "step": 1230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3486919857388392e-05,
+      "loss": 1.3579,
+      "step": 1231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.347692324643759e-05,
+      "loss": 1.2221,
+      "step": 1232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3466922681553636e-05,
+      "loss": 1.2839,
+      "step": 1233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3456918174109102e-05,
+      "loss": 1.2777,
+      "step": 1234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.344690973548104e-05,
+      "loss": 1.2537,
+      "step": 1235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3436897377050982e-05,
+      "loss": 1.3699,
+      "step": 1236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.342688111020491e-05,
+      "loss": 1.248,
+      "step": 1237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3416860946333256e-05,
+      "loss": 1.3886,
+      "step": 1238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3406836896830884e-05,
+      "loss": 1.1917,
+      "step": 1239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3396808973097066e-05,
+      "loss": 1.2582,
+      "step": 1240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3386777186535495e-05,
+      "loss": 1.2666,
+      "step": 1241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3376741548554248e-05,
+      "loss": 1.1695,
+      "step": 1242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3366702070565775e-05,
+      "loss": 1.2851,
+      "step": 1243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3356658763986919e-05,
+      "loss": 1.1393,
+      "step": 1244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.334661164023884e-05,
+      "loss": 1.3551,
+      "step": 1245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3336560710747073e-05,
+      "loss": 1.2921,
+      "step": 1246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3326505986941456e-05,
+      "loss": 1.2649,
+      "step": 1247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3316447480256153e-05,
+      "loss": 1.4216,
+      "step": 1248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3306385202129633e-05,
+      "loss": 1.2133,
+      "step": 1249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3296319164004646e-05,
+      "loss": 1.1427,
+      "step": 1250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3286249377328222e-05,
+      "loss": 1.0525,
+      "step": 1251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.327617585355165e-05,
+      "loss": 1.2209,
+      "step": 1252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3266098604130477e-05,
+      "loss": 1.2439,
+      "step": 1253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.325601764052448e-05,
+      "loss": 1.3988,
+      "step": 1254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.324593297419766e-05,
+      "loss": 1.3372,
+      "step": 1255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.323584461661823e-05,
+      "loss": 1.1898,
+      "step": 1256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3225752579258599e-05,
+      "loss": 1.2429,
+      "step": 1257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3215656873595366e-05,
+      "loss": 1.2746,
+      "step": 1258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3205557511109298e-05,
+      "loss": 1.1461,
+      "step": 1259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.319545450328532e-05,
+      "loss": 1.1718,
+      "step": 1260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3185347861612501e-05,
+      "loss": 1.2531,
+      "step": 1261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3175237597584045e-05,
+      "loss": 1.2999,
+      "step": 1262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3165123722697276e-05,
+      "loss": 1.2824,
+      "step": 1263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3155006248453615e-05,
+      "loss": 1.16,
+      "step": 1264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3144885186358596e-05,
+      "loss": 1.3179,
+      "step": 1265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3134760547921814e-05,
+      "loss": 1.234,
+      "step": 1266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3124632344656943e-05,
+      "loss": 1.3048,
+      "step": 1267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3114500588081698e-05,
+      "loss": 1.2577,
+      "step": 1268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3104365289717849e-05,
+      "loss": 1.2236,
+      "step": 1269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3094226461091186e-05,
+      "loss": 1.2877,
+      "step": 1270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3084084113731514e-05,
+      "loss": 1.3343,
+      "step": 1271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3073938259172645e-05,
+      "loss": 1.242,
+      "step": 1272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063788908952368e-05,
+      "loss": 1.3352,
+      "step": 1273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3053636074612457e-05,
+      "loss": 1.3005,
+      "step": 1274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3043479767698646e-05,
+      "loss": 1.1926,
+      "step": 1275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3033319999760619e-05,
+      "loss": 1.2242,
+      "step": 1276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3023156782351989e-05,
+      "loss": 1.2421,
+      "step": 1277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30129901270303e-05,
+      "loss": 1.158,
+      "step": 1278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3002820045357005e-05,
+      "loss": 1.2962,
+      "step": 1279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2992646548897442e-05,
+      "loss": 1.1178,
+      "step": 1280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2982469649220854e-05,
+      "loss": 1.3417,
+      "step": 1281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2972289357900332e-05,
+      "loss": 1.2901,
+      "step": 1282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2962105686512834e-05,
+      "loss": 1.2904,
+      "step": 1283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2951918646639163e-05,
+      "loss": 1.3182,
+      "step": 1284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.294172824986395e-05,
+      "loss": 1.226,
+      "step": 1285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.293153450777564e-05,
+      "loss": 1.2812,
+      "step": 1286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2921337431966494e-05,
+      "loss": 1.4035,
+      "step": 1287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2911137034032553e-05,
+      "loss": 1.3059,
+      "step": 1288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2900933325573636e-05,
+      "loss": 1.3659,
+      "step": 1289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2890726318193332e-05,
+      "loss": 1.2342,
+      "step": 1290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2880516023498981e-05,
+      "loss": 1.4161,
+      "step": 1291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2870302453101657e-05,
+      "loss": 1.2108,
+      "step": 1292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2860085618616169e-05,
+      "loss": 1.3012,
+      "step": 1293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.284986553166102e-05,
+      "loss": 1.2797,
+      "step": 1294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2839642203858433e-05,
+      "loss": 1.2384,
+      "step": 1295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2829415646834295e-05,
+      "loss": 1.3128,
+      "step": 1296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2819185872218184e-05,
+      "loss": 1.2666,
+      "step": 1297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2808952891643327e-05,
+      "loss": 1.3178,
+      "step": 1298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2798716716746602e-05,
+      "loss": 1.2278,
+      "step": 1299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2788477359168512e-05,
+      "loss": 1.3281,
+      "step": 1300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2778234830553183e-05,
+      "loss": 1.1671,
+      "step": 1301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.276798914254835e-05,
+      "loss": 1.3176,
+      "step": 1302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2757740306805345e-05,
+      "loss": 1.2479,
+      "step": 1303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2747488334979064e-05,
+      "loss": 1.176,
+      "step": 1304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2737233238727982e-05,
+      "loss": 1.3775,
+      "step": 1305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2726975029714124e-05,
+      "loss": 1.3324,
+      "step": 1306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2716713719603053e-05,
+      "loss": 1.23,
+      "step": 1307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2706449320063866e-05,
+      "loss": 1.21,
+      "step": 1308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2696181842769161e-05,
+      "loss": 1.3184,
+      "step": 1309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2685911299395046e-05,
+      "loss": 1.1431,
+      "step": 1310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2675637701621111e-05,
+      "loss": 1.2324,
+      "step": 1311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2665361061130416e-05,
+      "loss": 1.4506,
+      "step": 1312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2655081389609496e-05,
+      "loss": 1.2632,
+      "step": 1313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2644798698748314e-05,
+      "loss": 1.2199,
+      "step": 1314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2634513000240277e-05,
+      "loss": 1.1961,
+      "step": 1315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2624224305782215e-05,
+      "loss": 1.2649,
+      "step": 1316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2613932627074356e-05,
+      "loss": 1.3104,
+      "step": 1317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2603637975820324e-05,
+      "loss": 1.0901,
+      "step": 1318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2593340363727131e-05,
+      "loss": 1.3713,
+      "step": 1319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2583039802505148e-05,
+      "loss": 1.2994,
+      "step": 1320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2572736303868102e-05,
+      "loss": 1.307,
+      "step": 1321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.256242987953306e-05,
+      "loss": 1.2154,
+      "step": 1322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2552120541220419e-05,
+      "loss": 1.1019,
+      "step": 1323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2541808300653883e-05,
+      "loss": 1.2299,
+      "step": 1324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253149316956047e-05,
+      "loss": 1.1303,
+      "step": 1325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2521175159670464e-05,
+      "loss": 1.3876,
+      "step": 1326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2510854282717444e-05,
+      "loss": 1.2807,
+      "step": 1327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2500530550438232e-05,
+      "loss": 1.1717,
+      "step": 1328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2490203974572915e-05,
+      "loss": 1.3065,
+      "step": 1329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2479874566864799e-05,
+      "loss": 1.2305,
+      "step": 1330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2469542339060414e-05,
+      "loss": 1.0857,
+      "step": 1331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2459207302909499e-05,
+      "loss": 1.1372,
+      "step": 1332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2448869470164988e-05,
+      "loss": 1.4061,
+      "step": 1333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.243852885258299e-05,
+      "loss": 1.3289,
+      "step": 1334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2428185461922783e-05,
+      "loss": 1.3146,
+      "step": 1335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2417839309946806e-05,
+      "loss": 1.3308,
+      "step": 1336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2407490408420629e-05,
+      "loss": 1.1638,
+      "step": 1337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2397138769112949e-05,
+      "loss": 1.1167,
+      "step": 1338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2386784403795577e-05,
+      "loss": 1.1957,
+      "step": 1339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2376427324243432e-05,
+      "loss": 1.2913,
+      "step": 1340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2366067542234512e-05,
+      "loss": 1.3546,
+      "step": 1341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2355705069549885e-05,
+      "loss": 1.2994,
+      "step": 1342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2345339917973688e-05,
+      "loss": 1.4036,
+      "step": 1343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2334972099293102e-05,
+      "loss": 1.2087,
+      "step": 1344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2324601625298331e-05,
+      "loss": 1.4056,
+      "step": 1345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2314228507782614e-05,
+      "loss": 1.2082,
+      "step": 1346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2303852758542187e-05,
+      "loss": 1.4642,
+      "step": 1347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2293474389376281e-05,
+      "loss": 1.2091,
+      "step": 1348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2283093412087099e-05,
+      "loss": 1.2325,
+      "step": 1349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2272709838479824e-05,
+      "loss": 1.2466,
+      "step": 1350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2262323680362587e-05,
+      "loss": 1.2338,
+      "step": 1351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2251934949546446e-05,
+      "loss": 1.2539,
+      "step": 1352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2241543657845403e-05,
+      "loss": 1.1242,
+      "step": 1353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2231149817076358e-05,
+      "loss": 1.2605,
+      "step": 1354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2220753439059116e-05,
+      "loss": 1.3588,
+      "step": 1355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2210354535616366e-05,
+      "loss": 1.2776,
+      "step": 1356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2199953118573674e-05,
+      "loss": 1.241,
+      "step": 1357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2189549199759454e-05,
+      "loss": 1.1881,
+      "step": 1358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2179142791004974e-05,
+      "loss": 1.2892,
+      "step": 1359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2168733904144328e-05,
+      "loss": 1.1443,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2158322551014431e-05,
+      "loss": 1.1217,
+      "step": 1361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2147908743455005e-05,
+      "loss": 1.3153,
+      "step": 1362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2137492493308558e-05,
+      "loss": 1.0986,
+      "step": 1363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2127073812420376e-05,
+      "loss": 1.0793,
+      "step": 1364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2116652712638513e-05,
+      "loss": 1.326,
+      "step": 1365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.210622920581377e-05,
+      "loss": 1.3537,
+      "step": 1366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2095803303799693e-05,
+      "loss": 1.2164,
+      "step": 1367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2085375018452541e-05,
+      "loss": 1.0878,
+      "step": 1368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2074944361631288e-05,
+      "loss": 1.2595,
+      "step": 1369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2064511345197606e-05,
+      "loss": 1.2288,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.205407598101585e-05,
+      "loss": 1.2282,
+      "step": 1371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2043638280953043e-05,
+      "loss": 1.3438,
+      "step": 1372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2033198256878867e-05,
+      "loss": 1.271,
+      "step": 1373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2022755920665648e-05,
+      "loss": 1.2748,
+      "step": 1374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.201231128418833e-05,
+      "loss": 1.2505,
+      "step": 1375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.200186435932449e-05,
+      "loss": 1.2923,
+      "step": 1376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1991415157954293e-05,
+      "loss": 1.1689,
+      "step": 1377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1980963691960508e-05,
+      "loss": 1.2749,
+      "step": 1378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1970509973228459e-05,
+      "loss": 1.1473,
+      "step": 1379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1960054013646048e-05,
+      "loss": 1.2292,
+      "step": 1380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1949595825103721e-05,
+      "loss": 1.2904,
+      "step": 1381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1939135419494456e-05,
+      "loss": 1.2193,
+      "step": 1382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.192867280871376e-05,
+      "loss": 1.0881,
+      "step": 1383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1918208004659632e-05,
+      "loss": 1.3105,
+      "step": 1384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1907741019232584e-05,
+      "loss": 1.2666,
+      "step": 1385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1897271864335594e-05,
+      "loss": 1.2536,
+      "step": 1386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1886800551874111e-05,
+      "loss": 1.3282,
+      "step": 1387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1876327093756048e-05,
+      "loss": 1.1443,
+      "step": 1388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1865851501891741e-05,
+      "loss": 1.1805,
+      "step": 1389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1855373788193969e-05,
+      "loss": 1.1864,
+      "step": 1390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1844893964577906e-05,
+      "loss": 1.2669,
+      "step": 1391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1834412042961143e-05,
+      "loss": 1.3911,
+      "step": 1392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1823928035263646e-05,
+      "loss": 1.3024,
+      "step": 1393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1813441953407754e-05,
+      "loss": 1.1906,
+      "step": 1394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1802953809318173e-05,
+      "loss": 1.2009,
+      "step": 1395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1792463614921945e-05,
+      "loss": 1.1585,
+      "step": 1396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1781971382148446e-05,
+      "loss": 1.3063,
+      "step": 1397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.177147712292937e-05,
+      "loss": 1.3428,
+      "step": 1398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1760980849198717e-05,
+      "loss": 1.2162,
+      "step": 1399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1750482572892781e-05,
+      "loss": 1.2148,
+      "step": 1400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1739982305950124e-05,
+      "loss": 1.292,
+      "step": 1401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.172948006031158e-05,
+      "loss": 1.2038,
+      "step": 1402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1718975847920229e-05,
+      "loss": 1.2464,
+      "step": 1403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.170846968072139e-05,
+      "loss": 1.1437,
+      "step": 1404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1697961570662606e-05,
+      "loss": 1.2991,
+      "step": 1405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1687451529693626e-05,
+      "loss": 1.2902,
+      "step": 1406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.167693956976639e-05,
+      "loss": 1.2128,
+      "step": 1407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1666425702835033e-05,
+      "loss": 1.1326,
+      "step": 1408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1655909940855852e-05,
+      "loss": 1.1265,
+      "step": 1409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1645392295787299e-05,
+      "loss": 1.3404,
+      "step": 1410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1634872779589966e-05,
+      "loss": 1.2424,
+      "step": 1411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1624351404226573e-05,
+      "loss": 1.1055,
+      "step": 1412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1613828181661957e-05,
+      "loss": 1.3506,
+      "step": 1413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1603303123863054e-05,
+      "loss": 1.2615,
+      "step": 1414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1592776242798887e-05,
+      "loss": 1.2522,
+      "step": 1415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1582247550440555e-05,
+      "loss": 1.2939,
+      "step": 1416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1571717058761211e-05,
+      "loss": 1.2789,
+      "step": 1417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1561184779736061e-05,
+      "loss": 1.2443,
+      "step": 1418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1550650725342337e-05,
+      "loss": 1.3018,
+      "step": 1419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1540114907559294e-05,
+      "loss": 1.3083,
+      "step": 1420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1529577338368191e-05,
+      "loss": 1.0566,
+      "step": 1421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1519038029752281e-05,
+      "loss": 1.2826,
+      "step": 1422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1508496993696791e-05,
+      "loss": 1.1012,
+      "step": 1423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1497954242188913e-05,
+      "loss": 1.1776,
+      "step": 1424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.14874097872178e-05,
+      "loss": 1.462,
+      "step": 1425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1476863640774523e-05,
+      "loss": 1.2547,
+      "step": 1426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1466315814852094e-05,
+      "loss": 1.2628,
+      "step": 1427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.145576632144542e-05,
+      "loss": 1.2201,
+      "step": 1428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.144521517255132e-05,
+      "loss": 1.2119,
+      "step": 1429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1434662380168487e-05,
+      "loss": 1.2235,
+      "step": 1430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1424107956297479e-05,
+      "loss": 1.2132,
+      "step": 1431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1413551912940721e-05,
+      "loss": 1.2307,
+      "step": 1432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1402994262102468e-05,
+      "loss": 1.2471,
+      "step": 1433
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1392435015788805e-05,
+      "loss": 1.1631,
+      "step": 1434
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.138187418600764e-05,
+      "loss": 1.3305,
+      "step": 1435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1371311784768672e-05,
+      "loss": 1.1862,
+      "step": 1436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360747824083397e-05,
+      "loss": 1.1721,
+      "step": 1437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1350182315965071e-05,
+      "loss": 1.2731,
+      "step": 1438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1339615272428722e-05,
+      "loss": 1.0712,
+      "step": 1439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1329046705491111e-05,
+      "loss": 1.2964,
+      "step": 1440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1318476627170751e-05,
+      "loss": 1.2767,
+      "step": 1441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1307905049487855e-05,
+      "loss": 1.3459,
+      "step": 1442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.129733198446435e-05,
+      "loss": 1.3206,
+      "step": 1443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1286757444123854e-05,
+      "loss": 1.2021,
+      "step": 1444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1276181440491658e-05,
+      "loss": 1.302,
+      "step": 1445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1265603985594723e-05,
+      "loss": 1.3366,
+      "step": 1446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1255025091461657e-05,
+      "loss": 1.3548,
+      "step": 1447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1244444770122707e-05,
+      "loss": 1.2291,
+      "step": 1448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.123386303360974e-05,
+      "loss": 1.459,
+      "step": 1449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1223279893956234e-05,
+      "loss": 1.2989,
+      "step": 1450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1212695363197262e-05,
+      "loss": 1.2016,
+      "step": 1451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.120210945336948e-05,
+      "loss": 1.3596,
+      "step": 1452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1191522176511112e-05,
+      "loss": 1.2772,
+      "step": 1453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1180933544661936e-05,
+      "loss": 1.1584,
+      "step": 1454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1170343569863274e-05,
+      "loss": 1.0758,
+      "step": 1455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1159752264157963e-05,
+      "loss": 1.3042,
+      "step": 1456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1149159639590377e-05,
+      "loss": 1.1603,
+      "step": 1457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1138565708206364e-05,
+      "loss": 1.3142,
+      "step": 1458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1127970482053274e-05,
+      "loss": 1.0895,
+      "step": 1459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1117373973179925e-05,
+      "loss": 1.1043,
+      "step": 1460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1106776193636595e-05,
+      "loss": 1.3214,
+      "step": 1461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1096177155475006e-05,
+      "loss": 1.1407,
+      "step": 1462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1085576870748307e-05,
+      "loss": 1.2877,
+      "step": 1463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1074975351511075e-05,
+      "loss": 1.1184,
+      "step": 1464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1064372609819279e-05,
+      "loss": 1.2395,
+      "step": 1465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1053768657730284e-05,
+      "loss": 1.2231,
+      "step": 1466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1043163507302835e-05,
+      "loss": 1.2085,
+      "step": 1467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1032557170597034e-05,
+      "loss": 1.1212,
+      "step": 1468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1021949659674337e-05,
+      "loss": 1.0985,
+      "step": 1469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1011340986597528e-05,
+      "loss": 1.2408,
+      "step": 1470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.100073116343072e-05,
+      "loss": 1.2321,
+      "step": 1471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0990120202239324e-05,
+      "loss": 1.3038,
+      "step": 1472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.097950811509006e-05,
+      "loss": 1.2647,
+      "step": 1473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.096889491405092e-05,
+      "loss": 1.2462,
+      "step": 1474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0958280611191162e-05,
+      "loss": 1.1796,
+      "step": 1475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0947665218581293e-05,
+      "loss": 1.2872,
+      "step": 1476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0937048748293068e-05,
+      "loss": 1.3001,
+      "step": 1477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0926431212399467e-05,
+      "loss": 1.2903,
+      "step": 1478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0915812622974673e-05,
+      "loss": 1.1658,
+      "step": 1479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0905192992094075e-05,
+      "loss": 1.2011,
+      "step": 1480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0894572331834248e-05,
+      "loss": 1.1497,
+      "step": 1481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0883950654272925e-05,
+      "loss": 1.3064,
+      "step": 1482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0873327971489015e-05,
+      "loss": 1.1824,
+      "step": 1483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.086270429556255e-05,
+      "loss": 1.145,
+      "step": 1484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0852079638574708e-05,
+      "loss": 1.1142,
+      "step": 1485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.084145401260777e-05,
+      "loss": 1.4423,
+      "step": 1486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.083082742974513e-05,
+      "loss": 1.3293,
+      "step": 1487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.082019990207126e-05,
+      "loss": 1.2088,
+      "step": 1488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0809571441671716e-05,
+      "loss": 1.3175,
+      "step": 1489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0798942060633108e-05,
+      "loss": 1.1579,
+      "step": 1490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0788311771043092e-05,
+      "loss": 1.2273,
+      "step": 1491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0777680584990366e-05,
+      "loss": 1.4225,
+      "step": 1492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0767048514564633e-05,
+      "loss": 1.345,
+      "step": 1493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0756415571856617e-05,
+      "loss": 0.9828,
+      "step": 1494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.074578176895803e-05,
+      "loss": 1.276,
+      "step": 1495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.073514711796155e-05,
+      "loss": 1.2913,
+      "step": 1496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0724511630960837e-05,
+      "loss": 1.2195,
+      "step": 1497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0713875320050487e-05,
+      "loss": 1.2488,
+      "step": 1498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0703238197326042e-05,
+      "loss": 1.1805,
+      "step": 1499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0692600274883969e-05,
+      "loss": 1.2677,
+      "step": 1500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0681961564821632e-05,
+      "loss": 1.2276,
+      "step": 1501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0671322079237307e-05,
+      "loss": 1.1109,
+      "step": 1502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0660681830230137e-05,
+      "loss": 1.2044,
+      "step": 1503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0650040829900143e-05,
+      "loss": 1.2021,
+      "step": 1504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06393990903482e-05,
+      "loss": 1.0864,
+      "step": 1505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0628756623676016e-05,
+      "loss": 1.2712,
+      "step": 1506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.061811344198613e-05,
+      "loss": 1.3277,
+      "step": 1507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06074695573819e-05,
+      "loss": 1.1496,
+      "step": 1508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0596824981967475e-05,
+      "loss": 1.2536,
+      "step": 1509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0586179727847791e-05,
+      "loss": 1.3152,
+      "step": 1510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0575533807128567e-05,
+      "loss": 1.2341,
+      "step": 1511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0564887231916256e-05,
+      "loss": 1.2395,
+      "step": 1512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0554240014318085e-05,
+      "loss": 1.2996,
+      "step": 1513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0543592166441983e-05,
+      "loss": 1.1862,
+      "step": 1514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0532943700396616e-05,
+      "loss": 1.3286,
+      "step": 1515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0522294628291347e-05,
+      "loss": 1.2093,
+      "step": 1516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0511644962236228e-05,
+      "loss": 1.2798,
+      "step": 1517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0500994714341984e-05,
+      "loss": 1.2026,
+      "step": 1518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.049034389672e-05,
+      "loss": 1.1517,
+      "step": 1519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0479692521482316e-05,
+      "loss": 0.7474,
+      "step": 1520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0469040600741603e-05,
+      "loss": 0.8453,
+      "step": 1521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0458388146611151e-05,
+      "loss": 0.7812,
+      "step": 1522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.044773517120486e-05,
+      "loss": 0.7486,
+      "step": 1523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0437081686637216e-05,
+      "loss": 0.8325,
+      "step": 1524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0426427705023289e-05,
+      "loss": 0.8364,
+      "step": 1525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0415773238478715e-05,
+      "loss": 0.8639,
+      "step": 1526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0405118299119682e-05,
+      "loss": 0.9195,
+      "step": 1527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.039446289906291e-05,
+      "loss": 0.8059,
+      "step": 1528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0383807050425648e-05,
+      "loss": 0.7991,
+      "step": 1529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.037315076532565e-05,
+      "loss": 0.8506,
+      "step": 1530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0362494055881172e-05,
+      "loss": 0.798,
+      "step": 1531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0351836934210957e-05,
+      "loss": 0.7271,
+      "step": 1532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.03411794124342e-05,
+      "loss": 0.7752,
+      "step": 1533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0330521502670574e-05,
+      "loss": 0.7374,
+      "step": 1534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0319863217040166e-05,
+      "loss": 0.8382,
+      "step": 1535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0309204567663513e-05,
+      "loss": 0.8403,
+      "step": 1536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0298545566661558e-05,
+      "loss": 0.7714,
+      "step": 1537
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.028788622615564e-05,
+      "loss": 0.7968,
+      "step": 1538
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0277226558267494e-05,
+      "loss": 0.7852,
+      "step": 1539
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0266566575119213e-05,
+      "loss": 0.7557,
+      "step": 1540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0255906288833259e-05,
+      "loss": 0.7614,
+      "step": 1541
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0245245711532437e-05,
+      "loss": 0.8238,
+      "step": 1542
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0234584855339884e-05,
+      "loss": 0.6947,
+      "step": 1543
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0223923732379049e-05,
+      "loss": 0.7356,
+      "step": 1544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0213262354773689e-05,
+      "loss": 0.6816,
+      "step": 1545
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.020260073464785e-05,
+      "loss": 0.6796,
+      "step": 1546
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0191938884125847e-05,
+      "loss": 0.7914,
+      "step": 1547
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0181276815332273e-05,
+      "loss": 0.7402,
+      "step": 1548
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0170614540391952e-05,
+      "loss": 0.6915,
+      "step": 1549
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0159952071429953e-05,
+      "loss": 0.7947,
+      "step": 1550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0149289420571558e-05,
+      "loss": 0.7117,
+      "step": 1551
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0138626599942263e-05,
+      "loss": 0.7479,
+      "step": 1552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0127963621667754e-05,
+      "loss": 0.6167,
+      "step": 1553
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0117300497873897e-05,
+      "loss": 0.6633,
+      "step": 1554
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0106637240686723e-05,
+      "loss": 0.7158,
+      "step": 1555
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.009597386223241e-05,
+      "loss": 0.6957,
+      "step": 1556
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0085310374637288e-05,
+      "loss": 0.8345,
+      "step": 1557
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0074646790027792e-05,
+      "loss": 0.8166,
+      "step": 1558
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.006398312053048e-05,
+      "loss": 0.7793,
+      "step": 1559
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0053319378272008e-05,
+      "loss": 0.7357,
+      "step": 1560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0042655575379103e-05,
+      "loss": 0.7202,
+      "step": 1561
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0031991723978574e-05,
+      "loss": 0.8471,
+      "step": 1562
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0021327836197274e-05,
+      "loss": 0.7022,
+      "step": 1563
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.001066392416211e-05,
+      "loss": 0.7363,
+      "step": 1564
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.8191,
+      "step": 1565
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.989336075837893e-06,
+      "loss": 0.831,
+      "step": 1566
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.978672163802726e-06,
+      "loss": 0.846,
+      "step": 1567
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.96800827602143e-06,
+      "loss": 0.8937,
+      "step": 1568
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.9573444246209e-06,
+      "loss": 0.7693,
+      "step": 1569
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.946680621727995e-06,
+      "loss": 0.5848,
+      "step": 1570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.936016879469522e-06,
+      "loss": 0.76,
+      "step": 1571
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.925353209972213e-06,
+      "loss": 0.6796,
+      "step": 1572
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.914689625362717e-06,
+      "loss": 0.7551,
+      "step": 1573
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.904026137767591e-06,
+      "loss": 0.7341,
+      "step": 1574
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.89336275931328e-06,
+      "loss": 0.6881,
+      "step": 1575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.882699502126103e-06,
+      "loss": 0.7877,
+      "step": 1576
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.872036378332251e-06,
+      "loss": 0.6821,
+      "step": 1577
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.86137340005774e-06,
+      "loss": 0.6926,
+      "step": 1578
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.850710579428445e-06,
+      "loss": 0.7044,
+      "step": 1579
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.84004792857005e-06,
+      "loss": 0.6913,
+      "step": 1580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.82938545960805e-06,
+      "loss": 0.7573,
+      "step": 1581
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.818723184667732e-06,
+      "loss": 0.7585,
+      "step": 1582
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.808061115874156e-06,
+      "loss": 0.7767,
+      "step": 1583
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.797399265352154e-06,
+      "loss": 0.721,
+      "step": 1584
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.786737645226313e-06,
+      "loss": 0.7766,
+      "step": 1585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.776076267620955e-06,
+      "loss": 0.697,
+      "step": 1586
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.76541514466012e-06,
+      "loss": 0.7165,
+      "step": 1587
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.754754288467564e-06,
+      "loss": 0.821,
+      "step": 1588
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.744093711166745e-06,
+      "loss": 0.732,
+      "step": 1589
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.733433424880789e-06,
+      "loss": 0.705,
+      "step": 1590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.72277344173251e-06,
+      "loss": 0.7957,
+      "step": 1591
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.712113773844361e-06,
+      "loss": 0.7128,
+      "step": 1592
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.701454433338445e-06,
+      "loss": 0.7031,
+      "step": 1593
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.690795432336489e-06,
+      "loss": 0.7189,
+      "step": 1594
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.680136782959835e-06,
+      "loss": 0.7433,
+      "step": 1595
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.669478497329431e-06,
+      "loss": 0.7002,
+      "step": 1596
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.658820587565801e-06,
+      "loss": 0.7529,
+      "step": 1597
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.648163065789046e-06,
+      "loss": 0.6715,
+      "step": 1598
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.637505944118828e-06,
+      "loss": 0.7375,
+      "step": 1599
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.626849234674356e-06,
+      "loss": 0.7689,
+      "step": 1600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.616192949574357e-06,
+      "loss": 0.7926,
+      "step": 1601
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.605537100937094e-06,
+      "loss": 0.7911,
+      "step": 1602
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.59488170088032e-06,
+      "loss": 0.7955,
+      "step": 1603
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.584226761521285e-06,
+      "loss": 0.8466,
+      "step": 1604
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.573572294976714e-06,
+      "loss": 0.6859,
+      "step": 1605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.562918313362787e-06,
+      "loss": 0.6251,
+      "step": 1606
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.552264828795142e-06,
+      "loss": 0.7441,
+      "step": 1607
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.541611853388849e-06,
+      "loss": 0.7742,
+      "step": 1608
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.530959399258397e-06,
+      "loss": 0.755,
+      "step": 1609
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.520307478517686e-06,
+      "loss": 0.6872,
+      "step": 1610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.509656103280004e-06,
+      "loss": 0.7239,
+      "step": 1611
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.499005285658021e-06,
+      "loss": 0.8755,
+      "step": 1612
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.488355037763774e-06,
+      "loss": 0.8583,
+      "step": 1613
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.477705371708657e-06,
+      "loss": 0.7734,
+      "step": 1614
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.467056299603388e-06,
+      "loss": 0.7673,
+      "step": 1615
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.456407833558019e-06,
+      "loss": 0.7102,
+      "step": 1616
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.445759985681919e-06,
+      "loss": 0.7643,
+      "step": 1617
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.435112768083742e-06,
+      "loss": 0.7464,
+      "step": 1618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.42446619287144e-06,
+      "loss": 0.6291,
+      "step": 1619
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41382027215221e-06,
+      "loss": 0.8474,
+      "step": 1620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.403175018032528e-06,
+      "loss": 0.824,
+      "step": 1621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.3925304426181e-06,
+      "loss": 0.67,
+      "step": 1622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.38188655801387e-06,
+      "loss": 0.78,
+      "step": 1623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.37124337632399e-06,
+      "loss": 0.6538,
+      "step": 1624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.360600909651804e-06,
+      "loss": 0.6635,
+      "step": 1625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.349959170099859e-06,
+      "loss": 0.7027,
+      "step": 1626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.339318169769865e-06,
+      "loss": 0.7704,
+      "step": 1627
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.328677920762698e-06,
+      "loss": 0.7284,
+      "step": 1628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.318038435178371e-06,
+      "loss": 0.7295,
+      "step": 1629
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.307399725116035e-06,
+      "loss": 0.7458,
+      "step": 1630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.29676180267396e-06,
+      "loss": 0.701,
+      "step": 1631
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.286124679949515e-06,
+      "loss": 0.7679,
+      "step": 1632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.275488369039169e-06,
+      "loss": 0.7512,
+      "step": 1633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.264852882038453e-06,
+      "loss": 0.7408,
+      "step": 1634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.254218231041973e-06,
+      "loss": 0.7379,
+      "step": 1635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.243584428143383e-06,
+      "loss": 0.74,
+      "step": 1636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.232951485435367e-06,
+      "loss": 0.7516,
+      "step": 1637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.222319415009639e-06,
+      "loss": 0.8017,
+      "step": 1638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.21168822895691e-06,
+      "loss": 0.7807,
+      "step": 1639
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.201057939366896e-06,
+      "loss": 0.7438,
+      "step": 1640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.190428558328285e-06,
+      "loss": 0.6397,
+      "step": 1641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.179800097928743e-06,
+      "loss": 0.7843,
+      "step": 1642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.169172570254874e-06,
+      "loss": 0.7219,
+      "step": 1643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.158545987392233e-06,
+      "loss": 1.1518,
+      "step": 1644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.147920361425295e-06,
+      "loss": 0.8327,
+      "step": 1645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.13729570443745e-06,
+      "loss": 0.7701,
+      "step": 1646
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12667202851099e-06,
+      "loss": 0.693,
+      "step": 1647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.116049345727076e-06,
+      "loss": 0.6348,
+      "step": 1648
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.105427668165755e-06,
+      "loss": 0.7202,
+      "step": 1649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.094807007905927e-06,
+      "loss": 0.8613,
+      "step": 1650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084187377025334e-06,
+      "loss": 0.6858,
+      "step": 1651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.073568787600538e-06,
+      "loss": 0.7561,
+      "step": 1652
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.062951251706933e-06,
+      "loss": 0.5943,
+      "step": 1653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.052334781418708e-06,
+      "loss": 0.9754,
+      "step": 1654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.04171938880884e-06,
+      "loss": 0.8839,
+      "step": 1655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.031105085949083e-06,
+      "loss": 0.7718,
+      "step": 1656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.020491884909941e-06,
+      "loss": 0.7039,
+      "step": 1657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.009879797760679e-06,
+      "loss": 0.8086,
+      "step": 1658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.999268836569284e-06,
+      "loss": 0.8561,
+      "step": 1659
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.988659013402474e-06,
+      "loss": 0.7165,
+      "step": 1660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.978050340325668e-06,
+      "loss": 0.6422,
+      "step": 1661
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.967442829402968e-06,
+      "loss": 0.8316,
+      "step": 1662
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.956836492697167e-06,
+      "loss": 0.7368,
+      "step": 1663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.94623134226972e-06,
+      "loss": 0.7332,
+      "step": 1664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.935627390180728e-06,
+      "loss": 0.7195,
+      "step": 1665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.925024648488932e-06,
+      "loss": 0.7806,
+      "step": 1666
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.914423129251697e-06,
+      "loss": 0.6617,
+      "step": 1667
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.903822844524999e-06,
+      "loss": 0.6465,
+      "step": 1668
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.893223806363407e-06,
+      "loss": 0.7953,
+      "step": 1669
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.882626026820078e-06,
+      "loss": 0.7214,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.872029517946729e-06,
+      "loss": 0.754,
+      "step": 1671
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.861434291793639e-06,
+      "loss": 0.7778,
+      "step": 1672
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.850840360409627e-06,
+      "loss": 0.8362,
+      "step": 1673
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.840247735842037e-06,
+      "loss": 0.7572,
+      "step": 1674
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.829656430136732e-06,
+      "loss": 0.7194,
+      "step": 1675
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.819066455338067e-06,
+      "loss": 0.6776,
+      "step": 1676
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.80847782348889e-06,
+      "loss": 0.8073,
+      "step": 1677
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.797890546630522e-06,
+      "loss": 0.6929,
+      "step": 1678
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.787304636802743e-06,
+      "loss": 0.7384,
+      "step": 1679
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.77672010604377e-06,
+      "loss": 0.7707,
+      "step": 1680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.766136966390263e-06,
+      "loss": 0.6968,
+      "step": 1681
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.755555229877295e-06,
+      "loss": 0.8593,
+      "step": 1682
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.744974908538345e-06,
+      "loss": 0.7993,
+      "step": 1683
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.734396014405282e-06,
+      "loss": 0.7381,
+      "step": 1684
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.723818559508345e-06,
+      "loss": 0.7781,
+      "step": 1685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.713242555876149e-06,
+      "loss": 0.7275,
+      "step": 1686
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.702668015535651e-06,
+      "loss": 0.6601,
+      "step": 1687
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.692094950512145e-06,
+      "loss": 0.665,
+      "step": 1688
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.681523372829254e-06,
+      "loss": 0.8076,
+      "step": 1689
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.670953294508892e-06,
+      "loss": 0.7575,
+      "step": 1690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.660384727571282e-06,
+      "loss": 0.7832,
+      "step": 1691
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.64981768403493e-06,
+      "loss": 0.7434,
+      "step": 1692
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.639252175916608e-06,
+      "loss": 0.6912,
+      "step": 1693
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.62868821523133e-06,
+      "loss": 0.7807,
+      "step": 1694
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.618125813992364e-06,
+      "loss": 0.6817,
+      "step": 1695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.607564984211199e-06,
+      "loss": 0.7088,
+      "step": 1696
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.597005737897534e-06,
+      "loss": 0.7503,
+      "step": 1697
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.586448087059284e-06,
+      "loss": 0.701,
+      "step": 1698
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.575892043702523e-06,
+      "loss": 0.83,
+      "step": 1699
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.565337619831516e-06,
+      "loss": 0.6545,
+      "step": 1700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.554784827448681e-06,
+      "loss": 0.6521,
+      "step": 1701
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.544233678554581e-06,
+      "loss": 0.8336,
+      "step": 1702
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.533684185147913e-06,
+      "loss": 0.802,
+      "step": 1703
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.52313635922548e-06,
+      "loss": 0.6752,
+      "step": 1704
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.512590212782203e-06,
+      "loss": 0.6158,
+      "step": 1705
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.502045757811087e-06,
+      "loss": 0.8078,
+      "step": 1706
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.491503006303212e-06,
+      "loss": 0.7619,
+      "step": 1707
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.480961970247722e-06,
+      "loss": 0.7769,
+      "step": 1708
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.47042266163181e-06,
+      "loss": 0.6736,
+      "step": 1709
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.459885092440708e-06,
+      "loss": 0.8792,
+      "step": 1710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.449349274657665e-06,
+      "loss": 0.7375,
+      "step": 1711
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.438815220263942e-06,
+      "loss": 0.6718,
+      "step": 1712
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.42828294123879e-06,
+      "loss": 0.7983,
+      "step": 1713
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.417752449559447e-06,
+      "loss": 0.8696,
+      "step": 1714
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.407223757201113e-06,
+      "loss": 0.6301,
+      "step": 1715
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.396696876136948e-06,
+      "loss": 0.6936,
+      "step": 1716
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.386171818338048e-06,
+      "loss": 0.768,
+      "step": 1717
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.37564859577343e-06,
+      "loss": 0.82,
+      "step": 1718
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.365127220410037e-06,
+      "loss": 0.7359,
+      "step": 1719
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.354607704212703e-06,
+      "loss": 0.7153,
+      "step": 1720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.344090059144151e-06,
+      "loss": 0.8518,
+      "step": 1721
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.333574297164969e-06,
+      "loss": 0.6675,
+      "step": 1722
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.323060430233613e-06,
+      "loss": 0.8015,
+      "step": 1723
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.312548470306379e-06,
+      "loss": 0.7307,
+      "step": 1724
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.302038429337396e-06,
+      "loss": 0.722,
+      "step": 1725
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.291530319278614e-06,
+      "loss": 0.8462,
+      "step": 1726
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.281024152079775e-06,
+      "loss": 0.8232,
+      "step": 1727
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.270519939688421e-06,
+      "loss": 0.7586,
+      "step": 1728
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.260017694049878e-06,
+      "loss": 0.8015,
+      "step": 1729
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.249517427107225e-06,
+      "loss": 0.7307,
+      "step": 1730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.239019150801286e-06,
+      "loss": 0.6664,
+      "step": 1731
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.228522877070636e-06,
+      "loss": 0.7064,
+      "step": 1732
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.21802861785156e-06,
+      "loss": 0.758,
+      "step": 1733
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.207536385078057e-06,
+      "loss": 0.8387,
+      "step": 1734
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.197046190681832e-06,
+      "loss": 0.6874,
+      "step": 1735
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.186558046592247e-06,
+      "loss": 0.8579,
+      "step": 1736
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.176071964736358e-06,
+      "loss": 0.6297,
+      "step": 1737
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.16558795703886e-06,
+      "loss": 0.6555,
+      "step": 1738
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.155106035422095e-06,
+      "loss": 0.6886,
+      "step": 1739
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.144626211806038e-06,
+      "loss": 0.7178,
+      "step": 1740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.13414849810826e-06,
+      "loss": 0.7144,
+      "step": 1741
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.123672906243955e-06,
+      "loss": 0.7969,
+      "step": 1742
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11319944812589e-06,
+      "loss": 0.7901,
+      "step": 1743
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.102728135664411e-06,
+      "loss": 0.7181,
+      "step": 1744
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.09225898076742e-06,
+      "loss": 0.8622,
+      "step": 1745
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.08179199534037e-06,
+      "loss": 0.8609,
+      "step": 1746
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.071327191286244e-06,
+      "loss": 0.7806,
+      "step": 1747
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.060864580505542e-06,
+      "loss": 0.6724,
+      "step": 1748
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.050404174896282e-06,
+      "loss": 0.7419,
+      "step": 1749
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.039945986353954e-06,
+      "loss": 0.7907,
+      "step": 1750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.029490026771545e-06,
+      "loss": 0.6673,
+      "step": 1751
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.019036308039497e-06,
+      "loss": 0.8518,
+      "step": 1752
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.008584842045706e-06,
+      "loss": 0.7744,
+      "step": 1753
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.998135640675514e-06,
+      "loss": 0.8429,
+      "step": 1754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.987688715811672e-06,
+      "loss": 0.7407,
+      "step": 1755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.977244079334357e-06,
+      "loss": 0.7023,
+      "step": 1756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.966801743121133e-06,
+      "loss": 0.7511,
+      "step": 1757
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.95636171904696e-06,
+      "loss": 0.6814,
+      "step": 1758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.945924018984155e-06,
+      "loss": 0.8168,
+      "step": 1759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.935488654802395e-06,
+      "loss": 0.6846,
+      "step": 1760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.925055638368713e-06,
+      "loss": 0.6312,
+      "step": 1761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.914624981547459e-06,
+      "loss": 0.712,
+      "step": 1762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.90419669620031e-06,
+      "loss": 0.8883,
+      "step": 1763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.893770794186231e-06,
+      "loss": 0.7941,
+      "step": 1764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.88334728736149e-06,
+      "loss": 0.7611,
+      "step": 1765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.872926187579625e-06,
+      "loss": 0.729,
+      "step": 1766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.862507506691444e-06,
+      "loss": 0.7939,
+      "step": 1767
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.852091256545e-06,
+      "loss": 0.713,
+      "step": 1768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.841677448985572e-06,
+      "loss": 0.8089,
+      "step": 1769
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.831266095855675e-06,
+      "loss": 0.6924,
+      "step": 1770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.82085720899503e-06,
+      "loss": 0.6898,
+      "step": 1771
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.810450800240549e-06,
+      "loss": 0.7316,
+      "step": 1772
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.800046881426331e-06,
+      "loss": 0.7067,
+      "step": 1773
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.789645464383635e-06,
+      "loss": 0.6733,
+      "step": 1774
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.779246560940887e-06,
+      "loss": 0.8138,
+      "step": 1775
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.768850182923646e-06,
+      "loss": 0.7343,
+      "step": 1776
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.758456342154602e-06,
+      "loss": 0.8757,
+      "step": 1777
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.748065050453557e-06,
+      "loss": 0.7713,
+      "step": 1778
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.737676319637417e-06,
+      "loss": 0.7135,
+      "step": 1779
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.727290161520176e-06,
+      "loss": 0.8717,
+      "step": 1780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.716906587912903e-06,
+      "loss": 0.6306,
+      "step": 1781
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.706525610623726e-06,
+      "loss": 0.7388,
+      "step": 1782
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.696147241457817e-06,
+      "loss": 0.6993,
+      "step": 1783
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.685771492217387e-06,
+      "loss": 0.6875,
+      "step": 1784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.675398374701669e-06,
+      "loss": 0.821,
+      "step": 1785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.665027900706902e-06,
+      "loss": 0.7293,
+      "step": 1786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.654660082026313e-06,
+      "loss": 0.7566,
+      "step": 1787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.644294930450117e-06,
+      "loss": 0.8534,
+      "step": 1788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.633932457765491e-06,
+      "loss": 0.7217,
+      "step": 1789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.623572675756568e-06,
+      "loss": 0.7808,
+      "step": 1790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.613215596204425e-06,
+      "loss": 0.7292,
+      "step": 1791
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.602861230887055e-06,
+      "loss": 0.73,
+      "step": 1792
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.592509591579373e-06,
+      "loss": 0.8007,
+      "step": 1793
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.582160690053194e-06,
+      "loss": 0.9258,
+      "step": 1794
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.571814538077216e-06,
+      "loss": 0.7619,
+      "step": 1795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5614711474170155e-06,
+      "loss": 0.7193,
+      "step": 1796
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.551130529835016e-06,
+      "loss": 0.7072,
+      "step": 1797
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.540792697090502e-06,
+      "loss": 0.6891,
+      "step": 1798
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.530457660939588e-06,
+      "loss": 0.7172,
+      "step": 1799
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5201254331352056e-06,
+      "loss": 0.7301,
+      "step": 1800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.509796025427088e-06,
+      "loss": 0.7413,
+      "step": 1801
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.49946944956177e-06,
+      "loss": 0.7004,
+      "step": 1802
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.489145717282559e-06,
+      "loss": 0.5519,
+      "step": 1803
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.478824840329537e-06,
+      "loss": 0.7054,
+      "step": 1804
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4685068304395366e-06,
+      "loss": 0.8361,
+      "step": 1805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.45819169934612e-06,
+      "loss": 0.7112,
+      "step": 1806
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4478794587795855e-06,
+      "loss": 0.813,
+      "step": 1807
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.437570120466943e-06,
+      "loss": 0.7203,
+      "step": 1808
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.427263696131903e-06,
+      "loss": 0.7893,
+      "step": 1809
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.416960197494856e-06,
+      "loss": 0.7818,
+      "step": 1810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.406659636272872e-06,
+      "loss": 0.6633,
+      "step": 1811
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.396362024179677e-06,
+      "loss": 0.7338,
+      "step": 1812
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.386067372925647e-06,
+      "loss": 0.735,
+      "step": 1813
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.375775694217788e-06,
+      "loss": 0.81,
+      "step": 1814
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.365486999759723e-06,
+      "loss": 0.7522,
+      "step": 1815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.355201301251689e-06,
+      "loss": 0.6648,
+      "step": 1816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.344918610390506e-06,
+      "loss": 0.6333,
+      "step": 1817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.334638938869582e-06,
+      "loss": 0.68,
+      "step": 1818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.324362298378893e-06,
+      "loss": 0.7931,
+      "step": 1819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.314088700604958e-06,
+      "loss": 0.7953,
+      "step": 1820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.303818157230841e-06,
+      "loss": 0.6957,
+      "step": 1821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.293550679936135e-06,
+      "loss": 0.7854,
+      "step": 1822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.283286280396948e-06,
+      "loss": 0.6751,
+      "step": 1823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.27302497028588e-06,
+      "loss": 0.7078,
+      "step": 1824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.262766761272022e-06,
+      "loss": 0.7975,
+      "step": 1825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.252511665020939e-06,
+      "loss": 0.6619,
+      "step": 1826
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.242259693194658e-06,
+      "loss": 0.7909,
+      "step": 1827
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.232010857451653e-06,
+      "loss": 0.6512,
+      "step": 1828
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.221765169446819e-06,
+      "loss": 0.748,
+      "step": 1829
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.211522640831492e-06,
+      "loss": 0.6781,
+      "step": 1830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.2012832832534e-06,
+      "loss": 0.7517,
+      "step": 1831
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1910471083566725e-06,
+      "loss": 0.7194,
+      "step": 1832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.180814127781819e-06,
+      "loss": 0.658,
+      "step": 1833
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.170584353165708e-06,
+      "loss": 0.7403,
+      "step": 1834
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1603577961415704e-06,
+      "loss": 0.7236,
+      "step": 1835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.15013446833898e-06,
+      "loss": 0.8157,
+      "step": 1836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.139914381383837e-06,
+      "loss": 0.7308,
+      "step": 1837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.129697546898344e-06,
+      "loss": 0.7194,
+      "step": 1838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.119483976501021e-06,
+      "loss": 0.8648,
+      "step": 1839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.10927368180667e-06,
+      "loss": 0.6839,
+      "step": 1840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0990666744263646e-06,
+      "loss": 0.7048,
+      "step": 1841
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0888629659674525e-06,
+      "loss": 0.695,
+      "step": 1842
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.07866256803351e-06,
+      "loss": 0.7182,
+      "step": 1843
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.068465492224361e-06,
+      "loss": 0.6605,
+      "step": 1844
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.058271750136054e-06,
+      "loss": 0.725,
+      "step": 1845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.04808135336084e-06,
+      "loss": 0.7677,
+      "step": 1846
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.03789431348717e-06,
+      "loss": 0.6982,
+      "step": 1847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.027710642099672e-06,
+      "loss": 0.6249,
+      "step": 1848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.017530350779148e-06,
+      "loss": 0.8421,
+      "step": 1849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.007353451102557e-06,
+      "loss": 0.8239,
+      "step": 1850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.997179954642998e-06,
+      "loss": 0.6682,
+      "step": 1851
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9870098729697015e-06,
+      "loss": 0.7499,
+      "step": 1852
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.976843217648013e-06,
+      "loss": 0.6502,
+      "step": 1853
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.966680000239384e-06,
+      "loss": 0.6084,
+      "step": 1854
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.956520232301355e-06,
+      "loss": 0.762,
+      "step": 1855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.946363925387546e-06,
+      "loss": 0.7446,
+      "step": 1856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.936211091047636e-06,
+      "loss": 0.8434,
+      "step": 1857
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.926061740827359e-06,
+      "loss": 0.8147,
+      "step": 1858
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.915915886268486e-06,
+      "loss": 0.7941,
+      "step": 1859
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.905773538908814e-06,
+      "loss": 0.7455,
+      "step": 1860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.895634710282153e-06,
+      "loss": 0.8107,
+      "step": 1861
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.885499411918304e-06,
+      "loss": 0.7882,
+      "step": 1862
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8753676553430606e-06,
+      "loss": 0.8008,
+      "step": 1863
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8652394520781865e-06,
+      "loss": 0.7703,
+      "step": 1864
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.855114813641408e-06,
+      "loss": 0.8034,
+      "step": 1865
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.844993751546388e-06,
+      "loss": 0.7733,
+      "step": 1866
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8348762773027285e-06,
+      "loss": 0.6668,
+      "step": 1867
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.824762402415958e-06,
+      "loss": 0.8248,
+      "step": 1868
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.814652138387501e-06,
+      "loss": 0.7283,
+      "step": 1869
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8045454967146854e-06,
+      "loss": 0.7483,
+      "step": 1870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.794442488890705e-06,
+      "loss": 0.7233,
+      "step": 1871
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.784343126404636e-06,
+      "loss": 0.8548,
+      "step": 1872
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.774247420741401e-06,
+      "loss": 0.7449,
+      "step": 1873
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.764155383381772e-06,
+      "loss": 0.8191,
+      "step": 1874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.754067025802346e-06,
+      "loss": 0.7457,
+      "step": 1875
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.743982359475525e-06,
+      "loss": 0.647,
+      "step": 1876
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.733901395869527e-06,
+      "loss": 0.8944,
+      "step": 1877
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.723824146448352e-06,
+      "loss": 0.695,
+      "step": 1878
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.713750622671784e-06,
+      "loss": 0.596,
+      "step": 1879
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7036808359953585e-06,
+      "loss": 0.5971,
+      "step": 1880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.693614797870371e-06,
+      "loss": 0.6719,
+      "step": 1881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.68355251974385e-06,
+      "loss": 0.7233,
+      "step": 1882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.673494013058548e-06,
+      "loss": 0.7284,
+      "step": 1883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.663439289252932e-06,
+      "loss": 0.8362,
+      "step": 1884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.653388359761161e-06,
+      "loss": 0.8038,
+      "step": 1885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6433412360130855e-06,
+      "loss": 0.7687,
+      "step": 1886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.633297929434224e-06,
+      "loss": 0.6707,
+      "step": 1887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.623258451445758e-06,
+      "loss": 0.7155,
+      "step": 1888
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.613222813464508e-06,
+      "loss": 0.6336,
+      "step": 1889
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.603191026902936e-06,
+      "loss": 0.7285,
+      "step": 1890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.59316310316912e-06,
+      "loss": 0.6758,
+      "step": 1891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.583139053666745e-06,
+      "loss": 0.7682,
+      "step": 1892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.573118889795093e-06,
+      "loss": 0.6451,
+      "step": 1893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.563102622949021e-06,
+      "loss": 0.733,
+      "step": 1894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.553090264518962e-06,
+      "loss": 0.7325,
+      "step": 1895
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.543081825890901e-06,
+      "loss": 0.7829,
+      "step": 1896
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.533077318446365e-06,
+      "loss": 0.7278,
+      "step": 1897
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.523076753562411e-06,
+      "loss": 0.7916,
+      "step": 1898
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.513080142611612e-06,
+      "loss": 0.7904,
+      "step": 1899
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.503087496962044e-06,
+      "loss": 0.6089,
+      "step": 1900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.493098827977278e-06,
+      "loss": 0.7025,
+      "step": 1901
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.483114147016358e-06,
+      "loss": 0.7289,
+      "step": 1902
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.473133465433793e-06,
+      "loss": 0.754,
+      "step": 1903
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.463156794579543e-06,
+      "loss": 0.7923,
+      "step": 1904
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.453184145799015e-06,
+      "loss": 0.7206,
+      "step": 1905
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4432155304330335e-06,
+      "loss": 0.6077,
+      "step": 1906
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.433250959817845e-06,
+      "loss": 0.6976,
+      "step": 1907
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.42329044528508e-06,
+      "loss": 0.7523,
+      "step": 1908
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.413333998161772e-06,
+      "loss": 0.6929,
+      "step": 1909
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4033816297703245e-06,
+      "loss": 0.8903,
+      "step": 1910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.393433351428502e-06,
+      "loss": 0.6549,
+      "step": 1911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.383489174449422e-06,
+      "loss": 0.6684,
+      "step": 1912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.373549110141528e-06,
+      "loss": 0.6565,
+      "step": 1913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.363613169808593e-06,
+      "loss": 0.6944,
+      "step": 1914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.353681364749703e-06,
+      "loss": 0.7965,
+      "step": 1915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.343753706259239e-06,
+      "loss": 0.7501,
+      "step": 1916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.333830205626861e-06,
+      "loss": 0.7404,
+      "step": 1917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.323910874137509e-06,
+      "loss": 0.7629,
+      "step": 1918
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.313995723071377e-06,
+      "loss": 0.8053,
+      "step": 1919
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3040847637039075e-06,
+      "loss": 0.6441,
+      "step": 1920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.294178007305776e-06,
+      "loss": 0.718,
+      "step": 1921
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.284275465142875e-06,
+      "loss": 0.7312,
+      "step": 1922
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.27437714847631e-06,
+      "loss": 0.7502,
+      "step": 1923
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2644830685623756e-06,
+      "loss": 0.8266,
+      "step": 1924
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.254593236652554e-06,
+      "loss": 0.7382,
+      "step": 1925
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.244707663993494e-06,
+      "loss": 0.7314,
+      "step": 1926
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2348263618269975e-06,
+      "loss": 0.7311,
+      "step": 1927
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.224949341390017e-06,
+      "loss": 0.7672,
+      "step": 1928
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.215076613914628e-06,
+      "loss": 0.7134,
+      "step": 1929
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.205208190628034e-06,
+      "loss": 0.7885,
+      "step": 1930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1953440827525306e-06,
+      "loss": 0.7421,
+      "step": 1931
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.185484301505518e-06,
+      "loss": 0.7605,
+      "step": 1932
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.175628858099471e-06,
+      "loss": 0.7185,
+      "step": 1933
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.165777763741932e-06,
+      "loss": 0.7015,
+      "step": 1934
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1559310296355005e-06,
+      "loss": 0.6552,
+      "step": 1935
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.146088666977804e-06,
+      "loss": 0.7147,
+      "step": 1936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1362506869615205e-06,
+      "loss": 0.6034,
+      "step": 1937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.126417100774329e-06,
+      "loss": 0.6316,
+      "step": 1938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.116587919598915e-06,
+      "loss": 0.6678,
+      "step": 1939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.106763154612963e-06,
+      "loss": 0.6918,
+      "step": 1940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.096942816989116e-06,
+      "loss": 0.7406,
+      "step": 1941
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0871269178949986e-06,
+      "loss": 0.7056,
+      "step": 1942
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0773154684931875e-06,
+      "loss": 0.6214,
+      "step": 1943
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0675084799411975e-06,
+      "loss": 0.648,
+      "step": 1944
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.057705963391461e-06,
+      "loss": 0.6987,
+      "step": 1945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.047907929991333e-06,
+      "loss": 0.7254,
+      "step": 1946
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.038114390883073e-06,
+      "loss": 0.7628,
+      "step": 1947
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.028325357203822e-06,
+      "loss": 0.6533,
+      "step": 1948
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.018540840085611e-06,
+      "loss": 0.9035,
+      "step": 1949
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.008760850655316e-06,
+      "loss": 0.7052,
+      "step": 1950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.998985400034676e-06,
+      "loss": 0.6983,
+      "step": 1951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.989214499340267e-06,
+      "loss": 0.6291,
+      "step": 1952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.979448159683491e-06,
+      "loss": 0.7766,
+      "step": 1953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.96968639217056e-06,
+      "loss": 0.6956,
+      "step": 1954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.959929207902488e-06,
+      "loss": 0.742,
+      "step": 1955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.950176617975079e-06,
+      "loss": 0.7034,
+      "step": 1956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9404286334789115e-06,
+      "loss": 0.6267,
+      "step": 1957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.930685265499329e-06,
+      "loss": 0.589,
+      "step": 1958
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9209465251164175e-06,
+      "loss": 0.8011,
+      "step": 1959
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.911212423405008e-06,
+      "loss": 0.6309,
+      "step": 1960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.901482971434652e-06,
+      "loss": 0.6323,
+      "step": 1961
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.891758180269618e-06,
+      "loss": 0.71,
+      "step": 1962
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.882038060968871e-06,
+      "loss": 0.6835,
+      "step": 1963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.872322624586061e-06,
+      "loss": 0.7589,
+      "step": 1964
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.862611882169518e-06,
+      "loss": 0.7705,
+      "step": 1965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.852905844762229e-06,
+      "loss": 0.7749,
+      "step": 1966
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8432045234018365e-06,
+      "loss": 0.7817,
+      "step": 1967
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.833507929120612e-06,
+      "loss": 0.6737,
+      "step": 1968
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.823816072945458e-06,
+      "loss": 0.6447,
+      "step": 1969
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.814128965897887e-06,
+      "loss": 0.7379,
+      "step": 1970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.804446618994009e-06,
+      "loss": 0.613,
+      "step": 1971
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.794769043244527e-06,
+      "loss": 0.5713,
+      "step": 1972
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.785096249654705e-06,
+      "loss": 0.8384,
+      "step": 1973
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.775428249224382e-06,
+      "loss": 0.7849,
+      "step": 1974
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.765765052947941e-06,
+      "loss": 0.9222,
+      "step": 1975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.756106671814301e-06,
+      "loss": 0.8544,
+      "step": 1976
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7464531168069096e-06,
+      "loss": 0.7115,
+      "step": 1977
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7368043989037206e-06,
+      "loss": 0.7258,
+      "step": 1978
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7271605290771895e-06,
+      "loss": 0.7261,
+      "step": 1979
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.717521518294259e-06,
+      "loss": 0.6235,
+      "step": 1980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.70788737751635e-06,
+      "loss": 0.6413,
+      "step": 1981
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.698258117699334e-06,
+      "loss": 0.7574,
+      "step": 1982
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.68863374979354e-06,
+      "loss": 0.7668,
+      "step": 1983
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6790142847437355e-06,
+      "loss": 0.7314,
+      "step": 1984
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.669399733489108e-06,
+      "loss": 0.7755,
+      "step": 1985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.659790106963261e-06,
+      "loss": 0.7053,
+      "step": 1986
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.650185416094194e-06,
+      "loss": 0.6335,
+      "step": 1987
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.640585671804296e-06,
+      "loss": 0.6977,
+      "step": 1988
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.63099088501033e-06,
+      "loss": 0.7254,
+      "step": 1989
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.621401066623421e-06,
+      "loss": 0.75,
+      "step": 1990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.611816227549048e-06,
+      "loss": 0.7286,
+      "step": 1991
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.602236378687018e-06,
+      "loss": 0.7389,
+      "step": 1992
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.59266153093147e-06,
+      "loss": 0.7002,
+      "step": 1993
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.583091695170857e-06,
+      "loss": 0.8015,
+      "step": 1994
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5735268822879344e-06,
+      "loss": 0.6768,
+      "step": 1995
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.563967103159733e-06,
+      "loss": 0.7159,
+      "step": 1996
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.554412368657572e-06,
+      "loss": 0.6989,
+      "step": 1997
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.544862689647024e-06,
+      "loss": 0.7128,
+      "step": 1998
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.535318076987928e-06,
+      "loss": 0.5933,
+      "step": 1999
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.52577854153435e-06,
+      "loss": 0.7736,
+      "step": 2000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.516244094134578e-06,
+      "loss": 0.6531,
+      "step": 2001
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.506714745631124e-06,
+      "loss": 0.8514,
+      "step": 2002
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.497190506860695e-06,
+      "loss": 0.8415,
+      "step": 2003
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.487671388654192e-06,
+      "loss": 0.7159,
+      "step": 2004
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.478157401836694e-06,
+      "loss": 0.7302,
+      "step": 2005
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.468648557227434e-06,
+      "loss": 0.6675,
+      "step": 2006
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.459144865639806e-06,
+      "loss": 0.7683,
+      "step": 2007
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.449646337881345e-06,
+      "loss": 0.7269,
+      "step": 2008
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.440152984753709e-06,
+      "loss": 0.7207,
+      "step": 2009
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.430664817052678e-06,
+      "loss": 0.7395,
+      "step": 2010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.421181845568125e-06,
+      "loss": 0.7995,
+      "step": 2011
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.411704081084025e-06,
+      "loss": 0.666,
+      "step": 2012
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.402231534378424e-06,
+      "loss": 0.6999,
+      "step": 2013
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.392764216223441e-06,
+      "loss": 0.7382,
+      "step": 2014
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.383302137385238e-06,
+      "loss": 0.8523,
+      "step": 2015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.373845308624028e-06,
+      "loss": 0.6393,
+      "step": 2016
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.364393740694054e-06,
+      "loss": 0.6108,
+      "step": 2017
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3549474443435726e-06,
+      "loss": 0.6351,
+      "step": 2018
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.345506430314845e-06,
+      "loss": 0.6475,
+      "step": 2019
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.33607070934413e-06,
+      "loss": 0.6502,
+      "step": 2020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.326640292161665e-06,
+      "loss": 0.6687,
+      "step": 2021
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.317215189491651e-06,
+      "loss": 0.5928,
+      "step": 2022
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.307795412052259e-06,
+      "loss": 0.6536,
+      "step": 2023
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.298380970555584e-06,
+      "loss": 0.6886,
+      "step": 2024
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.288971875707669e-06,
+      "loss": 0.6748,
+      "step": 2025
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.279568138208473e-06,
+      "loss": 0.7961,
+      "step": 2026
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.270169768751858e-06,
+      "loss": 0.756,
+      "step": 2027
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.260776778025591e-06,
+      "loss": 0.7973,
+      "step": 2028
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.251389176711307e-06,
+      "loss": 0.7368,
+      "step": 2029
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.242006975484528e-06,
+      "loss": 0.706,
+      "step": 2030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2326301850146285e-06,
+      "loss": 0.7257,
+      "step": 2031
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.223258815964833e-06,
+      "loss": 0.7576,
+      "step": 2032
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.213892878992191e-06,
+      "loss": 0.7511,
+      "step": 2033
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.204532384747584e-06,
+      "loss": 0.7854,
+      "step": 2034
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.195177343875701e-06,
+      "loss": 0.6626,
+      "step": 2035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.185827767015031e-06,
+      "loss": 0.6775,
+      "step": 2036
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.176483664797852e-06,
+      "loss": 0.721,
+      "step": 2037
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1671450478502064e-06,
+      "loss": 0.7386,
+      "step": 2038
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.157811926791905e-06,
+      "loss": 0.738,
+      "step": 2039
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1484843122365125e-06,
+      "loss": 0.8146,
+      "step": 2040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.139162214791322e-06,
+      "loss": 0.5877,
+      "step": 2041
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.129845645057372e-06,
+      "loss": 0.7367,
+      "step": 2042
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.120534613629389e-06,
+      "loss": 0.7743,
+      "step": 2043
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1112291310958166e-06,
+      "loss": 0.8649,
+      "step": 2044
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.101929208038786e-06,
+      "loss": 0.7196,
+      "step": 2045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.09263485503411e-06,
+      "loss": 0.5693,
+      "step": 2046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.083346082651255e-06,
+      "loss": 0.7333,
+      "step": 2047
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.074062901453352e-06,
+      "loss": 0.7562,
+      "step": 2048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.06478532199717e-06,
+      "loss": 0.7743,
+      "step": 2049
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.05551335483311e-06,
+      "loss": 0.7186,
+      "step": 2050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.04624701050519e-06,
+      "loss": 0.6196,
+      "step": 2051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.03698629955103e-06,
+      "loss": 0.6444,
+      "step": 2052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.027731232501849e-06,
+      "loss": 0.8377,
+      "step": 2053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.018481819882446e-06,
+      "loss": 0.807,
+      "step": 2054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.009238072211187e-06,
+      "loss": 0.7461,
+      "step": 2055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.6753,
+      "step": 2056
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.990767613754358e-06,
+      "loss": 0.8889,
+      "step": 2057
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.981540923973263e-06,
+      "loss": 0.7549,
+      "step": 2058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.972319941149245e-06,
+      "loss": 0.745,
+      "step": 2059
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.963104675768345e-06,
+      "loss": 0.6283,
+      "step": 2060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.953895138310091e-06,
+      "loss": 0.6988,
+      "step": 2061
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.944691339247515e-06,
+      "loss": 0.7217,
+      "step": 2062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9354932890471096e-06,
+      "loss": 0.654,
+      "step": 2063
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.926300998168835e-06,
+      "loss": 0.7342,
+      "step": 2064
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.917114477066104e-06,
+      "loss": 0.6831,
+      "step": 2065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.907933736185757e-06,
+      "loss": 0.5879,
+      "step": 2066
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.898758785968075e-06,
+      "loss": 0.8035,
+      "step": 2067
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.889589636846747e-06,
+      "loss": 0.7861,
+      "step": 2068
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.880426299248867e-06,
+      "loss": 0.7847,
+      "step": 2069
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.871268783594922e-06,
+      "loss": 0.6659,
+      "step": 2070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.86211710029877e-06,
+      "loss": 0.7981,
+      "step": 2071
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.852971259767642e-06,
+      "loss": 0.6935,
+      "step": 2072
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.843831272402126e-06,
+      "loss": 0.7619,
+      "step": 2073
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.83469714859616e-06,
+      "loss": 0.6715,
+      "step": 2074
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.825568898736999e-06,
+      "loss": 0.7134,
+      "step": 2075
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.816446533205225e-06,
+      "loss": 0.741,
+      "step": 2076
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.807330062374733e-06,
+      "loss": 0.7401,
+      "step": 2077
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.79821949661271e-06,
+      "loss": 0.7026,
+      "step": 2078
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.789114846279631e-06,
+      "loss": 0.6982,
+      "step": 2079
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.780016121729236e-06,
+      "loss": 0.7182,
+      "step": 2080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7709233333085345e-06,
+      "loss": 0.7178,
+      "step": 2081
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.761836491357782e-06,
+      "loss": 0.7042,
+      "step": 2082
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7527556062104756e-06,
+      "loss": 0.6837,
+      "step": 2083
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7436806881933304e-06,
+      "loss": 0.7406,
+      "step": 2084
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.734611747626285e-06,
+      "loss": 0.7689,
+      "step": 2085
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7255487948224754e-06,
+      "loss": 0.7852,
+      "step": 2086
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.716491840088227e-06,
+      "loss": 0.7435,
+      "step": 2087
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.707440893723052e-06,
+      "loss": 0.6964,
+      "step": 2088
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.698395966019615e-06,
+      "loss": 0.7188,
+      "step": 2089
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.689357067263751e-06,
+      "loss": 0.7081,
+      "step": 2090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.680324207734434e-06,
+      "loss": 0.813,
+      "step": 2091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.671297397703766e-06,
+      "loss": 0.7185,
+      "step": 2092
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.662276647436977e-06,
+      "loss": 0.5856,
+      "step": 2093
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.653261967192398e-06,
+      "loss": 0.6918,
+      "step": 2094
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.644253367221467e-06,
+      "loss": 0.7439,
+      "step": 2095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635250857768696e-06,
+      "loss": 0.7696,
+      "step": 2096
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.62625444907168e-06,
+      "loss": 0.7711,
+      "step": 2097
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6172641513610765e-06,
+      "loss": 0.771,
+      "step": 2098
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.608279974860581e-06,
+      "loss": 0.8323,
+      "step": 2099
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.599301929786943e-06,
+      "loss": 0.6977,
+      "step": 2100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5903300263499315e-06,
+      "loss": 0.7317,
+      "step": 2101
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.581364274752338e-06,
+      "loss": 0.7067,
+      "step": 2102
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.572404685189946e-06,
+      "loss": 0.6092,
+      "step": 2103
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.563451267851542e-06,
+      "loss": 0.6164,
+      "step": 2104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.554504032918888e-06,
+      "loss": 0.7474,
+      "step": 2105
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.545562990566725e-06,
+      "loss": 0.6677,
+      "step": 2106
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536628150962744e-06,
+      "loss": 0.6281,
+      "step": 2107
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.527699524267576e-06,
+      "loss": 0.6745,
+      "step": 2108
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.518777120634798e-06,
+      "loss": 0.7419,
+      "step": 2109
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.509860950210905e-06,
+      "loss": 0.6872,
+      "step": 2110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.500951023135308e-06,
+      "loss": 0.6998,
+      "step": 2111
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.492047349540308e-06,
+      "loss": 0.6863,
+      "step": 2112
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.483149939551105e-06,
+      "loss": 0.6969,
+      "step": 2113
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.474258803285774e-06,
+      "loss": 0.7998,
+      "step": 2114
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.465373950855249e-06,
+      "loss": 0.6863,
+      "step": 2115
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4564953923633295e-06,
+      "loss": 0.6583,
+      "step": 2116
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4476231379066445e-06,
+      "loss": 0.6266,
+      "step": 2117
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4387571975746655e-06,
+      "loss": 0.7109,
+      "step": 2118
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.429897581449677e-06,
+      "loss": 0.7029,
+      "step": 2119
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4210442996067724e-06,
+      "loss": 0.7064,
+      "step": 2120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.412197362113848e-06,
+      "loss": 0.7868,
+      "step": 2121
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.403356779031572e-06,
+      "loss": 0.7483,
+      "step": 2122
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.394522560413399e-06,
+      "loss": 0.7122,
+      "step": 2123
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.385694716305541e-06,
+      "loss": 0.6946,
+      "step": 2124
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.376873256746962e-06,
+      "loss": 0.7781,
+      "step": 2125
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.368058191769363e-06,
+      "loss": 0.7361,
+      "step": 2126
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.359249531397177e-06,
+      "loss": 0.6632,
+      "step": 2127
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.350447285647552e-06,
+      "loss": 0.6804,
+      "step": 2128
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.341651464530341e-06,
+      "loss": 0.7049,
+      "step": 2129
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.332862078048095e-06,
+      "loss": 0.7433,
+      "step": 2130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.324079136196034e-06,
+      "loss": 0.7752,
+      "step": 2131
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.315302648962066e-06,
+      "loss": 0.7045,
+      "step": 2132
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30653262632675e-06,
+      "loss": 0.7195,
+      "step": 2133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.297769078263296e-06,
+      "loss": 0.8205,
+      "step": 2134
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2890120147375545e-06,
+      "loss": 0.7518,
+      "step": 2135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.280261445707985e-06,
+      "loss": 0.7425,
+      "step": 2136
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.271517381125687e-06,
+      "loss": 0.73,
+      "step": 2137
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.262779830934346e-06,
+      "loss": 0.7006,
+      "step": 2138
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.254048805070248e-06,
+      "loss": 0.6843,
+      "step": 2139
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2453243134622465e-06,
+      "loss": 0.6841,
+      "step": 2140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2366063660317805e-06,
+      "loss": 0.6899,
+      "step": 2141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2278949726928354e-06,
+      "loss": 0.7306,
+      "step": 2142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.219190143351951e-06,
+      "loss": 0.6526,
+      "step": 2143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210491887908201e-06,
+      "loss": 0.7064,
+      "step": 2144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.201800216253176e-06,
+      "loss": 0.7287,
+      "step": 2145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.193115138270986e-06,
+      "loss": 0.7231,
+      "step": 2146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.184436663838245e-06,
+      "loss": 0.6422,
+      "step": 2147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.175764802824049e-06,
+      "loss": 0.7402,
+      "step": 2148
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.167099565089989e-06,
+      "loss": 0.7965,
+      "step": 2149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.158440960490104e-06,
+      "loss": 0.6677,
+      "step": 2150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.149788998870902e-06,
+      "loss": 0.7098,
+      "step": 2151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.141143690071335e-06,
+      "loss": 0.8215,
+      "step": 2152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.132505043922792e-06,
+      "loss": 0.7118,
+      "step": 2153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.123873070249074e-06,
+      "loss": 0.6459,
+      "step": 2154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.115247778866407e-06,
+      "loss": 0.7414,
+      "step": 2155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.106629179583412e-06,
+      "loss": 0.7304,
+      "step": 2156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0980172822010985e-06,
+      "loss": 0.6996,
+      "step": 2157
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0894120965128615e-06,
+      "loss": 0.7429,
+      "step": 2158
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.080813632304455e-06,
+      "loss": 0.6918,
+      "step": 2159
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.072221899353995e-06,
+      "loss": 0.7265,
+      "step": 2160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.063636907431942e-06,
+      "loss": 0.6741,
+      "step": 2161
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.055058666301087e-06,
+      "loss": 0.7794,
+      "step": 2162
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.046487185716552e-06,
+      "loss": 0.7572,
+      "step": 2163
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.037922475425759e-06,
+      "loss": 0.6487,
+      "step": 2164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.029364545168438e-06,
+      "loss": 0.6464,
+      "step": 2165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.020813404676612e-06,
+      "loss": 0.7904,
+      "step": 2166
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.012269063674581e-06,
+      "loss": 0.663,
+      "step": 2167
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0037315318789e-06,
+      "loss": 0.7922,
+      "step": 2168
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9952008189984025e-06,
+      "loss": 0.6516,
+      "step": 2169
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.986676934734151e-06,
+      "loss": 0.7492,
+      "step": 2170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.978159888779451e-06,
+      "loss": 0.6673,
+      "step": 2171
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.96964969081983e-06,
+      "loss": 0.6747,
+      "step": 2172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.961146350533021e-06,
+      "loss": 0.5325,
+      "step": 2173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9526498775889645e-06,
+      "loss": 0.7031,
+      "step": 2174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.944160281649793e-06,
+      "loss": 0.6815,
+      "step": 2175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.935677572369817e-06,
+      "loss": 0.6953,
+      "step": 2176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.927201759395518e-06,
+      "loss": 0.7131,
+      "step": 2177
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.918732852365524e-06,
+      "loss": 0.6739,
+      "step": 2178
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.910270860910623e-06,
+      "loss": 0.6494,
+      "step": 2179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.901815794653729e-06,
+      "loss": 0.6589,
+      "step": 2180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.893367663209896e-06,
+      "loss": 0.6446,
+      "step": 2181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.884926476186269e-06,
+      "loss": 0.6824,
+      "step": 2182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.876492243182113e-06,
+      "loss": 0.7825,
+      "step": 2183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8680649737887784e-06,
+      "loss": 0.6295,
+      "step": 2184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.859644677589699e-06,
+      "loss": 0.5602,
+      "step": 2185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.851231364160379e-06,
+      "loss": 0.6399,
+      "step": 2186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.842825043068377e-06,
+      "loss": 0.7782,
+      "step": 2187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.834425723873303e-06,
+      "loss": 0.6725,
+      "step": 2188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.826033416126807e-06,
+      "loss": 0.6962,
+      "step": 2189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.817648129372563e-06,
+      "loss": 0.7005,
+      "step": 2190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.809269873146263e-06,
+      "loss": 0.7011,
+      "step": 2191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.800898656975599e-06,
+      "loss": 0.658,
+      "step": 2192
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.792534490380263e-06,
+      "loss": 0.7342,
+      "step": 2193
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7841773828719277e-06,
+      "loss": 0.7488,
+      "step": 2194
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.775827343954239e-06,
+      "loss": 0.6383,
+      "step": 2195
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.767484383122799e-06,
+      "loss": 0.6609,
+      "step": 2196
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.759148509865168e-06,
+      "loss": 0.6105,
+      "step": 2197
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.750819733660844e-06,
+      "loss": 0.7654,
+      "step": 2198
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7424980639812534e-06,
+      "loss": 0.756,
+      "step": 2199
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.734183510289742e-06,
+      "loss": 0.5883,
+      "step": 2200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.725876082041563e-06,
+      "loss": 0.8041,
+      "step": 2201
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.717575788683866e-06,
+      "loss": 0.5943,
+      "step": 2202
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7092826396556882e-06,
+      "loss": 0.7253,
+      "step": 2203
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.700996644387944e-06,
+      "loss": 0.7387,
+      "step": 2204
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6927178123034034e-06,
+      "loss": 0.7315,
+      "step": 2205
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.684446152816702e-06,
+      "loss": 0.801,
+      "step": 2206
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6761816753343114e-06,
+      "loss": 0.7474,
+      "step": 2207
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6679243892545414e-06,
+      "loss": 0.6163,
+      "step": 2208
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.659674303967522e-06,
+      "loss": 0.6888,
+      "step": 2209
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6514314288551877e-06,
+      "loss": 0.7455,
+      "step": 2210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6431957732912793e-06,
+      "loss": 0.7855,
+      "step": 2211
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.634967346641334e-06,
+      "loss": 0.7015,
+      "step": 2212
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6267461582626584e-06,
+      "loss": 0.7168,
+      "step": 2213
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6185322175043336e-06,
+      "loss": 0.7709,
+      "step": 2214
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6103255337071918e-06,
+      "loss": 0.7098,
+      "step": 2215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.602126116203819e-06,
+      "loss": 0.717,
+      "step": 2216
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5939339743185376e-06,
+      "loss": 0.6801,
+      "step": 2217
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5857491173673976e-06,
+      "loss": 0.6834,
+      "step": 2218
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5775715546581546e-06,
+      "loss": 0.7986,
+      "step": 2219
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5694012954902803e-06,
+      "loss": 0.8078,
+      "step": 2220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5612383491549384e-06,
+      "loss": 0.7479,
+      "step": 2221
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.553082724934973e-06,
+      "loss": 0.6753,
+      "step": 2222
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.544934432104905e-06,
+      "loss": 0.7859,
+      "step": 2223
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5367934799309167e-06,
+      "loss": 0.7151,
+      "step": 2224
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5286598776708436e-06,
+      "loss": 0.7203,
+      "step": 2225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5205336345741603e-06,
+      "loss": 0.7241,
+      "step": 2226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.512414759881976e-06,
+      "loss": 0.719,
+      "step": 2227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5043032628270225e-06,
+      "loss": 0.6583,
+      "step": 2228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.49619915263363e-06,
+      "loss": 0.8169,
+      "step": 2229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.488102438517742e-06,
+      "loss": 0.6847,
+      "step": 2230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4800131296868842e-06,
+      "loss": 0.5998,
+      "step": 2231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4719312353401647e-06,
+      "loss": 0.7476,
+      "step": 2232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4638567646682576e-06,
+      "loss": 0.753,
+      "step": 2233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4557897268533935e-06,
+      "loss": 0.6882,
+      "step": 2234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4477301310693544e-06,
+      "loss": 0.756,
+      "step": 2235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4396779864814568e-06,
+      "loss": 0.5769,
+      "step": 2236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4316333022465465e-06,
+      "loss": 0.7097,
+      "step": 2237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.423596087512976e-06,
+      "loss": 0.5898,
+      "step": 2238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.415566351420615e-06,
+      "loss": 0.6184,
+      "step": 2239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.407544103100824e-06,
+      "loss": 0.6735,
+      "step": 2240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.399529351676447e-06,
+      "loss": 0.7342,
+      "step": 2241
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.39152210626181e-06,
+      "loss": 0.5299,
+      "step": 2242
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3835223759626845e-06,
+      "loss": 0.7169,
+      "step": 2243
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3755301698763197e-06,
+      "loss": 0.7386,
+      "step": 2244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.367545497091396e-06,
+      "loss": 0.6406,
+      "step": 2245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3595683666880286e-06,
+      "loss": 0.7105,
+      "step": 2246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3515987877377498e-06,
+      "loss": 0.7014,
+      "step": 2247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.343636769303512e-06,
+      "loss": 0.7209,
+      "step": 2248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3356823204396706e-06,
+      "loss": 0.6988,
+      "step": 2249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3277354501919677e-06,
+      "loss": 0.7941,
+      "step": 2250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.319796167597533e-06,
+      "loss": 0.7455,
+      "step": 2251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3118644816848577e-06,
+      "loss": 0.7725,
+      "step": 2252
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.303940401473804e-06,
+      "loss": 0.6013,
+      "step": 2253
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2960239359755808e-06,
+      "loss": 0.759,
+      "step": 2254
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.288115094192734e-06,
+      "loss": 0.6683,
+      "step": 2255
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2802138851191558e-06,
+      "loss": 0.8061,
+      "step": 2256
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2723203177400352e-06,
+      "loss": 0.7195,
+      "step": 2257
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.264434401031887e-06,
+      "loss": 0.7414,
+      "step": 2258
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2565561439625205e-06,
+      "loss": 0.6787,
+      "step": 2259
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.248685555491039e-06,
+      "loss": 0.8007,
+      "step": 2260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.240822644567815e-06,
+      "loss": 0.7207,
+      "step": 2261
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2329674201344997e-06,
+      "loss": 0.6228,
+      "step": 2262
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2251198911240002e-06,
+      "loss": 0.6543,
+      "step": 2263
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.217280066460472e-06,
+      "loss": 0.8337,
+      "step": 2264
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2094479550593107e-06,
+      "loss": 0.6155,
+      "step": 2265
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.201623565827139e-06,
+      "loss": 0.7811,
+      "step": 2266
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1938069076617983e-06,
+      "loss": 0.6261,
+      "step": 2267
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1859979894523375e-06,
+      "loss": 0.8209,
+      "step": 2268
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1781968200790094e-06,
+      "loss": 0.686,
+      "step": 2269
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.170403408413243e-06,
+      "loss": 0.788,
+      "step": 2270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1626177633176557e-06,
+      "loss": 0.7175,
+      "step": 2271
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.154839893646028e-06,
+      "loss": 0.7192,
+      "step": 2272
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1470698082433014e-06,
+      "loss": 0.7721,
+      "step": 2273
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.139307515945563e-06,
+      "loss": 0.7285,
+      "step": 2274
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1315530255800384e-06,
+      "loss": 0.6929,
+      "step": 2275
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1238063459650802e-06,
+      "loss": 0.6265,
+      "step": 2276
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.116067485910159e-06,
+      "loss": 0.6691,
+      "step": 2277
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1083364542158535e-06,
+      "loss": 0.7067,
+      "step": 2278
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1006132596738426e-06,
+      "loss": 0.7177,
+      "step": 2279
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.092897911066883e-06,
+      "loss": 0.661,
+      "step": 2280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0851904171688196e-06,
+      "loss": 0.659,
+      "step": 2281
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0774907867445623e-06,
+      "loss": 0.6698,
+      "step": 2282
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0697990285500802e-06,
+      "loss": 0.698,
+      "step": 2283
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.06211515133238e-06,
+      "loss": 0.5568,
+      "step": 2284
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.054439163829519e-06,
+      "loss": 0.6433,
+      "step": 2285
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0467710747705735e-06,
+      "loss": 0.678,
+      "step": 2286
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0391108928756486e-06,
+      "loss": 0.6623,
+      "step": 2287
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0314586268558486e-06,
+      "loss": 0.6125,
+      "step": 2288
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0238142854132722e-06,
+      "loss": 0.7927,
+      "step": 2289
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0161778772410153e-06,
+      "loss": 0.73,
+      "step": 2290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0085494110231493e-06,
+      "loss": 0.7054,
+      "step": 2291
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0009288954347115e-06,
+      "loss": 0.6187,
+      "step": 2292
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9933163391417042e-06,
+      "loss": 0.7471,
+      "step": 2293
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.985711750801068e-06,
+      "loss": 0.6812,
+      "step": 2294
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.97811513906069e-06,
+      "loss": 0.81,
+      "step": 2295
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9705265125593863e-06,
+      "loss": 0.7013,
+      "step": 2296
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9629458799268897e-06,
+      "loss": 0.6364,
+      "step": 2297
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.955373249783844e-06,
+      "loss": 0.7189,
+      "step": 2298
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.947808630741791e-06,
+      "loss": 0.6564,
+      "step": 2299
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9402520314031646e-06,
+      "loss": 0.6106,
+      "step": 2300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.932703460361277e-06,
+      "loss": 0.7337,
+      "step": 2301
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9251629262003135e-06,
+      "loss": 0.5722,
+      "step": 2302
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9176304374953123e-06,
+      "loss": 0.6796,
+      "step": 2303
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.910106002812172e-06,
+      "loss": 0.7378,
+      "step": 2304
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.902589630707624e-06,
+      "loss": 0.5941,
+      "step": 2305
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.895081329729239e-06,
+      "loss": 0.6849,
+      "step": 2306
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8875811084154025e-06,
+      "loss": 0.7581,
+      "step": 2307
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8800889752953156e-06,
+      "loss": 0.6943,
+      "step": 2308
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.872604938888981e-06,
+      "loss": 0.655,
+      "step": 2309
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8651290077071914e-06,
+      "loss": 0.6559,
+      "step": 2310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8576611902515283e-06,
+      "loss": 0.6779,
+      "step": 2311
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8502014950143376e-06,
+      "loss": 0.6959,
+      "step": 2312
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.842749930478734e-06,
+      "loss": 0.6175,
+      "step": 2313
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8353065051185867e-06,
+      "loss": 0.7628,
+      "step": 2314
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.827871227398508e-06,
+      "loss": 0.7496,
+      "step": 2315
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.820444105773845e-06,
+      "loss": 0.6525,
+      "step": 2316
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8130251486906666e-06,
+      "loss": 0.7134,
+      "step": 2317
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.805614364585758e-06,
+      "loss": 0.7566,
+      "step": 2318
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7982117618866177e-06,
+      "loss": 0.7241,
+      "step": 2319
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.790817349011432e-06,
+      "loss": 0.6949,
+      "step": 2320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.783431134369079e-06,
+      "loss": 0.6533,
+      "step": 2321
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.776053126359105e-06,
+      "loss": 0.6817,
+      "step": 2322
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7686833333717334e-06,
+      "loss": 0.6128,
+      "step": 2323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.761321763787841e-06,
+      "loss": 0.6683,
+      "step": 2324
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.753968425978958e-06,
+      "loss": 0.6115,
+      "step": 2325
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.746623328307243e-06,
+      "loss": 0.6595,
+      "step": 2326
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.739286479125494e-06,
+      "loss": 0.7621,
+      "step": 2327
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7319578867771237e-06,
+      "loss": 0.6654,
+      "step": 2328
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7246375595961584e-06,
+      "loss": 0.6504,
+      "step": 2329
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7173255059072234e-06,
+      "loss": 0.6322,
+      "step": 2330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7100217340255352e-06,
+      "loss": 0.6879,
+      "step": 2331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7027262522568944e-06,
+      "loss": 0.736,
+      "step": 2332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6954390688976706e-06,
+      "loss": 0.6613,
+      "step": 2333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6881601922348e-06,
+      "loss": 0.729,
+      "step": 2334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6808896305457733e-06,
+      "loss": 0.5926,
+      "step": 2335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6736273920986166e-06,
+      "loss": 0.6028,
+      "step": 2336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6663734851519017e-06,
+      "loss": 0.66,
+      "step": 2337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.65912791795472e-06,
+      "loss": 0.6421,
+      "step": 2338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.651890698746682e-06,
+      "loss": 0.7681,
+      "step": 2339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6446618357579035e-06,
+      "loss": 0.5866,
+      "step": 2340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6374413372089968e-06,
+      "loss": 0.637,
+      "step": 2341
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.630229211311064e-06,
+      "loss": 0.6654,
+      "step": 2342
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.623025466265684e-06,
+      "loss": 0.5687,
+      "step": 2343
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6158301102649085e-06,
+      "loss": 0.6714,
+      "step": 2344
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6086431514912438e-06,
+      "loss": 0.7063,
+      "step": 2345
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6014645981176488e-06,
+      "loss": 0.7165,
+      "step": 2346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.594294458307528e-06,
+      "loss": 0.6265,
+      "step": 2347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5871327402147173e-06,
+      "loss": 0.6936,
+      "step": 2348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5799794519834632e-06,
+      "loss": 0.5665,
+      "step": 2349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5728346017484453e-06,
+      "loss": 0.7499,
+      "step": 2350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565698197634735e-06,
+      "loss": 0.6924,
+      "step": 2351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5585702477578024e-06,
+      "loss": 0.7012,
+      "step": 2352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5514507602235063e-06,
+      "loss": 0.7002,
+      "step": 2353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5443397431280703e-06,
+      "loss": 0.6393,
+      "step": 2354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5372372045581006e-06,
+      "loss": 0.6277,
+      "step": 2355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.530143152590553e-06,
+      "loss": 0.7114,
+      "step": 2356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.523057595292734e-06,
+      "loss": 0.6846,
+      "step": 2357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.515980540722295e-06,
+      "loss": 0.6611,
+      "step": 2358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.508911996927207e-06,
+      "loss": 0.5545,
+      "step": 2359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5018519719457725e-06,
+      "loss": 0.6492,
+      "step": 2360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4948004738066013e-06,
+      "loss": 0.582,
+      "step": 2361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4877575105286146e-06,
+      "loss": 0.6536,
+      "step": 2362
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4807230901210155e-06,
+      "loss": 0.5477,
+      "step": 2363
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4736972205832997e-06,
+      "loss": 0.7073,
+      "step": 2364
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4666799099052384e-06,
+      "loss": 0.6569,
+      "step": 2365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.459671166066869e-06,
+      "loss": 0.6961,
+      "step": 2366
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.45267099703849e-06,
+      "loss": 0.7883,
+      "step": 2367
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.445679410780638e-06,
+      "loss": 0.7122,
+      "step": 2368
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4386964152441016e-06,
+      "loss": 0.8225,
+      "step": 2369
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.431722018369894e-06,
+      "loss": 0.6341,
+      "step": 2370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.424756228089251e-06,
+      "loss": 0.5397,
+      "step": 2371
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4177990523236216e-06,
+      "loss": 0.6188,
+      "step": 2372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410850498984659e-06,
+      "loss": 0.7613,
+      "step": 2373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4039105759742098e-06,
+      "loss": 0.6933,
+      "step": 2374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.396979291184306e-06,
+      "loss": 0.6117,
+      "step": 2375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3900566524971593e-06,
+      "loss": 0.6098,
+      "step": 2376
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3831426677851423e-06,
+      "loss": 0.7346,
+      "step": 2377
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3762373449107935e-06,
+      "loss": 0.6459,
+      "step": 2378
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.369340691726798e-06,
+      "loss": 0.7263,
+      "step": 2379
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3624527160759812e-06,
+      "loss": 0.6671,
+      "step": 2380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3555734257913045e-06,
+      "loss": 0.6921,
+      "step": 2381
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3487028286958458e-06,
+      "loss": 0.603,
+      "step": 2382
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3418409326028037e-06,
+      "loss": 0.6765,
+      "step": 2383
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.334987745315478e-06,
+      "loss": 0.7422,
+      "step": 2384
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.328143274627266e-06,
+      "loss": 0.825,
+      "step": 2385
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3213075283216547e-06,
+      "loss": 0.646,
+      "step": 2386
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3144805141722038e-06,
+      "loss": 0.7573,
+      "step": 2387
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3076622399425485e-06,
+      "loss": 0.7257,
+      "step": 2388
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3008527133863836e-06,
+      "loss": 0.7305,
+      "step": 2389
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.294051942247457e-06,
+      "loss": 0.7919,
+      "step": 2390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2872599342595557e-06,
+      "loss": 0.6848,
+      "step": 2391
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2804766971465054e-06,
+      "loss": 0.571,
+      "step": 2392
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2737022386221552e-06,
+      "loss": 0.6643,
+      "step": 2393
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2669365663903764e-06,
+      "loss": 0.6848,
+      "step": 2394
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2601796881450468e-06,
+      "loss": 0.6377,
+      "step": 2395
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.253431611570035e-06,
+      "loss": 0.6544,
+      "step": 2396
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.24669234433921e-06,
+      "loss": 0.8275,
+      "step": 2397
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2399618941164202e-06,
+      "loss": 0.7129,
+      "step": 2398
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2332402685554876e-06,
+      "loss": 0.753,
+      "step": 2399
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2265274753002008e-06,
+      "loss": 0.5883,
+      "step": 2400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2198235219842967e-06,
+      "loss": 0.6828,
+      "step": 2401
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.213128416231468e-06,
+      "loss": 0.6688,
+      "step": 2402
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2064421656553415e-06,
+      "loss": 0.6618,
+      "step": 2403
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.199764777859477e-06,
+      "loss": 0.7577,
+      "step": 2404
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1930962604373517e-06,
+      "loss": 0.8434,
+      "step": 2405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1864366209723597e-06,
+      "loss": 0.6372,
+      "step": 2406
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.179785867037797e-06,
+      "loss": 0.7029,
+      "step": 2407
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1731440061968536e-06,
+      "loss": 0.6978,
+      "step": 2408
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.166511046002612e-06,
+      "loss": 0.6433,
+      "step": 2409
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1598869939980225e-06,
+      "loss": 0.6724,
+      "step": 2410
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.153271857715915e-06,
+      "loss": 0.6688,
+      "step": 2411
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1466656446789757e-06,
+      "loss": 0.532,
+      "step": 2412
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.140068362399745e-06,
+      "loss": 0.775,
+      "step": 2413
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.133480018380608e-06,
+      "loss": 0.7806,
+      "step": 2414
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1269006201137833e-06,
+      "loss": 0.6316,
+      "step": 2415
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1203301750813166e-06,
+      "loss": 0.7329,
+      "step": 2416
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1137686907550737e-06,
+      "loss": 0.6746,
+      "step": 2417
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1072161745967313e-06,
+      "loss": 0.6451,
+      "step": 2418
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1006726340577597e-06,
+      "loss": 0.6995,
+      "step": 2419
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.094138076579433e-06,
+      "loss": 0.695,
+      "step": 2420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0876125095928033e-06,
+      "loss": 0.7856,
+      "step": 2421
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0810959405187005e-06,
+      "loss": 0.6353,
+      "step": 2422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0745883767677246e-06,
+      "loss": 0.7336,
+      "step": 2423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.068089825740225e-06,
+      "loss": 0.7266,
+      "step": 2424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0616002948263157e-06,
+      "loss": 0.7532,
+      "step": 2425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0551197914058464e-06,
+      "loss": 0.6806,
+      "step": 2426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0486483228484012e-06,
+      "loss": 0.7049,
+      "step": 2427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0421858965132856e-06,
+      "loss": 0.6314,
+      "step": 2428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.035732519749528e-06,
+      "loss": 0.6876,
+      "step": 2429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0292881998958634e-06,
+      "loss": 0.6627,
+      "step": 2430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.022852944280731e-06,
+      "loss": 0.7102,
+      "step": 2431
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.016426760222259e-06,
+      "loss": 0.6204,
+      "step": 2432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0100096550282543e-06,
+      "loss": 0.5164,
+      "step": 2433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0036016359962084e-06,
+      "loss": 0.7561,
+      "step": 2434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9972027104132763e-06,
+      "loss": 0.7087,
+      "step": 2435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.990812885556269e-06,
+      "loss": 0.7155,
+      "step": 2436
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9844321686916602e-06,
+      "loss": 0.7704,
+      "step": 2437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9780605670755473e-06,
+      "loss": 0.6917,
+      "step": 2438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.971698087953676e-06,
+      "loss": 0.6903,
+      "step": 2439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.965344738561414e-06,
+      "loss": 0.6161,
+      "step": 2440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9590005261237475e-06,
+      "loss": 0.6375,
+      "step": 2441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9526654578552673e-06,
+      "loss": 0.7197,
+      "step": 2442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463395409601713e-06,
+      "loss": 0.6918,
+      "step": 2443
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.940022782632248e-06,
+      "loss": 0.6887,
+      "step": 2444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9337151900548734e-06,
+      "loss": 0.7296,
+      "step": 2445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9274167704009973e-06,
+      "loss": 0.6461,
+      "step": 2446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.921127530833138e-06,
+      "loss": 0.7237,
+      "step": 2447
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9148474785033768e-06,
+      "loss": 0.6848,
+      "step": 2448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.908576620553345e-06,
+      "loss": 0.6442,
+      "step": 2449
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.902314964114219e-06,
+      "loss": 0.7891,
+      "step": 2450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8960625163067158e-06,
+      "loss": 0.6919,
+      "step": 2451
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.889819284241069e-06,
+      "loss": 0.7969,
+      "step": 2452
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8835852750170414e-06,
+      "loss": 0.7172,
+      "step": 2453
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8773604957239067e-06,
+      "loss": 0.6268,
+      "step": 2454
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8711449534404424e-06,
+      "loss": 0.6722,
+      "step": 2455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8649386552349136e-06,
+      "loss": 0.7591,
+      "step": 2456
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8587416081650866e-06,
+      "loss": 0.684,
+      "step": 2457
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8525538192782e-06,
+      "loss": 0.554,
+      "step": 2458
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8463752956109616e-06,
+      "loss": 0.6304,
+      "step": 2459
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8402060441895508e-06,
+      "loss": 0.675,
+      "step": 2460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8340460720295917e-06,
+      "loss": 0.6858,
+      "step": 2461
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.827895386136166e-06,
+      "loss": 0.6739,
+      "step": 2462
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8217539935037897e-06,
+      "loss": 0.7592,
+      "step": 2463
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8156219011164121e-06,
+      "loss": 0.6757,
+      "step": 2464
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8094991159474107e-06,
+      "loss": 0.6849,
+      "step": 2465
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8033856449595689e-06,
+      "loss": 0.608,
+      "step": 2466
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7972814951050855e-06,
+      "loss": 0.8048,
+      "step": 2467
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7911866733255556e-06,
+      "loss": 0.5433,
+      "step": 2468
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.785101186551975e-06,
+      "loss": 0.698,
+      "step": 2469
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7790250417047116e-06,
+      "loss": 0.6862,
+      "step": 2470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.772958245693518e-06,
+      "loss": 0.668,
+      "step": 2471
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7669008054175108e-06,
+      "loss": 0.6286,
+      "step": 2472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7608527277651711e-06,
+      "loss": 0.6147,
+      "step": 2473
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7548140196143338e-06,
+      "loss": 0.6577,
+      "step": 2474
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7487846878321691e-06,
+      "loss": 0.7557,
+      "step": 2475
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7427647392751967e-06,
+      "loss": 0.5973,
+      "step": 2476
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7367541807892586e-06,
+      "loss": 0.5939,
+      "step": 2477
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7307530192095212e-06,
+      "loss": 0.6397,
+      "step": 2478
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7247612613604637e-06,
+      "loss": 0.7555,
+      "step": 2479
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.718778914055873e-06,
+      "loss": 0.714,
+      "step": 2480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.712805984098832e-06,
+      "loss": 0.6922,
+      "step": 2481
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7068424782817162e-06,
+      "loss": 0.8039,
+      "step": 2482
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.700888403386186e-06,
+      "loss": 0.6169,
+      "step": 2483
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.694943766183168e-06,
+      "loss": 0.6418,
+      "step": 2484
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6890085734328666e-06,
+      "loss": 0.7807,
+      "step": 2485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6830828318847415e-06,
+      "loss": 0.6801,
+      "step": 2486
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6771665482775058e-06,
+      "loss": 0.7606,
+      "step": 2487
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6712597293391153e-06,
+      "loss": 0.6118,
+      "step": 2488
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6653623817867638e-06,
+      "loss": 0.6275,
+      "step": 2489
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6594745123268751e-06,
+      "loss": 0.6102,
+      "step": 2490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6535961276550948e-06,
+      "loss": 0.6771,
+      "step": 2491
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6477272344562789e-06,
+      "loss": 0.5991,
+      "step": 2492
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6418678394044962e-06,
+      "loss": 0.6772,
+      "step": 2493
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6360179491630056e-06,
+      "loss": 0.6505,
+      "step": 2494
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6301775703842637e-06,
+      "loss": 0.7815,
+      "step": 2495
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6243467097099097e-06,
+      "loss": 0.614,
+      "step": 2496
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6185253737707607e-06,
+      "loss": 0.6995,
+      "step": 2497
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6127135691867945e-06,
+      "loss": 0.5755,
+      "step": 2498
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6069113025671546e-06,
+      "loss": 0.6053,
+      "step": 2499
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6011185805101447e-06,
+      "loss": 0.6471,
+      "step": 2500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5953354096032048e-06,
+      "loss": 0.5996,
+      "step": 2501
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5895617964229215e-06,
+      "loss": 0.6437,
+      "step": 2502
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5837977475350008e-06,
+      "loss": 0.6752,
+      "step": 2503
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5780432694942815e-06,
+      "loss": 0.8052,
+      "step": 2504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.572298368844719e-06,
+      "loss": 0.6576,
+      "step": 2505
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5665630521193753e-06,
+      "loss": 0.7189,
+      "step": 2506
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5608373258404087e-06,
+      "loss": 0.747,
+      "step": 2507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5551211965190783e-06,
+      "loss": 0.7151,
+      "step": 2508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5494146706557257e-06,
+      "loss": 0.6727,
+      "step": 2509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5437177547397742e-06,
+      "loss": 0.6923,
+      "step": 2510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.538030455249716e-06,
+      "loss": 0.655,
+      "step": 2511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5323527786531113e-06,
+      "loss": 0.5931,
+      "step": 2512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5266847314065726e-06,
+      "loss": 0.6519,
+      "step": 2513
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.521026319955764e-06,
+      "loss": 0.741,
+      "step": 2514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.515377550735394e-06,
+      "loss": 0.6867,
+      "step": 2515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5097384301692041e-06,
+      "loss": 0.6973,
+      "step": 2516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5041089646699603e-06,
+      "loss": 0.5613,
+      "step": 2517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4984891606394535e-06,
+      "loss": 0.6384,
+      "step": 2518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4928790244684866e-06,
+      "loss": 0.7297,
+      "step": 2519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.487278562536868e-06,
+      "loss": 0.663,
+      "step": 2520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4816877812134034e-06,
+      "loss": 0.7282,
+      "step": 2521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4761066868558915e-06,
+      "loss": 0.7148,
+      "step": 2522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4705352858111144e-06,
+      "loss": 0.7311,
+      "step": 2523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4649735844148304e-06,
+      "loss": 0.7404,
+      "step": 2524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4594215889917707e-06,
+      "loss": 0.6626,
+      "step": 2525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4538793058556222e-06,
+      "loss": 0.6253,
+      "step": 2526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4483467413090324e-06,
+      "loss": 0.6016,
+      "step": 2527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4428239016435953e-06,
+      "loss": 0.6774,
+      "step": 2528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4373107931398477e-06,
+      "loss": 0.6316,
+      "step": 2529
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4318074220672595e-06,
+      "loss": 0.6527,
+      "step": 2530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.426313794684221e-06,
+      "loss": 0.659,
+      "step": 2531
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4208299172380512e-06,
+      "loss": 0.6483,
+      "step": 2532
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4153557959649788e-06,
+      "loss": 0.6334,
+      "step": 2533
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4098914370901385e-06,
+      "loss": 0.6306,
+      "step": 2534
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4044368468275549e-06,
+      "loss": 0.7679,
+      "step": 2535
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3989920313801552e-06,
+      "loss": 0.6251,
+      "step": 2536
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3935569969397434e-06,
+      "loss": 0.5506,
+      "step": 2537
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3881317496870062e-06,
+      "loss": 0.672,
+      "step": 2538
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3827162957914975e-06,
+      "loss": 0.5093,
+      "step": 2539
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.37731064141163e-06,
+      "loss": 0.5369,
+      "step": 2540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3719147926946797e-06,
+      "loss": 0.6313,
+      "step": 2541
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3665287557767682e-06,
+      "loss": 0.7615,
+      "step": 2542
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.361152536782856e-06,
+      "loss": 0.7162,
+      "step": 2543
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3557861418267514e-06,
+      "loss": 0.5893,
+      "step": 2544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3504295770110731e-06,
+      "loss": 0.6642,
+      "step": 2545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3450828484272727e-06,
+      "loss": 0.6711,
+      "step": 2546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.6091,
+      "step": 2547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3344189242651672e-06,
+      "loss": 0.7028,
+      "step": 2548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3291017408138007e-06,
+      "loss": 0.7104,
+      "step": 2549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.32379441784818e-06,
+      "loss": 0.6543,
+      "step": 2550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3184969614037567e-06,
+      "loss": 0.6472,
+      "step": 2551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3132093775047616e-06,
+      "loss": 0.8348,
+      "step": 2552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3079316721641987e-06,
+      "loss": 0.7482,
+      "step": 2553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3026638513838386e-06,
+      "loss": 0.588,
+      "step": 2554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.29740592115421e-06,
+      "loss": 0.5869,
+      "step": 2555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.292157887454596e-06,
+      "loss": 0.6731,
+      "step": 2556
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2869197562530223e-06,
+      "loss": 0.7313,
+      "step": 2557
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2816915335062596e-06,
+      "loss": 0.72,
+      "step": 2558
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2764732251598022e-06,
+      "loss": 0.5977,
+      "step": 2559
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2712648371478763e-06,
+      "loss": 0.7444,
+      "step": 2560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2660663753934256e-06,
+      "loss": 0.6558,
+      "step": 2561
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.260877845808105e-06,
+      "loss": 0.6358,
+      "step": 2562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2556992542922708e-06,
+      "loss": 0.7264,
+      "step": 2563
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2505306067349853e-06,
+      "loss": 0.7023,
+      "step": 2564
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2453719090139994e-06,
+      "loss": 0.7608,
+      "step": 2565
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2402231669957465e-06,
+      "loss": 0.6659,
+      "step": 2566
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2350843865353447e-06,
+      "loss": 0.5094,
+      "step": 2567
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2299555734765733e-06,
+      "loss": 0.7642,
+      "step": 2568
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2248367336518874e-06,
+      "loss": 0.6149,
+      "step": 2569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2197278728823947e-06,
+      "loss": 0.7276,
+      "step": 2570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.214628996977857e-06,
+      "loss": 0.6943,
+      "step": 2571
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2095401117366823e-06,
+      "loss": 0.7002,
+      "step": 2572
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.204461222945913e-06,
+      "loss": 0.6412,
+      "step": 2573
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1993923363812288e-06,
+      "loss": 0.6831,
+      "step": 2574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1943334578069277e-06,
+      "loss": 0.8331,
+      "step": 2575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1892845929759412e-06,
+      "loss": 0.712,
+      "step": 2576
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1842457476297953e-06,
+      "loss": 0.7024,
+      "step": 2577
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.179216927498632e-06,
+      "loss": 0.6374,
+      "step": 2578
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1741981383011924e-06,
+      "loss": 0.6785,
+      "step": 2579
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1691893857448089e-06,
+      "loss": 0.7947,
+      "step": 2580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1641906755254007e-06,
+      "loss": 0.8104,
+      "step": 2581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.159202013327464e-06,
+      "loss": 0.671,
+      "step": 2582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.154223404824073e-06,
+      "loss": 0.7137,
+      "step": 2583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1492548556768668e-06,
+      "loss": 0.6039,
+      "step": 2584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1442963715360444e-06,
+      "loss": 0.721,
+      "step": 2585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.139347958040361e-06,
+      "loss": 0.6585,
+      "step": 2586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1344096208171186e-06,
+      "loss": 0.6241,
+      "step": 2587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1294813654821591e-06,
+      "loss": 0.877,
+      "step": 2588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1245631976398608e-06,
+      "loss": 0.6719,
+      "step": 2589
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1196551228831321e-06,
+      "loss": 0.723,
+      "step": 2590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1147571467933983e-06,
+      "loss": 0.6761,
+      "step": 2591
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1098692749406058e-06,
+      "loss": 0.7836,
+      "step": 2592
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.10499151288321e-06,
+      "loss": 0.6073,
+      "step": 2593
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1001238661681657e-06,
+      "loss": 0.5906,
+      "step": 2594
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0952663403309294e-06,
+      "loss": 0.8204,
+      "step": 2595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0904189408954445e-06,
+      "loss": 0.7257,
+      "step": 2596
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.085581673374142e-06,
+      "loss": 0.6876,
+      "step": 2597
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0807545432679278e-06,
+      "loss": 0.6238,
+      "step": 2598
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0759375560661834e-06,
+      "loss": 0.7294,
+      "step": 2599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.07113071724675e-06,
+      "loss": 0.6923,
+      "step": 2600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0663340322759318e-06,
+      "loss": 0.5949,
+      "step": 2601
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.061547506608488e-06,
+      "loss": 0.673,
+      "step": 2602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0567711456876218e-06,
+      "loss": 0.6742,
+      "step": 2603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0520049549449796e-06,
+      "loss": 0.7142,
+      "step": 2604
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.047248939800637e-06,
+      "loss": 0.7914,
+      "step": 2605
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0425031056631007e-06,
+      "loss": 0.6933,
+      "step": 2606
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0377674579293062e-06,
+      "loss": 0.649,
+      "step": 2607
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0330420019845965e-06,
+      "loss": 0.6101,
+      "step": 2608
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.02832674320273e-06,
+      "loss": 0.7183,
+      "step": 2609
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0236216869458626e-06,
+      "loss": 0.7793,
+      "step": 2610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0189268385645524e-06,
+      "loss": 0.6259,
+      "step": 2611
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0142422033977506e-06,
+      "loss": 0.6424,
+      "step": 2612
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0095677867727927e-06,
+      "loss": 0.6323,
+      "step": 2613
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0049035940053897e-06,
+      "loss": 0.6899,
+      "step": 2614
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0002496303996312e-06,
+      "loss": 0.6545,
+      "step": 2615
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.956059012479725e-07,
+      "loss": 0.6734,
+      "step": 2616
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.909724118312304e-07,
+      "loss": 0.589,
+      "step": 2617
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.863491674185776e-07,
+      "loss": 0.7629,
+      "step": 2618
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.817361732675346e-07,
+      "loss": 0.6731,
+      "step": 2619
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.77133434623968e-07,
+      "loss": 0.7219,
+      "step": 2620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.7254095672208e-07,
+      "loss": 0.6475,
+      "step": 2621
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.679587447844052e-07,
+      "loss": 0.7196,
+      "step": 2622
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.633868040218043e-07,
+      "loss": 0.634,
+      "step": 2623
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.588251396334524e-07,
+      "loss": 0.717,
+      "step": 2624
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.542737568068472e-07,
+      "loss": 0.715,
+      "step": 2625
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.497326607177881e-07,
+      "loss": 0.7918,
+      "step": 2626
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.452018565303789e-07,
+      "loss": 0.524,
+      "step": 2627
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.40681349397019e-07,
+      "loss": 0.6232,
+      "step": 2628
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.361711444583965e-07,
+      "loss": 0.6308,
+      "step": 2629
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.316712468434874e-07,
+      "loss": 0.743,
+      "step": 2630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.271816616695439e-07,
+      "loss": 0.7237,
+      "step": 2631
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.22702394042092e-07,
+      "loss": 0.6994,
+      "step": 2632
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.182334490549205e-07,
+      "loss": 0.7045,
+      "step": 2633
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.137748317900852e-07,
+      "loss": 0.5457,
+      "step": 2634
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.093265473178958e-07,
+      "loss": 0.732,
+      "step": 2635
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.048886006969093e-07,
+      "loss": 0.6142,
+      "step": 2636
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.004609969739308e-07,
+      "loss": 0.6379,
+      "step": 2637
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.960437411839939e-07,
+      "loss": 0.6108,
+      "step": 2638
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.916368383503782e-07,
+      "loss": 0.6138,
+      "step": 2639
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.872402934845814e-07,
+      "loss": 0.714,
+      "step": 2640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.828541115863253e-07,
+      "loss": 0.7185,
+      "step": 2641
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.784782976435425e-07,
+      "loss": 0.7271,
+      "step": 2642
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.741128566323797e-07,
+      "loss": 0.7367,
+      "step": 2643
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.697577935171864e-07,
+      "loss": 0.6225,
+      "step": 2644
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.654131132505106e-07,
+      "loss": 0.6469,
+      "step": 2645
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.610788207730958e-07,
+      "loss": 0.7463,
+      "step": 2646
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.567549210138637e-07,
+      "loss": 0.5993,
+      "step": 2647
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.524414188899266e-07,
+      "loss": 0.6791,
+      "step": 2648
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.481383193065684e-07,
+      "loss": 0.7428,
+      "step": 2649
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.438456271572437e-07,
+      "loss": 0.7658,
+      "step": 2650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.395633473235764e-07,
+      "loss": 0.7767,
+      "step": 2651
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.352914846753413e-07,
+      "loss": 0.6969,
+      "step": 2652
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.310300440704732e-07,
+      "loss": 0.5453,
+      "step": 2653
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.267790303550527e-07,
+      "loss": 0.8743,
+      "step": 2654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.225384483633036e-07,
+      "loss": 0.6619,
+      "step": 2655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.183083029175853e-07,
+      "loss": 0.541,
+      "step": 2656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.140885988283898e-07,
+      "loss": 0.8292,
+      "step": 2657
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.098793408943361e-07,
+      "loss": 0.7288,
+      "step": 2658
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.056805339021645e-07,
+      "loss": 0.6667,
+      "step": 2659
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.014921826267286e-07,
+      "loss": 0.7894,
+      "step": 2660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.973142918309928e-07,
+      "loss": 0.7047,
+      "step": 2661
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.931468662660246e-07,
+      "loss": 0.6641,
+      "step": 2662
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.88989910670993e-07,
+      "loss": 0.6732,
+      "step": 2663
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8484342977316e-07,
+      "loss": 0.6306,
+      "step": 2664
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807074282878723e-07,
+      "loss": 0.6054,
+      "step": 2665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.765819109185635e-07,
+      "loss": 0.6509,
+      "step": 2666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724668823567439e-07,
+      "loss": 0.6439,
+      "step": 2667
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.683623472819956e-07,
+      "loss": 0.767,
+      "step": 2668
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.642683103619686e-07,
+      "loss": 0.7687,
+      "step": 2669
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.601847762523718e-07,
+      "loss": 0.6927,
+      "step": 2670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.561117495969717e-07,
+      "loss": 0.671,
+      "step": 2671
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.520492350275876e-07,
+      "loss": 0.6598,
+      "step": 2672
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.479972371640831e-07,
+      "loss": 0.6614,
+      "step": 2673
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.439557606143633e-07,
+      "loss": 0.716,
+      "step": 2674
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.39924809974365e-07,
+      "loss": 0.6041,
+      "step": 2675
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.359043898280571e-07,
+      "loss": 0.6687,
+      "step": 2676
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.318945047474357e-07,
+      "loss": 0.7462,
+      "step": 2677
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.278951592925154e-07,
+      "loss": 0.6986,
+      "step": 2678
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.239063580113203e-07,
+      "loss": 0.5393,
+      "step": 2679
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.19928105439891e-07,
+      "loss": 0.6693,
+      "step": 2680
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.159604061022663e-07,
+      "loss": 0.6553,
+      "step": 2681
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.120032645104891e-07,
+      "loss": 0.6441,
+      "step": 2682
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.080566851645954e-07,
+      "loss": 0.6279,
+      "step": 2683
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.041206725526029e-07,
+      "loss": 0.6176,
+      "step": 2684
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.0019523115052e-07,
+      "loss": 0.7056,
+      "step": 2685
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.962803654223327e-07,
+      "loss": 0.7062,
+      "step": 2686
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.923760798199986e-07,
+      "loss": 0.6981,
+      "step": 2687
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.884823787834471e-07,
+      "loss": 0.7004,
+      "step": 2688
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.845992667405621e-07,
+      "loss": 0.5681,
+      "step": 2689
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.807267481071966e-07,
+      "loss": 0.724,
+      "step": 2690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.768648272871503e-07,
+      "loss": 0.6296,
+      "step": 2691
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.730135086721746e-07,
+      "loss": 0.7343,
+      "step": 2692
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.691727966419603e-07,
+      "loss": 0.564,
+      "step": 2693
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.653426955641396e-07,
+      "loss": 0.6441,
+      "step": 2694
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.615232097942769e-07,
+      "loss": 0.6008,
+      "step": 2695
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.577143436758659e-07,
+      "loss": 0.5842,
+      "step": 2696
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.53916101540324e-07,
+      "loss": 0.7402,
+      "step": 2697
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.501284877069824e-07,
+      "loss": 0.6289,
+      "step": 2698
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.463515064830916e-07,
+      "loss": 0.6598,
+      "step": 2699
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.42585162163808e-07,
+      "loss": 0.5905,
+      "step": 2700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.38829459032193e-07,
+      "loss": 0.7595,
+      "step": 2701
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.350844013592062e-07,
+      "loss": 0.715,
+      "step": 2702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.313499934037015e-07,
+      "loss": 0.7258,
+      "step": 2703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.276262394124222e-07,
+      "loss": 0.5514,
+      "step": 2704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.239131436199941e-07,
+      "loss": 0.6828,
+      "step": 2705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.202107102489274e-07,
+      "loss": 0.6106,
+      "step": 2706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.165189435095997e-07,
+      "loss": 0.6825,
+      "step": 2707
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.12837847600265e-07,
+      "loss": 0.7687,
+      "step": 2708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.091674267070391e-07,
+      "loss": 0.6662,
+      "step": 2709
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.05507685003901e-07,
+      "loss": 0.8031,
+      "step": 2710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.018586266526838e-07,
+      "loss": 0.6603,
+      "step": 2711
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.982202558030692e-07,
+      "loss": 0.7243,
+      "step": 2712
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.945925765925875e-07,
+      "loss": 0.5889,
+      "step": 2713
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.909755931466121e-07,
+      "loss": 0.5792,
+      "step": 2714
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.873693095783529e-07,
+      "loss": 0.706,
+      "step": 2715
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.837737299888502e-07,
+      "loss": 0.6878,
+      "step": 2716
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.80188858466969e-07,
+      "loss": 0.761,
+      "step": 2717
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.766146990894039e-07,
+      "loss": 0.6688,
+      "step": 2718
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.730512559206625e-07,
+      "loss": 0.6699,
+      "step": 2719
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.694985330130698e-07,
+      "loss": 0.6538,
+      "step": 2720
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.659565344067552e-07,
+      "loss": 0.7084,
+      "step": 2721
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.624252641296568e-07,
+      "loss": 0.6618,
+      "step": 2722
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.5890472619751e-07,
+      "loss": 0.6645,
+      "step": 2723
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.553949246138468e-07,
+      "loss": 0.6506,
+      "step": 2724
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.5189586336999e-07,
+      "loss": 0.6687,
+      "step": 2725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.484075464450456e-07,
+      "loss": 0.6963,
+      "step": 2726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.449299778059069e-07,
+      "loss": 0.6859,
+      "step": 2727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.414631614072385e-07,
+      "loss": 0.6964,
+      "step": 2728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.380071011914823e-07,
+      "loss": 0.6917,
+      "step": 2729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.345618010888431e-07,
+      "loss": 0.6154,
+      "step": 2730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.311272650172949e-07,
+      "loss": 0.7428,
+      "step": 2731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.277034968825667e-07,
+      "loss": 0.6999,
+      "step": 2732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.242905005781451e-07,
+      "loss": 0.6909,
+      "step": 2733
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.208882799852666e-07,
+      "loss": 0.7756,
+      "step": 2734
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.174968389729118e-07,
+      "loss": 0.6534,
+      "step": 2735
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.141161813978046e-07,
+      "loss": 0.654,
+      "step": 2736
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.107463111044053e-07,
+      "loss": 0.6402,
+      "step": 2737
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.073872319249073e-07,
+      "loss": 0.6297,
+      "step": 2738
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.04038947679234e-07,
+      "loss": 0.7351,
+      "step": 2739
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.007014621750272e-07,
+      "loss": 0.6511,
+      "step": 2740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.973747792076555e-07,
+      "loss": 0.579,
+      "step": 2741
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.940589025601983e-07,
+      "loss": 0.7071,
+      "step": 2742
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.907538360034502e-07,
+      "loss": 0.6111,
+      "step": 2743
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.874595832959061e-07,
+      "loss": 0.6774,
+      "step": 2744
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.841761481837703e-07,
+      "loss": 0.7553,
+      "step": 2745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.809035344009416e-07,
+      "loss": 0.6495,
+      "step": 2746
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.776417456690152e-07,
+      "loss": 0.7881,
+      "step": 2747
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.74390785697274e-07,
+      "loss": 0.6295,
+      "step": 2748
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7115065818268636e-07,
+      "loss": 0.7305,
+      "step": 2749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6792136680990363e-07,
+      "loss": 0.7171,
+      "step": 2750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6470291525125386e-07,
+      "loss": 0.6919,
+      "step": 2751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.614953071667383e-07,
+      "loss": 0.7825,
+      "step": 2752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.582985462040268e-07,
+      "loss": 0.7472,
+      "step": 2753
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5511263599845145e-07,
+      "loss": 0.6995,
+      "step": 2754
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5193758017301084e-07,
+      "loss": 0.6419,
+      "step": 2755
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4877338233835223e-07,
+      "loss": 0.6327,
+      "step": 2756
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.456200460927862e-07,
+      "loss": 0.6755,
+      "step": 2757
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4247757502225983e-07,
+      "loss": 0.6664,
+      "step": 2758
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3934597270037017e-07,
+      "loss": 0.6987,
+      "step": 2759
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3622524268835507e-07,
+      "loss": 0.6946,
+      "step": 2760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3311538853508697e-07,
+      "loss": 0.7739,
+      "step": 2761
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3001641377707125e-07,
+      "loss": 0.6093,
+      "step": 2762
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2692832193843794e-07,
+      "loss": 0.6907,
+      "step": 2763
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2385111653094667e-07,
+      "loss": 0.7846,
+      "step": 2764
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.207848010539717e-07,
+      "loss": 0.8014,
+      "step": 2765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1772937899450804e-07,
+      "loss": 0.6396,
+      "step": 2766
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1468485382716085e-07,
+      "loss": 0.6481,
+      "step": 2767
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1165122901414055e-07,
+      "loss": 0.742,
+      "step": 2768
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0862850800526655e-07,
+      "loss": 0.7864,
+      "step": 2769
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.056166942379558e-07,
+      "loss": 0.7122,
+      "step": 2770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.026157911372242e-07,
+      "loss": 0.7419,
+      "step": 2771
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9962580211567405e-07,
+      "loss": 0.7055,
+      "step": 2772
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9664673057350223e-07,
+      "loss": 0.7459,
+      "step": 2773
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9367857989848765e-07,
+      "loss": 0.6831,
+      "step": 2774
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.907213534659926e-07,
+      "loss": 0.6474,
+      "step": 2775
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.877750546389536e-07,
+      "loss": 0.7284,
+      "step": 2776
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8483968676788055e-07,
+      "loss": 0.6127,
+      "step": 2777
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.819152531908543e-07,
+      "loss": 0.726,
+      "step": 2778
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.790017572335214e-07,
+      "loss": 0.6844,
+      "step": 2779
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7609920220908813e-07,
+      "loss": 0.5946,
+      "step": 2780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.73207591418322e-07,
+      "loss": 0.6261,
+      "step": 2781
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.703269281495392e-07,
+      "loss": 0.8017,
+      "step": 2782
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.674572156786127e-07,
+      "loss": 0.6984,
+      "step": 2783
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6459845726895984e-07,
+      "loss": 0.6563,
+      "step": 2784
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6175065617154113e-07,
+      "loss": 0.6191,
+      "step": 2785
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5891381562485504e-07,
+      "loss": 0.7208,
+      "step": 2786
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5608793885493766e-07,
+      "loss": 0.6748,
+      "step": 2787
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.53273029075355e-07,
+      "loss": 0.5896,
+      "step": 2788
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.504690894872065e-07,
+      "loss": 0.6115,
+      "step": 2789
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.476761232791115e-07,
+      "loss": 0.84,
+      "step": 2790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4489413362721156e-07,
+      "loss": 0.6417,
+      "step": 2791
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4212312369516496e-07,
+      "loss": 0.6939,
+      "step": 2792
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.393630966341477e-07,
+      "loss": 0.6228,
+      "step": 2793
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3661405558284124e-07,
+      "loss": 0.6105,
+      "step": 2794
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.338760036674382e-07,
+      "loss": 0.6057,
+      "step": 2795
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.311489440016302e-07,
+      "loss": 0.541,
+      "step": 2796
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.284328796866121e-07,
+      "loss": 0.7407,
+      "step": 2797
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.25727813811072e-07,
+      "loss": 0.6918,
+      "step": 2798
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.230337494511937e-07,
+      "loss": 0.6731,
+      "step": 2799
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2035068967064876e-07,
+      "loss": 0.6961,
+      "step": 2800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.176786375205931e-07,
+      "loss": 0.7355,
+      "step": 2801
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.150175960396673e-07,
+      "loss": 0.6905,
+      "step": 2802
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.123675682539884e-07,
+      "loss": 0.6961,
+      "step": 2803
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0972855717715134e-07,
+      "loss": 0.5558,
+      "step": 2804
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0710056581021886e-07,
+      "loss": 0.5556,
+      "step": 2805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0448359714172593e-07,
+      "loss": 0.6529,
+      "step": 2806
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.018776541476709e-07,
+      "loss": 0.6608,
+      "step": 2807
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9928273979151433e-07,
+      "loss": 0.779,
+      "step": 2808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.966988570241747e-07,
+      "loss": 0.6715,
+      "step": 2809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9412600878402696e-07,
+      "loss": 0.56,
+      "step": 2810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.915641979968953e-07,
+      "loss": 0.8199,
+      "step": 2811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8901342757605256e-07,
+      "loss": 0.7911,
+      "step": 2812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8647370042222064e-07,
+      "loss": 0.6145,
+      "step": 2813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.839450194235549e-07,
+      "loss": 0.745,
+      "step": 2814
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8142738745565724e-07,
+      "loss": 0.7071,
+      "step": 2815
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.789208073815608e-07,
+      "loss": 0.7781,
+      "step": 2816
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.764252820517299e-07,
+      "loss": 0.6849,
+      "step": 2817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.739408143040623e-07,
+      "loss": 0.7383,
+      "step": 2818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.714674069638712e-07,
+      "loss": 0.6071,
+      "step": 2819
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6900506284390446e-07,
+      "loss": 0.725,
+      "step": 2820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6655378474432e-07,
+      "loss": 0.6599,
+      "step": 2821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.641135754526958e-07,
+      "loss": 0.6231,
+      "step": 2822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6168443774401976e-07,
+      "loss": 0.6586,
+      "step": 2823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.592663743806911e-07,
+      "loss": 0.8006,
+      "step": 2824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5685938811251476e-07,
+      "loss": 0.6761,
+      "step": 2825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.544634816766989e-07,
+      "loss": 0.6951,
+      "step": 2826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.520786577978518e-07,
+      "loss": 0.6571,
+      "step": 2827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4970491918797855e-07,
+      "loss": 0.6903,
+      "step": 2828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4734226854647883e-07,
+      "loss": 0.7043,
+      "step": 2829
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.449907085601422e-07,
+      "loss": 0.6137,
+      "step": 2830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.426502419031451e-07,
+      "loss": 0.8411,
+      "step": 2831
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4032087123705414e-07,
+      "loss": 0.7115,
+      "step": 2832
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3800259921081037e-07,
+      "loss": 0.5807,
+      "step": 2833
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.35695428460736e-07,
+      "loss": 0.7759,
+      "step": 2834
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3339936161053235e-07,
+      "loss": 0.6828,
+      "step": 2835
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.311144012712707e-07,
+      "loss": 0.6282,
+      "step": 2836
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2884055004139038e-07,
+      "loss": 0.7145,
+      "step": 2837
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.265778105066996e-07,
+      "loss": 0.6468,
+      "step": 2838
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.243261852403711e-07,
+      "loss": 0.6459,
+      "step": 2839
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.220856768029367e-07,
+      "loss": 0.6344,
+      "step": 2840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1985628774228717e-07,
+      "loss": 0.5841,
+      "step": 2841
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1763802059367012e-07,
+      "loss": 0.6143,
+      "step": 2842
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.154308778796821e-07,
+      "loss": 0.5321,
+      "step": 2843
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1323486211026979e-07,
+      "loss": 0.7149,
+      "step": 2844
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1104997578273e-07,
+      "loss": 0.683,
+      "step": 2845
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.088762213816986e-07,
+      "loss": 0.7292,
+      "step": 2846
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0671360137915375e-07,
+      "loss": 0.7957,
+      "step": 2847
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0456211823441153e-07,
+      "loss": 0.6577,
+      "step": 2848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0242177439412592e-07,
+      "loss": 0.684,
+      "step": 2849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0029257229227993e-07,
+      "loss": 0.7029,
+      "step": 2850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9817451435018676e-07,
+      "loss": 0.7119,
+      "step": 2851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9606760297648742e-07,
+      "loss": 0.6767,
+      "step": 2852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9397184056714868e-07,
+      "loss": 0.6691,
+      "step": 2853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9188722950545636e-07,
+      "loss": 0.6608,
+      "step": 2854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8981377216201636e-07,
+      "loss": 0.6504,
+      "step": 2855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.877514708947481e-07,
+      "loss": 0.6051,
+      "step": 2856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8570032804888783e-07,
+      "loss": 0.6595,
+      "step": 2857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.836603459569808e-07,
+      "loss": 0.7658,
+      "step": 2858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.816315269388802e-07,
+      "loss": 0.7564,
+      "step": 2859
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7961387330174606e-07,
+      "loss": 0.5965,
+      "step": 2860
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7760738734003858e-07,
+      "loss": 0.631,
+      "step": 2861
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.756120713355214e-07,
+      "loss": 0.6346,
+      "step": 2862
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7362792755725278e-07,
+      "loss": 0.6879,
+      "step": 2863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7165495826158896e-07,
+      "loss": 0.6362,
+      "step": 2864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.696931656921752e-07,
+      "loss": 0.5911,
+      "step": 2865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6774255207994915e-07,
+      "loss": 0.7731,
+      "step": 2866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6580311964313645e-07,
+      "loss": 0.6895,
+      "step": 2867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6387487058724504e-07,
+      "loss": 0.7009,
+      "step": 2868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6195780710506648e-07,
+      "loss": 0.7321,
+      "step": 2869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.600519313766724e-07,
+      "loss": 0.6184,
+      "step": 2870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5815724556941135e-07,
+      "loss": 0.6653,
+      "step": 2871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.562737518379065e-07,
+      "loss": 0.6153,
+      "step": 2872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5440145232405557e-07,
+      "loss": 0.7641,
+      "step": 2873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.525403491570221e-07,
+      "loss": 0.662,
+      "step": 2874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.506904444532431e-07,
+      "loss": 0.7431,
+      "step": 2875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.488517403164147e-07,
+      "loss": 0.7146,
+      "step": 2876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4702423883749983e-07,
+      "loss": 0.6645,
+      "step": 2877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4520794209472056e-07,
+      "loss": 0.672,
+      "step": 2878
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4340285215355577e-07,
+      "loss": 0.6396,
+      "step": 2879
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4160897106674343e-07,
+      "loss": 0.6892,
+      "step": 2880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3982630087427062e-07,
+      "loss": 0.651,
+      "step": 2881
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3805484360337906e-07,
+      "loss": 0.6363,
+      "step": 2882
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.362946012685573e-07,
+      "loss": 0.5743,
+      "step": 2883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3454557587154083e-07,
+      "loss": 0.6203,
+      "step": 2884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3280776940130858e-07,
+      "loss": 0.7716,
+      "step": 2885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3108118383408418e-07,
+      "loss": 0.7167,
+      "step": 2886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2936582113332707e-07,
+      "loss": 0.6412,
+      "step": 2887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.276616832497346e-07,
+      "loss": 0.6639,
+      "step": 2888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2596877212124327e-07,
+      "loss": 0.7092,
+      "step": 2889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2428708967301973e-07,
+      "loss": 0.6886,
+      "step": 2890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.226166378174598e-07,
+      "loss": 0.7492,
+      "step": 2891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2095741845419174e-07,
+      "loss": 0.5752,
+      "step": 2892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1930943347006618e-07,
+      "loss": 0.6927,
+      "step": 2893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1767268473916182e-07,
+      "loss": 0.7242,
+      "step": 2894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1604717412277866e-07,
+      "loss": 0.6736,
+      "step": 2895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.144329034694347e-07,
+      "loss": 0.727,
+      "step": 2896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1282987461486927e-07,
+      "loss": 0.5983,
+      "step": 2897
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1123808938203306e-07,
+      "loss": 0.7715,
+      "step": 2898
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0965754958109587e-07,
+      "loss": 0.6424,
+      "step": 2899
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0808825700943437e-07,
+      "loss": 0.7114,
+      "step": 2900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0653021345163995e-07,
+      "loss": 0.668,
+      "step": 2901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.049834206795064e-07,
+      "loss": 0.6584,
+      "step": 2902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0344788045203668e-07,
+      "loss": 0.5823,
+      "step": 2903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0192359451543731e-07,
+      "loss": 0.6654,
+      "step": 2904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0041056460311393e-07,
+      "loss": 0.5543,
+      "step": 2905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.890879243567686e-08,
+      "loss": 0.6764,
+      "step": 2906
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.741827972092777e-08,
+      "loss": 0.6297,
+      "step": 2907
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.593902815386968e-08,
+      "loss": 0.6081,
+      "step": 2908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.447103941669588e-08,
+      "loss": 0.6845,
+      "step": 2909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.301431517879323e-08,
+      "loss": 0.6775,
+      "step": 2910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.156885709674101e-08,
+      "loss": 0.6371,
+      "step": 2911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.013466681429994e-08,
+      "loss": 0.6572,
+      "step": 2912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.871174596242427e-08,
+      "loss": 0.6672,
+      "step": 2913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.73000961592485e-08,
+      "loss": 0.6575,
+      "step": 2914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.589971901009186e-08,
+      "loss": 0.6667,
+      "step": 2915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.451061610745381e-08,
+      "loss": 0.6693,
+      "step": 2916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.313278903101075e-08,
+      "loss": 0.6127,
+      "step": 2917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.17662393476204e-08,
+      "loss": 0.6235,
+      "step": 2918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.041096861131303e-08,
+      "loss": 0.5349,
+      "step": 2919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.906697836329247e-08,
+      "loss": 0.8139,
+      "step": 2920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.773427013193502e-08,
+      "loss": 0.7365,
+      "step": 2921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.641284543278838e-08,
+      "loss": 0.7617,
+      "step": 2922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.510270576856605e-08,
+      "loss": 0.6038,
+      "step": 2923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.38038526291518e-08,
+      "loss": 0.7505,
+      "step": 2924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.25162874915919e-08,
+      "loss": 0.7365,
+      "step": 2925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.12400118200951e-08,
+      "loss": 0.6976,
+      "step": 2926
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.997502706603487e-08,
+      "loss": 0.6077,
+      "step": 2927
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.872133466794163e-08,
+      "loss": 0.6821,
+      "step": 2928
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.747893605150824e-08,
+      "loss": 0.6679,
+      "step": 2929
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.624783262958012e-08,
+      "loss": 0.6501,
+      "step": 2930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.50280258021596e-08,
+      "loss": 0.7814,
+      "step": 2931
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.381951695640154e-08,
+      "loss": 0.6804,
+      "step": 2932
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.262230746661657e-08,
+      "loss": 0.5597,
+      "step": 2933
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.143639869426232e-08,
+      "loss": 0.6584,
+      "step": 2934
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.026179198794335e-08,
+      "loss": 0.838,
+      "step": 2935
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9098488683417834e-08,
+      "loss": 0.5844,
+      "step": 2936
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.794649010358422e-08,
+      "loss": 0.836,
+      "step": 2937
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.680579755848681e-08,
+      "loss": 0.7431,
+      "step": 2938
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.5676412345316844e-08,
+      "loss": 0.647,
+      "step": 2939
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.455833574839919e-08,
+      "loss": 0.5488,
+      "step": 2940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3451569039205674e-08,
+      "loss": 0.6635,
+      "step": 2941
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.235611347634173e-08,
+      "loss": 0.6223,
+      "step": 2942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.12719703055542e-08,
+      "loss": 0.6356,
+      "step": 2943
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0199140759721324e-08,
+      "loss": 0.7753,
+      "step": 2944
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.913762605885719e-08,
+      "loss": 0.6892,
+      "step": 2945
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8087427410110635e-08,
+      "loss": 0.6796,
+      "step": 2946
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7048546007759656e-08,
+      "loss": 0.7213,
+      "step": 2947
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.602098303321256e-08,
+      "loss": 0.6885,
+      "step": 2948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5004739655006844e-08,
+      "loss": 0.6352,
+      "step": 2949
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.399981702880918e-08,
+      "loss": 0.6829,
+      "step": 2950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3006216297407645e-08,
+      "loss": 0.666,
+      "step": 2951
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2023938590720625e-08,
+      "loss": 0.7365,
+      "step": 2952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1052985025787914e-08,
+      "loss": 0.6313,
+      "step": 2953
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.009335670676851e-08,
+      "loss": 0.7312,
+      "step": 2954
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.914505472494723e-08,
+      "loss": 0.7549,
+      "step": 2955
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.820808015872701e-08,
+      "loss": 0.6622,
+      "step": 2956
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.728243407362664e-08,
+      "loss": 0.6694,
+      "step": 2957
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.636811752228519e-08,
+      "loss": 0.6484,
+      "step": 2958
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.546513154445874e-08,
+      "loss": 0.7049,
+      "step": 2959
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.457347716701587e-08,
+      "loss": 0.6954,
+      "step": 2960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3693155403938805e-08,
+      "loss": 0.7506,
+      "step": 2961
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.282416725632454e-08,
+      "loss": 0.7205,
+      "step": 2962
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1966513712380356e-08,
+      "loss": 0.6839,
+      "step": 2963
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1120195747422755e-08,
+      "loss": 0.7295,
+      "step": 2964
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0285214323879655e-08,
+      "loss": 0.6969,
+      "step": 2965
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9461570391287053e-08,
+      "loss": 0.5346,
+      "step": 2966
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.864926488628572e-08,
+      "loss": 0.6258,
+      "step": 2967
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.78482987326234e-08,
+      "loss": 0.6871,
+      "step": 2968
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7058672841154822e-08,
+      "loss": 0.7879,
+      "step": 2969
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.628038810983613e-08,
+      "loss": 0.6921,
+      "step": 2970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5513445423728246e-08,
+      "loss": 0.6992,
+      "step": 2971
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4757845654992396e-08,
+      "loss": 0.6271,
+      "step": 2972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4013589662892356e-08,
+      "loss": 0.5533,
+      "step": 2973
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3280678293788883e-08,
+      "loss": 0.6439,
+      "step": 2974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2559112381146386e-08,
+      "loss": 0.6682,
+      "step": 2975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1848892745522932e-08,
+      "loss": 0.6897,
+      "step": 2976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1150020194574682e-08,
+      "loss": 0.555,
+      "step": 2977
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0462495523057013e-08,
+      "loss": 0.6069,
+      "step": 2978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.978631951281673e-08,
+      "loss": 0.6374,
+      "step": 2979
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9121492932796526e-08,
+      "loss": 0.6606,
+      "step": 2980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.846801653903274e-08,
+      "loss": 0.7836,
+      "step": 2981
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7825891074654267e-08,
+      "loss": 0.5954,
+      "step": 2982
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7195117269881433e-08,
+      "loss": 0.7067,
+      "step": 2983
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6575695842027116e-08,
+      "loss": 0.6521,
+      "step": 2984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.59676274954923e-08,
+      "loss": 0.6676,
+      "step": 2985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.537091292176829e-08,
+      "loss": 0.6783,
+      "step": 2986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4785552799433388e-08,
+      "loss": 0.6663,
+      "step": 2987
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4211547794157344e-08,
+      "loss": 0.6096,
+      "step": 2988
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3648898558693558e-08,
+      "loss": 0.6837,
+      "step": 2989
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3097605732882435e-08,
+      "loss": 0.5759,
+      "step": 2990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2557669943650264e-08,
+      "loss": 0.6818,
+      "step": 2991
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2029091805010328e-08,
+      "loss": 0.7021,
+      "step": 2992
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.151187191805625e-08,
+      "loss": 0.6455,
+      "step": 2993
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1006010870966422e-08,
+      "loss": 0.587,
+      "step": 2994
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0511509239004013e-08,
+      "loss": 0.7055,
+      "step": 2995
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0028367584512532e-08,
+      "loss": 0.7165,
+      "step": 2996
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.556586456916928e-09,
+      "loss": 0.7311,
+      "step": 2997
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.096166392723593e-09,
+      "loss": 0.572,
+      "step": 2998
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.647107915519259e-09,
+      "loss": 0.6315,
+      "step": 2999
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.209411535969881e-09,
+      "loss": 0.5715,
+      "step": 3000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.783077751820633e-09,
+      "loss": 0.5621,
+      "step": 3001
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.3681070478948125e-09,
+      "loss": 0.6666,
+      "step": 3002
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.964499896096044e-09,
+      "loss": 0.672,
+      "step": 3003
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.572256755401629e-09,
+      "loss": 0.7563,
+      "step": 3004
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.191378071869203e-09,
+      "loss": 0.7738,
+      "step": 3005
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.821864278631184e-09,
+      "loss": 0.7616,
+      "step": 3006
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463715795895885e-09,
+      "loss": 0.74,
+      "step": 3007
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.116933030946403e-09,
+      "loss": 0.6748,
+      "step": 3008
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.781516378142837e-09,
+      "loss": 0.7536,
+      "step": 3009
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.45746621891785e-09,
+      "loss": 0.6959,
+      "step": 3010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.144782921779999e-09,
+      "loss": 0.6405,
+      "step": 3011
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.843466842310406e-09,
+      "loss": 0.6195,
+      "step": 3012
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5535183231627523e-09,
+      "loss": 0.5632,
+      "step": 3013
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.274937694065505e-09,
+      "loss": 0.7038,
+      "step": 3014
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0077252718185844e-09,
+      "loss": 0.5748,
+      "step": 3015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.751881360292252e-09,
+      "loss": 0.6874,
+      "step": 3016
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.507406250433775e-09,
+      "loss": 0.584,
+      "step": 3017
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.27430022025521e-09,
+      "loss": 0.6519,
+      "step": 3018
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0525635348456195e-09,
+      "loss": 0.6832,
+      "step": 3019
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8421964463610776e-09,
+      "loss": 0.7397,
+      "step": 3020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6431991940302206e-09,
+      "loss": 0.728,
+      "step": 3021
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.455572004150918e-09,
+      "loss": 0.7637,
+      "step": 3022
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.279315090091382e-09,
+      "loss": 0.6522,
+      "step": 3023
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.114428652290167e-09,
+      "loss": 0.7245,
+      "step": 3024
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.609128782550604e-10,
+      "loss": 0.7767,
+      "step": 3025
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.187679425630813e-10,
+      "loss": 0.6937,
+      "step": 3026
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.879940068604818e-10,
+      "loss": 0.6176,
+      "step": 3027
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.685912198627464e-10,
+      "loss": 0.7871,
+      "step": 3028
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.605597173534815e-10,
+      "loss": 0.6383,
+      "step": 3029
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.63899622184416e-10,
+      "loss": 0.6543,
+      "step": 3030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.786110442787315e-10,
+      "loss": 0.6971,
+      "step": 3031
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0469408062440131e-10,
+      "loss": 0.7434,
+      "step": 3032
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4214881527974123e-10,
+      "loss": 0.7192,
+      "step": 3033
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.097531937118931e-11,
+      "loss": 0.6138,
+      "step": 3034
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.1173651093305854e-11,
+      "loss": 0.6704,
+      "step": 3035
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.274385570655291e-11,
+      "loss": 0.7346,
+      "step": 3036
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.685965542845395e-12,
+      "loss": 0.6479,
+      "step": 3037
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.5033,
+      "step": 3038
+    },
+    {
+      "epoch": 2.0,
+      "step": 3038,
+      "total_flos": 148653022003200.0,
+      "train_loss": 1.016740475225166,
+      "train_runtime": 19333.4248,
+      "train_samples_per_second": 5.027,
+      "train_steps_per_second": 0.157
+    }
+  ],
+  "max_steps": 3038,
+  "num_train_epochs": 2,
+  "total_flos": 148653022003200.0,
+  "trial_name": null,
+  "trial_params": null
+}