diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,33289 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999398785546805,
+  "global_step": 5544,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1976047904191619e-06,
+      "loss": 0.1931,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3952095808383237e-06,
+      "loss": 1.3743,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.592814371257485e-06,
+      "loss": 1.3377,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7904191616766475e-06,
+      "loss": 1.2551,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9880239520958085e-06,
+      "loss": 1.2965,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.18562874251497e-06,
+      "loss": 0.2164,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.383233532934131e-06,
+      "loss": 1.3616,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.580838323353295e-06,
+      "loss": 1.2608,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0778443113772455e-05,
+      "loss": 1.1563,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1976047904191617e-05,
+      "loss": 1.1863,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.317365269461078e-05,
+      "loss": 1.1664,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.437125748502994e-05,
+      "loss": 1.1808,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5568862275449103e-05,
+      "loss": 1.1097,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6766467065868263e-05,
+      "loss": 1.1028,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7964071856287426e-05,
+      "loss": 0.9888,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.916167664670659e-05,
+      "loss": 1.0575,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.035928143712575e-05,
+      "loss": 1.0839,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.155688622754491e-05,
+      "loss": 1.0698,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.275449101796407e-05,
+      "loss": 1.0798,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3952095808383234e-05,
+      "loss": 1.0481,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5149700598802394e-05,
+      "loss": 0.9923,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.634730538922156e-05,
+      "loss": 1.0392,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.754491017964072e-05,
+      "loss": 1.0505,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.874251497005988e-05,
+      "loss": 0.9962,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.994011976047904e-05,
+      "loss": 0.2573,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1137724550898205e-05,
+      "loss": 1.0226,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.233532934131737e-05,
+      "loss": 1.0482,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3532934131736525e-05,
+      "loss": 0.9714,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473053892215569e-05,
+      "loss": 1.0218,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.592814371257485e-05,
+      "loss": 0.9491,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.712574850299401e-05,
+      "loss": 1.0131,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.832335329341318e-05,
+      "loss": 0.9702,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9520958083832336e-05,
+      "loss": 1.0166,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.07185628742515e-05,
+      "loss": 0.9833,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.191616766467066e-05,
+      "loss": 0.9282,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.311377245508982e-05,
+      "loss": 0.965,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4311377245508984e-05,
+      "loss": 1.0056,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.550898203592814e-05,
+      "loss": 1.0725,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.670658682634731e-05,
+      "loss": 1.0104,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.790419161676647e-05,
+      "loss": 0.9522,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.910179640718563e-05,
+      "loss": 0.9258,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.029940119760479e-05,
+      "loss": 1.0006,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.149700598802395e-05,
+      "loss": 0.9703,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.269461077844312e-05,
+      "loss": 0.9147,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.389221556886228e-05,
+      "loss": 0.9121,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.508982035928144e-05,
+      "loss": 0.9871,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.62874251497006e-05,
+      "loss": 0.9708,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.748502994011976e-05,
+      "loss": 0.9465,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.868263473053892e-05,
+      "loss": 0.899,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.988023952095808e-05,
+      "loss": 1.002,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.107784431137725e-05,
+      "loss": 1.0416,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.227544910179641e-05,
+      "loss": 0.9716,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.347305389221557e-05,
+      "loss": 0.9345,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.467065868263474e-05,
+      "loss": 0.9486,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.58682634730539e-05,
+      "loss": 0.9368,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.706586826347305e-05,
+      "loss": 0.2048,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.826347305389222e-05,
+      "loss": 0.9669,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.946107784431138e-05,
+      "loss": 0.9514,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.065868263473055e-05,
+      "loss": 0.8638,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.18562874251497e-05,
+      "loss": 0.236,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.305389221556886e-05,
+      "loss": 0.9676,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.425149700598802e-05,
+      "loss": 0.9365,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.544910179640719e-05,
+      "loss": 0.8909,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.664670658682636e-05,
+      "loss": 0.9203,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.784431137724552e-05,
+      "loss": 0.9289,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.904191616766467e-05,
+      "loss": 0.8843,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.023952095808383e-05,
+      "loss": 0.9639,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.1437125748503e-05,
+      "loss": 0.8832,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.263473053892216e-05,
+      "loss": 0.9084,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.383233532934131e-05,
+      "loss": 0.915,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.502994011976048e-05,
+      "loss": 0.9729,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.622754491017964e-05,
+      "loss": 0.9575,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.742514970059881e-05,
+      "loss": 0.9545,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.862275449101797e-05,
+      "loss": 0.8899,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.982035928143712e-05,
+      "loss": 0.9796,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.101796407185628e-05,
+      "loss": 0.8899,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.221556886227547e-05,
+      "loss": 1.0012,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.341317365269462e-05,
+      "loss": 0.2431,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.461077844311378e-05,
+      "loss": 0.8991,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.580838323353294e-05,
+      "loss": 0.9005,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.700598802395209e-05,
+      "loss": 0.896,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.820359281437126e-05,
+      "loss": 0.934,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.940119760479042e-05,
+      "loss": 0.8957,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010059880239520958,
+      "loss": 0.9757,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010179640718562875,
+      "loss": 0.9084,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001029940119760479,
+      "loss": 0.9227,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010419161676646707,
+      "loss": 0.9393,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010538922155688624,
+      "loss": 0.9603,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010658682634730539,
+      "loss": 0.8957,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010778443113772456,
+      "loss": 0.9141,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010898203592814371,
+      "loss": 0.217,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011017964071856288,
+      "loss": 0.9497,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011137724550898203,
+      "loss": 0.9075,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001125748502994012,
+      "loss": 0.8833,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011377245508982037,
+      "loss": 0.9363,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011497005988023953,
+      "loss": 0.8636,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001161676646706587,
+      "loss": 0.8929,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011736526946107784,
+      "loss": 0.9333,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011856287425149701,
+      "loss": 0.9106,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011976047904191617,
+      "loss": 0.2182,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012095808383233534,
+      "loss": 0.8972,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001221556886227545,
+      "loss": 0.9146,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012335329341317365,
+      "loss": 0.8784,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012455089820359282,
+      "loss": 0.9177,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012574850299401196,
+      "loss": 0.9035,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012694610778443113,
+      "loss": 0.9095,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001281437125748503,
+      "loss": 0.9087,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012934131736526947,
+      "loss": 0.9243,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013053892215568865,
+      "loss": 0.9231,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001317365269461078,
+      "loss": 0.8938,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013293413173652696,
+      "loss": 0.8634,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001341317365269461,
+      "loss": 0.2034,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013532934131736527,
+      "loss": 0.9693,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013652694610778444,
+      "loss": 0.8426,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013772455089820359,
+      "loss": 0.9055,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013892215568862276,
+      "loss": 0.9545,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014011976047904193,
+      "loss": 0.9492,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001413173652694611,
+      "loss": 0.9213,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014251497005988024,
+      "loss": 0.9203,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001437125748502994,
+      "loss": 0.9233,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014491017964071858,
+      "loss": 0.8702,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014610778443113772,
+      "loss": 0.9006,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001473053892215569,
+      "loss": 0.9182,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014850299401197604,
+      "loss": 0.9019,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001497005988023952,
+      "loss": 0.9392,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015089820359281438,
+      "loss": 0.9406,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015209580838323355,
+      "loss": 0.9184,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015329341317365272,
+      "loss": 0.9177,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015449101796407186,
+      "loss": 0.8913,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015568862275449103,
+      "loss": 0.8859,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015688622754491018,
+      "loss": 0.9404,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015808383233532935,
+      "loss": 0.9114,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001592814371257485,
+      "loss": 0.8994,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016047904191616766,
+      "loss": 0.9571,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016167664670658683,
+      "loss": 0.9249,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000162874251497006,
+      "loss": 0.2314,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016407185628742517,
+      "loss": 0.8755,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016526946107784431,
+      "loss": 0.9068,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016646706586826348,
+      "loss": 0.2587,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016766467065868263,
+      "loss": 0.8826,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001688622754491018,
+      "loss": 0.9043,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017005988023952097,
+      "loss": 0.888,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001712574850299401,
+      "loss": 0.917,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017245508982035928,
+      "loss": 0.9564,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017365269461077845,
+      "loss": 0.8473,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017485029940119762,
+      "loss": 0.8923,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001760479041916168,
+      "loss": 0.8613,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017724550898203594,
+      "loss": 0.9051,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001784431137724551,
+      "loss": 0.8382,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017964071856287425,
+      "loss": 0.8643,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018083832335329342,
+      "loss": 0.881,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018203592814371256,
+      "loss": 0.85,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018323353293413173,
+      "loss": 0.8809,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018443113772455093,
+      "loss": 0.8555,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018562874251497007,
+      "loss": 0.8924,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018682634730538924,
+      "loss": 0.912,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001880239520958084,
+      "loss": 0.9539,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018922155688622756,
+      "loss": 0.9448,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001904191616766467,
+      "loss": 0.9068,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019161676646706587,
+      "loss": 0.9427,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019281437125748504,
+      "loss": 0.8924,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019401197604790419,
+      "loss": 0.8717,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019520958083832338,
+      "loss": 0.9385,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019640718562874253,
+      "loss": 0.2177,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001976047904191617,
+      "loss": 0.9092,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019880239520958084,
+      "loss": 0.8779,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8766,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999998293172372,
+      "loss": 0.8626,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999993172690065,
+      "loss": 0.8899,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999998463855483,
+      "loss": 0.8819,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999972690769584,
+      "loss": 0.9483,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999957329338398,
+      "loss": 0.8634,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999938554266522,
+      "loss": 0.9004,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999991636556036,
+      "loss": 0.9226,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999989076322749,
+      "loss": 0.9136,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999861747276653,
+      "loss": 0.9101,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999982931771775,
+      "loss": 0.9418,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999793474561854,
+      "loss": 0.8999,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999754217821198,
+      "loss": 0.8755,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999711547509188,
+      "loss": 0.9069,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999665463640382,
+      "loss": 0.8804,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999615966230521,
+      "loss": 0.9633,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999563055296497,
+      "loss": 0.955,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999950673085637,
+      "loss": 0.896,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999944699292937,
+      "loss": 0.8836,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999383841535888,
+      "loss": 0.9281,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999317276697483,
+      "loss": 0.9027,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999247298436879,
+      "loss": 0.8967,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999173906777958,
+      "loss": 0.9162,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999097101745786,
+      "loss": 0.8778,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999016883366565,
+      "loss": 0.8995,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998933251667693,
+      "loss": 0.8478,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998846206677713,
+      "loss": 0.9112,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998755748426336,
+      "loss": 0.9187,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999866187694445,
+      "loss": 0.885,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998564592264088,
+      "loss": 0.9026,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999846389441847,
+      "loss": 0.8924,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998359783441965,
+      "loss": 0.8582,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998252259370117,
+      "loss": 0.2544,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998141322239625,
+      "loss": 0.8827,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998026972088364,
+      "loss": 0.8412,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999790920895537,
+      "loss": 0.919,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997788032880839,
+      "loss": 0.9133,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999766344390614,
+      "loss": 0.9247,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997535442073803,
+      "loss": 0.8843,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999740402742752,
+      "loss": 0.2407,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997269200012154,
+      "loss": 0.817,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999713095987373,
+      "loss": 0.9256,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999698930705944,
+      "loss": 0.9077,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999684424161764,
+      "loss": 0.9138,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996695763597843,
+      "loss": 0.8897,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999654387305074,
+      "loss": 0.8749,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996388570028186,
+      "loss": 0.896,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996229854583186,
+      "loss": 0.8532,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996067726769928,
+      "loss": 0.8968,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995902186643752,
+      "loss": 0.9114,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999573323426117,
+      "loss": 0.8961,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995560869679856,
+      "loss": 0.8684,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999538509295865,
+      "loss": 0.2585,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995205904157554,
+      "loss": 0.8998,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995023303337738,
+      "loss": 0.2061,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994837290561538,
+      "loss": 0.9261,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999464786589245,
+      "loss": 0.9395,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994455029395135,
+      "loss": 0.829,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994258781135424,
+      "loss": 0.8271,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994059121180308,
+      "loss": 0.2896,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993856049597945,
+      "loss": 0.904,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993649566457658,
+      "loss": 0.9824,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999343967182993,
+      "loss": 0.9478,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993226365786412,
+      "loss": 0.8612,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993009648399922,
+      "loss": 0.844,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992789519744438,
+      "loss": 0.8718,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992565979895104,
+      "loss": 0.8795,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992339028928227,
+      "loss": 0.8523,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999210866692129,
+      "loss": 0.8756,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991874893952915,
+      "loss": 0.8749,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991637710102916,
+      "loss": 0.9157,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991397115452254,
+      "loss": 0.8812,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991153110083065,
+      "loss": 0.9141,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990905694078638,
+      "loss": 0.8656,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990654867523436,
+      "loss": 0.888,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990400630503079,
+      "loss": 0.8705,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990142983104358,
+      "loss": 0.8786,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989881925415226,
+      "loss": 0.8547,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989617457524795,
+      "loss": 0.89,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989349579523346,
+      "loss": 0.9447,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989078291502327,
+      "loss": 0.8189,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988803593554345,
+      "loss": 0.902,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998852548577317,
+      "loss": 0.9142,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998824396825374,
+      "loss": 0.8728,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987959041092154,
+      "loss": 0.9018,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987670704385677,
+      "loss": 0.8851,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998737895823274,
+      "loss": 0.9428,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998708380273293,
+      "loss": 0.9249,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986785237987009,
+      "loss": 0.9207,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998648326409689,
+      "loss": 0.9154,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986177881165662,
+      "loss": 0.894,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985869089297572,
+      "loss": 0.2522,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985556888598026,
+      "loss": 0.8439,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199852412791736,
+      "loss": 0.8956,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984922261132037,
+      "loss": 0.8532,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984599834582233,
+      "loss": 0.903,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998427399963426,
+      "loss": 0.8896,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998394475639934,
+      "loss": 0.9326,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983612104989868,
+      "loss": 0.9036,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983276045519398,
+      "loss": 0.91,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982936578102654,
+      "loss": 0.8541,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998259370285551,
+      "loss": 0.9035,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982247419895023,
+      "loss": 0.8819,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998189772933939,
+      "loss": 0.908,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981544631307997,
+      "loss": 0.8985,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981188125921365,
+      "loss": 0.8666,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980828213301204,
+      "loss": 0.859,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998046489357037,
+      "loss": 0.8978,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998009816685289,
+      "loss": 0.873,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979728033273948,
+      "loss": 0.8405,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979354492959898,
+      "loss": 0.8736,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978977546038256,
+      "loss": 0.9216,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978597192637696,
+      "loss": 0.8908,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978213432888056,
+      "loss": 0.8743,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997782626692034,
+      "loss": 0.9,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977435694866716,
+      "loss": 0.9029,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977041716860507,
+      "loss": 0.8854,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976644333036204,
+      "loss": 0.2259,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976243543529468,
+      "loss": 0.8292,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975839348477105,
+      "loss": 0.903,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975431748017096,
+      "loss": 0.8767,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997502074228858,
+      "loss": 0.854,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974606331431865,
+      "loss": 0.895,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974188515588413,
+      "loss": 0.8557,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973767294900857,
+      "loss": 0.8844,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997334266951298,
+      "loss": 0.9369,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997291463956974,
+      "loss": 0.8788,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997248320521725,
+      "loss": 0.2345,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972048366602785,
+      "loss": 0.8902,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971610123874784,
+      "loss": 0.814,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997116847718285,
+      "loss": 0.2435,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970723426677745,
+      "loss": 0.8641,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970274972511395,
+      "loss": 0.9089,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969823114836886,
+      "loss": 0.9083,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969367853808466,
+      "loss": 0.8299,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968909189581547,
+      "loss": 0.9149,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199684471223127,
+      "loss": 0.8684,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019967981652159656,
+      "loss": 0.897,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967512779281313,
+      "loss": 0.9298,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967040503837734,
+      "loss": 0.8335,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966564825990126,
+      "loss": 0.952,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966085745900877,
+      "loss": 0.9317,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996560326373353,
+      "loss": 0.8388,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996511737965278,
+      "loss": 0.8773,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964628093824496,
+      "loss": 0.8696,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964135406415703,
+      "loss": 0.9247,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963639317594587,
+      "loss": 0.8677,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963139827530496,
+      "loss": 0.8663,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996263693639394,
+      "loss": 0.9191,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996213064435658,
+      "loss": 0.8655,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961620951591263,
+      "loss": 0.9456,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961107858271966,
+      "loss": 0.88,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996059136457385,
+      "loss": 0.8054,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960071470673222,
+      "loss": 0.8737,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959548176747558,
+      "loss": 0.8776,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959021482975496,
+      "loss": 0.9124,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958491389536826,
+      "loss": 0.8693,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957957896612504,
+      "loss": 0.2474,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957421004384652,
+      "loss": 0.8767,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995688071303654,
+      "loss": 0.9031,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956337022752609,
+      "loss": 0.8921,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955789933718453,
+      "loss": 0.8993,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955239446120833,
+      "loss": 0.8764,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954685560147662,
+      "loss": 0.8213,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995412827598802,
+      "loss": 0.9241,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953567593832146,
+      "loss": 0.8594,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953003513871432,
+      "loss": 0.8728,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952436036298443,
+      "loss": 0.8628,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951865161306893,
+      "loss": 0.7883,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951290889091657,
+      "loss": 0.8244,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950713219848775,
+      "loss": 0.894,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950132153775442,
+      "loss": 0.8934,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994954769107001,
+      "loss": 0.9233,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948959831932005,
+      "loss": 0.9366,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994836857656209,
+      "loss": 0.8745,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947773925162103,
+      "loss": 0.8654,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947175877935042,
+      "loss": 0.9135,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946574435085056,
+      "loss": 0.8496,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994596959681746,
+      "loss": 0.878,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945361363338715,
+      "loss": 0.8976,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944749734856462,
+      "loss": 0.8759,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944134711579486,
+      "loss": 0.2377,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943516293717734,
+      "loss": 0.8717,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942894481482317,
+      "loss": 0.8975,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942269275085493,
+      "loss": 0.2352,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941640674740692,
+      "loss": 0.9442,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941008680662492,
+      "loss": 0.2335,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940373293066634,
+      "loss": 0.8521,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939734512170024,
+      "loss": 0.8809,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939092338190712,
+      "loss": 0.9157,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938446771347915,
+      "loss": 0.8581,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937797811862015,
+      "loss": 0.8469,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937145459954537,
+      "loss": 0.8556,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936489715848173,
+      "loss": 0.8969,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993583057976677,
+      "loss": 0.8735,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993516805193534,
+      "loss": 0.8896,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993450213258004,
+      "loss": 0.852,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933832821928196,
+      "loss": 0.8033,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933160120208285,
+      "loss": 0.8489,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993248402764995,
+      "loss": 0.9447,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993180454448398,
+      "loss": 0.8689,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931121670942328,
+      "loss": 0.8762,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930435407258104,
+      "loss": 0.8634,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929745753665575,
+      "loss": 0.8596,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929052710400164,
+      "loss": 0.8266,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928356277698454,
+      "loss": 0.8555,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927656455798185,
+      "loss": 0.8041,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926953244938246,
+      "loss": 0.8883,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926246645358693,
+      "loss": 0.8783,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925536657300734,
+      "loss": 0.8787,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924823281006733,
+      "loss": 0.8786,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924106516720216,
+      "loss": 0.8253,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923386364685856,
+      "loss": 0.8226,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922662825149496,
+      "loss": 0.8615,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921935898358116,
+      "loss": 0.8509,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921205584559878,
+      "loss": 0.8646,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920471884004071,
+      "loss": 0.8399,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919734796941166,
+      "loss": 0.8383,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918994323622776,
+      "loss": 0.8811,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991825046430167,
+      "loss": 0.8701,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917503219231782,
+      "loss": 0.9255,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991675258866819,
+      "loss": 0.8786,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915998572867137,
+      "loss": 0.8556,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915241172086012,
+      "loss": 0.8728,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914480386583376,
+      "loss": 0.9044,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913716216618925,
+      "loss": 0.9159,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912948662453525,
+      "loss": 0.8888,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912177724349195,
+      "loss": 0.8604,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199114034025691,
+      "loss": 0.9292,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991062569737757,
+      "loss": 0.8706,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990984460904009,
+      "loss": 0.8333,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990906013782329,
+      "loss": 0.8422,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908272283994968,
+      "loss": 0.884,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907481047824068,
+      "loss": 0.8797,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906686429580687,
+      "loss": 0.851,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905888429536087,
+      "loss": 0.799,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990508704796267,
+      "loss": 0.843,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904282285134007,
+      "loss": 0.8715,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990347414132481,
+      "loss": 0.8954,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902662616810957,
+      "loss": 0.8617,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901847711869473,
+      "loss": 0.8242,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901029426778536,
+      "loss": 0.8854,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900207761817482,
+      "loss": 0.8806,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198993827172668,
+      "loss": 0.8996,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989855429340813,
+      "loss": 0.8974,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989772249052427,
+      "loss": 0.8079,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896887308899164,
+      "loss": 0.8773,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989604874881792,
+      "loss": 0.8999,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019895206810566788,
+      "loss": 0.871,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019894361494433182,
+      "loss": 0.8977,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989351280070566,
+      "loss": 0.8416,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892660729673939,
+      "loss": 0.8186,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891805281628885,
+      "loss": 0.8247,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989094645686252,
+      "loss": 0.8807,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890084255668014,
+      "loss": 0.8415,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889218678339698,
+      "loss": 0.9081,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888349725173047,
+      "loss": 0.8637,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988747739646469,
+      "loss": 0.9079,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886601692512413,
+      "loss": 0.8471,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988572261361515,
+      "loss": 0.8473,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988484016007299,
+      "loss": 0.8009,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883954332187168,
+      "loss": 0.7851,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988306513026008,
+      "loss": 0.9099,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882172554595265,
+      "loss": 0.8755,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881276605497418,
+      "loss": 0.8897,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988037728327239,
+      "loss": 0.8139,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879474588227169,
+      "loss": 0.9024,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878568520669915,
+      "loss": 0.8096,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877659080909922,
+      "loss": 0.8218,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987674626925764,
+      "loss": 0.8654,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019875830086024676,
+      "loss": 0.8429,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987491053152378,
+      "loss": 0.8462,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987398760606886,
+      "loss": 0.8836,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873061309974964,
+      "loss": 0.8244,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872131643558304,
+      "loss": 0.8238,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871198607136236,
+      "loss": 0.8694,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870262201027262,
+      "loss": 0.7736,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986932242555104,
+      "loss": 0.8289,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868379281028382,
+      "loss": 0.9005,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867432767781238,
+      "loss": 0.8918,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866482886132718,
+      "loss": 0.8626,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865529636407082,
+      "loss": 0.8841,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986457301892973,
+      "loss": 0.8403,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863613034027224,
+      "loss": 0.2277,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862649682027267,
+      "loss": 0.8894,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861682963258717,
+      "loss": 0.8535,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860712878051574,
+      "loss": 0.8674,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859739426736993,
+      "loss": 0.8688,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019858762609647278,
+      "loss": 0.8564,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985778242711588,
+      "loss": 0.8933,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856798879477403,
+      "loss": 0.8864,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985581196706759,
+      "loss": 0.8368,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854821690223342,
+      "loss": 0.8503,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019853828049282707,
+      "loss": 0.8995,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852831044584878,
+      "loss": 0.8413,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851830676470198,
+      "loss": 0.8947,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850826945280153,
+      "loss": 0.8999,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849819851357393,
+      "loss": 0.8925,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198488093950457,
+      "loss": 0.8878,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847795576690008,
+      "loss": 0.9015,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198467783966364,
+      "loss": 0.8509,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845757855232104,
+      "loss": 0.8682,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844733952825502,
+      "loss": 0.8373,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843706689766117,
+      "loss": 0.8646,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842676066404618,
+      "loss": 0.8734,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841642083092835,
+      "loss": 0.887,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840604740183721,
+      "loss": 0.8548,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839564038031396,
+      "loss": 0.907,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838519976991118,
+      "loss": 0.8017,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837472557419295,
+      "loss": 0.8475,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019836421779673483,
+      "loss": 0.8558,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835367644112373,
+      "loss": 0.8876,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019834310151095816,
+      "loss": 0.8537,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833249300984802,
+      "loss": 0.8276,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983218509414147,
+      "loss": 0.853,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831117530929102,
+      "loss": 0.2602,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830046611712127,
+      "loss": 0.8666,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828972336856127,
+      "loss": 0.8509,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982789470672781,
+      "loss": 0.8491,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982681372169505,
+      "loss": 0.8703,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825729382126855,
+      "loss": 0.8171,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824641688393383,
+      "loss": 0.8266,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823550640865937,
+      "loss": 0.8447,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982245623991696,
+      "loss": 0.8484,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821358485920044,
+      "loss": 0.8514,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982025737924992,
+      "loss": 0.8085,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819152920282472,
+      "loss": 0.8179,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818045109394725,
+      "loss": 0.8587,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816933946964846,
+      "loss": 0.8718,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981581943337215,
+      "loss": 0.83,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814701568997086,
+      "loss": 0.8837,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813580354221264,
+      "loss": 0.8829,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812455789427422,
+      "loss": 0.8815,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811327874999447,
+      "loss": 0.8897,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810196611322377,
+      "loss": 0.8533,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980906199878238,
+      "loss": 0.8633,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807924037766773,
+      "loss": 0.885,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806782728664022,
+      "loss": 0.8051,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805638071863726,
+      "loss": 0.2598,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804490067756633,
+      "loss": 0.8806,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803338716734635,
+      "loss": 0.8854,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802184019190758,
+      "loss": 0.8472,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801025975519179,
+      "loss": 0.8568,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799864586115214,
+      "loss": 0.8873,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979869985137532,
+      "loss": 0.9092,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019797531771697098,
+      "loss": 0.8031,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796360347479293,
+      "loss": 0.8603,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795185579121784,
+      "loss": 0.8213,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197940074670256,
+      "loss": 0.8317,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792826011592903,
+      "loss": 0.8719,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979164121322701,
+      "loss": 0.888,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790453072332365,
+      "loss": 0.9038,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789261589314553,
+      "loss": 0.8367,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788066764580314,
+      "loss": 0.912,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978686859853752,
+      "loss": 0.8806,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785667091595173,
+      "loss": 0.8385,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019784462244163437,
+      "loss": 0.8111,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197832540566536,
+      "loss": 0.9074,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782042529478103,
+      "loss": 0.8282,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780827663050507,
+      "loss": 0.8306,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019779609457785534,
+      "loss": 0.8653,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778387914099038,
+      "loss": 0.9327,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777163032408005,
+      "loss": 0.8268,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019775934813130575,
+      "loss": 0.8962,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774703256686017,
+      "loss": 0.8711,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977346836349474,
+      "loss": 0.8747,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772230133978296,
+      "loss": 0.9002,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770988568559373,
+      "loss": 0.8451,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197697436676618,
+      "loss": 0.8348,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976849543171054,
+      "loss": 0.816,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019767243861131703,
+      "loss": 0.8171,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765988956352528,
+      "loss": 0.8759,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764730717801397,
+      "loss": 0.8093,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763469145907828,
+      "loss": 0.2435,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976220424110248,
+      "loss": 0.7744,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976093600381715,
+      "loss": 0.8853,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019759664434484763,
+      "loss": 0.8885,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197583895335394,
+      "loss": 0.8395,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757111301416255,
+      "loss": 0.854,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019755829738551687,
+      "loss": 0.8717,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754544845383167,
+      "loss": 0.883,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753256622349313,
+      "loss": 0.8047,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751965069889887,
+      "loss": 0.8166,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750670188445777,
+      "loss": 0.866,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749371978459009,
+      "loss": 0.8704,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019748070440372747,
+      "loss": 0.876,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019746765574631296,
+      "loss": 0.8284,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745457381680086,
+      "loss": 0.8857,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744145861965695,
+      "loss": 0.8774,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742831015935822,
+      "loss": 0.8677,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741512844039322,
+      "loss": 0.8911,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740191346726165,
+      "loss": 0.8483,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019738866524447466,
+      "loss": 0.8745,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019737538377655477,
+      "loss": 0.8157,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736206906803577,
+      "loss": 0.8411,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019734872112346286,
+      "loss": 0.802,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019733533994739256,
+      "loss": 0.8461,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732192554439279,
+      "loss": 0.878,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019730847791904268,
+      "loss": 0.7835,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729499707593284,
+      "loss": 0.9014,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019728148301966517,
+      "loss": 0.8468,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726793575485288,
+      "loss": 0.8371,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972543552861206,
+      "loss": 0.8262,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019724074161810414,
+      "loss": 0.9113,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972270947554508,
+      "loss": 0.776,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972134147028191,
+      "loss": 0.8425,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719970146487897,
+      "loss": 0.8102,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019718595504631168,
+      "loss": 0.8889,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971721754518097,
+      "loss": 0.9111,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715836268607696,
+      "loss": 0.8994,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019714451675382865,
+      "loss": 0.8955,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713063765979127,
+      "loss": 0.7795,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971167254087027,
+      "loss": 0.8656,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710278000531207,
+      "loss": 0.8619,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970888014543799,
+      "loss": 0.8686,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707478976067796,
+      "loss": 0.9179,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019706074492898938,
+      "loss": 0.8274,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704666696410855,
+      "loss": 0.8144,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019703255587084123,
+      "loss": 0.8179,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701841165400443,
+      "loss": 0.8524,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019700423431842653,
+      "loss": 0.8483,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699002386894715,
+      "loss": 0.8344,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969757803104173,
+      "loss": 0.8269,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696150364769918,
+      "loss": 0.9038,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969471938856664,
+      "loss": 0.8461,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019693285102920376,
+      "loss": 0.9213,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019691847508320748,
+      "loss": 0.8737,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019690406605258498,
+      "loss": 0.7841,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019688962394225502,
+      "loss": 0.8294,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968751487571476,
+      "loss": 0.8723,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686064050220413,
+      "loss": 0.2493,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684609918237716,
+      "loss": 0.8575,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968315248026306,
+      "loss": 0.8645,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681691736793964,
+      "loss": 0.8938,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968022768832908,
+      "loss": 0.8457,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967876033536818,
+      "loss": 0.8866,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019677289678412169,
+      "loss": 0.8152,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675815717963072,
+      "loss": 0.2518,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967433845452406,
+      "loss": 0.8004,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019672857888599414,
+      "loss": 0.839,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671374020694546,
+      "loss": 0.8249,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019669886851316,
+      "loss": 0.8751,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019668396380971443,
+      "loss": 0.8497,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019666902610169674,
+      "loss": 0.8086,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665405539420609,
+      "loss": 0.8182,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196639051692353,
+      "loss": 0.8483,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662401500125922,
+      "loss": 0.8275,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019660894532605775,
+      "loss": 0.8249,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019659384267189283,
+      "loss": 0.8855,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019657870704392,
+      "loss": 0.857,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019656353844730608,
+      "loss": 0.8618,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019654833688722908,
+      "loss": 0.8353,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019653310236887825,
+      "loss": 0.84,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965178348974542,
+      "loss": 0.8652,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019650253447816868,
+      "loss": 0.8594,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019648720111624472,
+      "loss": 0.8739,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647183481691661,
+      "loss": 0.8544,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019645643558542989,
+      "loss": 0.8582,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964410034270413,
+      "loss": 0.8931,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019642553834701886,
+      "loss": 0.8538,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641004035064183,
+      "loss": 0.8703,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019639450944320067,
+      "loss": 0.8951,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019637894562999706,
+      "loss": 0.2227,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019636334891634402,
+      "loss": 0.916,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019634771930756572,
+      "loss": 0.8724,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019633205680899755,
+      "loss": 0.8392,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019631636142598613,
+      "loss": 0.8226,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019630063316388935,
+      "loss": 0.2147,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962848720280763,
+      "loss": 0.887,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626907802392722,
+      "loss": 0.8761,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019625325115683374,
+      "loss": 0.8415,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019623739143219857,
+      "loss": 0.835,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622149885543567,
+      "loss": 0.8225,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001962055734319702,
+      "loss": 0.855,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019618961516723853,
+      "loss": 0.8428,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961736240666883,
+      "loss": 0.8662,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019615760013577837,
+      "loss": 0.8363,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019614154337997864,
+      "loss": 0.2546,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612545380477046,
+      "loss": 0.2471,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610933141564612,
+      "loss": 0.9128,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019609317621810938,
+      "loss": 0.7862,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000196076988217675,
+      "loss": 0.9135,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000196060767419869,
+      "loss": 0.8904,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604451383022856,
+      "loss": 0.8321,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960282274543022,
+      "loss": 0.2317,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019601190829764947,
+      "loss": 0.8856,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959955563658412,
+      "loss": 0.8343,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597917166445933,
+      "loss": 0.8999,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019596275419909705,
+      "loss": 0.8743,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959463039753587,
+      "loss": 0.8534,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019592982099885987,
+      "loss": 0.8634,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019591330527522722,
+      "loss": 0.231,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958967568100987,
+      "loss": 0.8496,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019588017560912335,
+      "loss": 0.8423,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019586356167796142,
+      "loss": 0.8792,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019584691502228435,
+      "loss": 0.8689,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019583023564777473,
+      "loss": 0.906,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019581352356012635,
+      "loss": 0.8703,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579677876504408,
+      "loss": 0.8874,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578000126824408,
+      "loss": 0.815,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019576319107545353,
+      "loss": 0.828,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019574634819241094,
+      "loss": 0.8541,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019572947262486582,
+      "loss": 0.2405,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571256437857896,
+      "loss": 0.8285,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019569562345932222,
+      "loss": 0.8095,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019567864987287865,
+      "loss": 0.8544,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566164362504247,
+      "loss": 0.7802,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195644604721619,
+      "loss": 0.8478,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019562753316842472,
+      "loss": 0.9221,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019561042897128732,
+      "loss": 0.8973,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559329213604553,
+      "loss": 0.7988,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955761226685493,
+      "loss": 0.8687,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955589205746597,
+      "loss": 0.9031,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019554168586024894,
+      "loss": 0.8246,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019552441853120035,
+      "loss": 0.8183,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019550711859340835,
+      "loss": 0.9032,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019548978605277864,
+      "loss": 0.837,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019547242091522786,
+      "loss": 0.852,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019545502318668391,
+      "loss": 0.8095,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954375928730858,
+      "loss": 0.832,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954201299803836,
+      "loss": 0.8612,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954026345145385,
+      "loss": 0.8746,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538510648152294,
+      "loss": 0.8544,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019536754588732035,
+      "loss": 0.8352,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019534995273792533,
+      "loss": 0.9039,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953323270393435,
+      "loss": 0.8061,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019531466879759175,
+      "loss": 0.8144,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019529697801869795,
+      "loss": 0.8952,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527925470870115,
+      "loss": 0.8321,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019526149887365144,
+      "loss": 0.836,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019524371051961005,
+      "loss": 0.8691,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952258896526494,
+      "loss": 0.7993,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952080362788528,
+      "loss": 0.8565,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019519015040431482,
+      "loss": 0.8149,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019517223203514112,
+      "loss": 0.2159,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019515428117744836,
+      "loss": 0.8847,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019513629783736441,
+      "loss": 0.8019,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019511828202102807,
+      "loss": 0.9255,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019510023373458938,
+      "loss": 0.8463,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950821529842094,
+      "loss": 0.8236,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019506403977606027,
+      "loss": 0.8707,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504589411632517,
+      "loss": 0.2736,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502771601119848,
+      "loss": 0.881,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019500950546688555,
+      "loss": 0.9384,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019499126248960277,
+      "loss": 0.8691,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949729870855778,
+      "loss": 0.7926,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495467926104908,
+      "loss": 0.8468,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019493633902226636,
+      "loss": 0.8115,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491796637549034,
+      "loss": 0.7699,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019489956132699281,
+      "loss": 0.8615,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948811238830566,
+      "loss": 0.8471,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948626540499757,
+      "loss": 0.8217,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194844151834055,
+      "loss": 0.8248,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948256172416105,
+      "loss": 0.844,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948070502789693,
+      "loss": 0.8253,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019478845095246956,
+      "loss": 0.8332,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947698192684604,
+      "loss": 0.7975,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019475115523330205,
+      "loss": 0.8721,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019473245885336576,
+      "loss": 0.8698,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471373013503384,
+      "loss": 0.8602,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019469496908469962,
+      "loss": 0.8926,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946761757087675,
+      "loss": 0.9012,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019465735001365284,
+      "loss": 0.8329,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019463849200578213,
+      "loss": 0.8289,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019461960169159283,
+      "loss": 0.8197,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019460067907753342,
+      "loss": 0.8884,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019458172417006347,
+      "loss": 0.8597,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456273697565353,
+      "loss": 0.8271,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019454371750078511,
+      "loss": 0.8956,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019452466575195087,
+      "loss": 0.8978,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945055817356544,
+      "loss": 0.8359,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944864654584103,
+      "loss": 0.8997,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019446731692674427,
+      "loss": 0.8063,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944481361471929,
+      "loss": 0.8553,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442892312630385,
+      "loss": 0.8283,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019440967787063583,
+      "loss": 0.7762,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019439040038675843,
+      "loss": 0.8458,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943710906812524,
+      "loss": 0.8655,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435174876070933,
+      "loss": 0.7859,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194332374631732,
+      "loss": 0.8409,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019431296830093393,
+      "loss": 0.8848,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019429352977493988,
+      "loss": 0.2763,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019427405906038544,
+      "loss": 0.8236,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019425455616391723,
+      "loss": 0.8082,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019423502109219292,
+      "loss": 0.8685,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019421545385188104,
+      "loss": 0.8863,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019419585444966123,
+      "loss": 0.8632,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019417622289222405,
+      "loss": 0.9362,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019415655918627103,
+      "loss": 0.8492,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413686333851464,
+      "loss": 0.7891,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019411713535567838,
+      "loss": 0.8803,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019409737524449673,
+      "loss": 0.8409,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940775830117151,
+      "loss": 0.8126,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019405775866408986,
+      "loss": 0.8668,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940379022083884,
+      "loss": 0.7841,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019401801365138895,
+      "loss": 0.8688,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019399809299988085,
+      "loss": 0.848,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939781402606643,
+      "loss": 0.8587,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939581554405505,
+      "loss": 0.8362,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019393813854636153,
+      "loss": 0.7832,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019391808958493055,
+      "loss": 0.8744,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019389800856310147,
+      "loss": 0.8351,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019387789548772934,
+      "loss": 0.8146,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385775036568004,
+      "loss": 0.8322,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019383757320383044,
+      "loss": 0.7945,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019381736400906832,
+      "loss": 0.8627,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937971227882924,
+      "loss": 0.8177,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937768495484123,
+      "loss": 0.2506,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019375654429634867,
+      "loss": 0.8156,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373620703903296,
+      "loss": 0.7989,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019371583778340768,
+      "loss": 0.8559,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019369543653642612,
+      "loss": 0.8627,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936750033050526,
+      "loss": 0.8953,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019365453809626234,
+      "loss": 0.2427,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936340409170414,
+      "loss": 0.8037,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019361351177438687,
+      "loss": 0.8364,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019359295067530665,
+      "loss": 0.8332,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019357235762681963,
+      "loss": 0.8614,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935517326359555,
+      "loss": 0.8206,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193531075709755,
+      "loss": 0.8552,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019351038685526964,
+      "loss": 0.8693,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019348966607956187,
+      "loss": 0.8403,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934689133897051,
+      "loss": 0.7997,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344812879278352,
+      "loss": 0.8556,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019342731229589238,
+      "loss": 0.8674,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340646390613762,
+      "loss": 0.8074,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019338558363063616,
+      "loss": 0.8126,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019336467147651587,
+      "loss": 0.8827,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019334372745091538,
+      "loss": 0.879,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001933227515609843,
+      "loss": 0.9114,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019330174381388304,
+      "loss": 0.8498,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019328070421678292,
+      "loss": 0.8377,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019325963277686618,
+      "loss": 0.8551,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019323852950132585,
+      "loss": 0.8452,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019321739439736587,
+      "loss": 0.8493,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019319622747220104,
+      "loss": 0.8375,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019317502873305697,
+      "loss": 0.8141,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019315379818717025,
+      "loss": 0.857,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019313253584178822,
+      "loss": 0.8707,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311124170416912,
+      "loss": 0.9068,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019308991578158202,
+      "loss": 0.8374,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019306855808130685,
+      "loss": 0.7971,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019304716861063444,
+      "loss": 0.8547,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019302574737686636,
+      "loss": 0.8605,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930042943873151,
+      "loss": 0.8331,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019298280964930403,
+      "loss": 0.8118,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929612931701672,
+      "loss": 0.8155,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019293974495724966,
+      "loss": 0.8678,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019291816501790717,
+      "loss": 0.8918,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019289655335950648,
+      "loss": 0.8051,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019287490998942496,
+      "loss": 0.8112,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019285323491505097,
+      "loss": 0.2436,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928315281437836,
+      "loss": 0.8723,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019280978968303278,
+      "loss": 0.7964,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019278801954021932,
+      "loss": 0.8717,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927662177227748,
+      "loss": 0.884,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274438423814156,
+      "loss": 0.8187,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927225190937728,
+      "loss": 0.8418,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927006222971326,
+      "loss": 0.8338,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926786938556957,
+      "loss": 0.8243,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019265673377694774,
+      "loss": 0.855,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926347420683851,
+      "loss": 0.8515,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019261271873751503,
+      "loss": 0.8191,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019259066379185552,
+      "loss": 0.8061,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925685772389354,
+      "loss": 0.8644,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019254645908629422,
+      "loss": 0.8858,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019252430934148235,
+      "loss": 0.7904,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192502128012061,
+      "loss": 0.8117,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019247991510560204,
+      "loss": 0.806,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245767062968825,
+      "loss": 0.8557,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019243539459191307,
+      "loss": 0.7802,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019241308699988085,
+      "loss": 0.8908,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019239074786120655,
+      "loss": 0.8533,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019236837718351603,
+      "loss": 0.2495,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923459749744459,
+      "loss": 0.8491,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923235412416434,
+      "loss": 0.8124,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923010759927667,
+      "loss": 0.8073,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019227857923548468,
+      "loss": 0.8571,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019225605097747693,
+      "loss": 0.8478,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019223349122643384,
+      "loss": 0.8321,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019221089999005647,
+      "loss": 0.7984,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019218827727605678,
+      "loss": 0.8617,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921656230921573,
+      "loss": 0.8198,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019214293744609147,
+      "loss": 0.8289,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921202203456033,
+      "loss": 0.8059,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019209747179844765,
+      "loss": 0.844,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019207469181239018,
+      "loss": 0.8341,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019205188039520706,
+      "loss": 0.8101,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920290375546854,
+      "loss": 0.7879,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019200616329862293,
+      "loss": 0.8565,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019198325763482817,
+      "loss": 0.827,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019196032057112027,
+      "loss": 0.8264,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019193735211532915,
+      "loss": 0.8089,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019191435227529553,
+      "loss": 0.787,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019189132105887067,
+      "loss": 0.8614,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918682584739167,
+      "loss": 0.8405,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019184516452830637,
+      "loss": 0.8042,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019182203922992312,
+      "loss": 0.8581,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917988825866612,
+      "loss": 0.8935,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019177569460642545,
+      "loss": 0.8352,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917524752971314,
+      "loss": 0.7687,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917292246667054,
+      "loss": 0.837,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019170594272308438,
+      "loss": 0.2744,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191682629474216,
+      "loss": 0.8019,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916592849280586,
+      "loss": 0.8553,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019163590909258117,
+      "loss": 0.8199,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019161250197576347,
+      "loss": 0.7767,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019158906358559587,
+      "loss": 0.2384,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019156559393007937,
+      "loss": 0.8668,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915420930172258,
+      "loss": 0.8334,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019151856085505747,
+      "loss": 0.8526,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914949974516075,
+      "loss": 0.8353,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914714028149196,
+      "loss": 0.8559,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019144777695304817,
+      "loss": 0.8576,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914241198740583,
+      "loss": 0.861,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019140043158602564,
+      "loss": 0.8124,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913767120970366,
+      "loss": 0.8301,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913529614151882,
+      "loss": 0.8352,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019132917954858806,
+      "loss": 0.8838,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019130536650535453,
+      "loss": 0.2234,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019128152229361655,
+      "loss": 0.8511,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019125764692151366,
+      "loss": 0.2412,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019123374039719618,
+      "loss": 0.8638,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120980272882494,
+      "loss": 0.815,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911858339245714,
+      "loss": 0.8386,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019116183399261774,
+      "loss": 0.7883,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019113780294115668,
+      "loss": 0.8248,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019111374077839156,
+      "loss": 0.8812,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019108964751253643,
+      "loss": 0.8521,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019106552315181584,
+      "loss": 0.8295,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019104136770446508,
+      "loss": 0.8614,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019101718117872995,
+      "loss": 0.8431,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019099296358286692,
+      "loss": 0.8521,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019096871492514298,
+      "loss": 0.8249,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019094443521383588,
+      "loss": 0.8607,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909201244572338,
+      "loss": 0.8101,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019089578266363562,
+      "loss": 0.2802,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019087140984135083,
+      "loss": 0.8642,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908470059986994,
+      "loss": 0.8819,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019082257114401202,
+      "loss": 0.8387,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019079810528562982,
+      "loss": 0.7974,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907736084319047,
+      "loss": 0.2467,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190749080591199,
+      "loss": 0.8669,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019072452177188568,
+      "loss": 0.7911,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019069993198234827,
+      "loss": 0.7842,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906753112309809,
+      "loss": 0.8947,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906506595261882,
+      "loss": 0.8664,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019062597687638546,
+      "loss": 0.8631,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019060126328999846,
+      "loss": 0.8839,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019057651877546356,
+      "loss": 0.889,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055174334122772,
+      "loss": 0.85,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905269369957484,
+      "loss": 0.8842,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905020997474936,
+      "loss": 0.827,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047723160494194,
+      "loss": 0.7812,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019045233257658254,
+      "loss": 0.8775,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019042740267091502,
+      "loss": 0.8623,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019040244189644966,
+      "loss": 0.8467,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903774502617072,
+      "loss": 0.2414,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903524277752189,
+      "loss": 0.8344,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019032737444552654,
+      "loss": 0.7936,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019030229028118255,
+      "loss": 0.8772,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019027717529074972,
+      "loss": 0.8615,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902520294828015,
+      "loss": 0.8693,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022685286592176,
+      "loss": 0.8691,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902016454487049,
+      "loss": 0.8482,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019017640723975596,
+      "loss": 0.8197,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019015113824769034,
+      "loss": 0.9017,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190125838481134,
+      "loss": 0.8421,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019010050794872338,
+      "loss": 0.8213,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019007514665910554,
+      "loss": 0.8083,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019004975462093784,
+      "loss": 0.8255,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900243318428883,
+      "loss": 0.8585,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899988783336354,
+      "loss": 0.7889,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018997339410186808,
+      "loss": 0.8058,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018994787915628574,
+      "loss": 0.8699,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018992233350559833,
+      "loss": 0.2462,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018989675715852625,
+      "loss": 0.7614,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018987115012380042,
+      "loss": 0.8307,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018984551241016216,
+      "loss": 0.8222,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898198440263633,
+      "loss": 0.8739,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018979414498116614,
+      "loss": 0.8438,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018976841528334348,
+      "loss": 0.8742,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018974265494167851,
+      "loss": 0.8834,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018971686396496496,
+      "loss": 0.8702,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018969104236200694,
+      "loss": 0.7737,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018966519014161908,
+      "loss": 0.8796,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018963930731262646,
+      "loss": 0.8448,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018961339388386453,
+      "loss": 0.8647,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895874498641793,
+      "loss": 0.818,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018956147526242708,
+      "loss": 0.865,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895354700874748,
+      "loss": 0.8258,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895094343481997,
+      "loss": 0.8665,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018948336805348946,
+      "loss": 0.8354,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018945727121224222,
+      "loss": 0.7975,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018943114383336655,
+      "loss": 0.8719,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940498592578145,
+      "loss": 0.2362,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893787974984163,
+      "loss": 0.8485,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018935257856021097,
+      "loss": 0.8254,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018932632912011562,
+      "loss": 0.8144,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018930004918709097,
+      "loss": 0.8257,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001892737387701081,
+      "loss": 0.8664,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018924739787814842,
+      "loss": 0.7647,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001892210265202038,
+      "loss": 0.7923,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001891946247052766,
+      "loss": 0.8926,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018916819244237942,
+      "loss": 0.8619,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018914172974053532,
+      "loss": 0.8343,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891152366087778,
+      "loss": 0.8248,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018908871305615063,
+      "loss": 0.2534,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890621590917081,
+      "loss": 0.834,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018903557472451477,
+      "loss": 0.8775,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018900895996364566,
+      "loss": 0.8734,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018898231481818617,
+      "loss": 0.826,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018895563929723194,
+      "loss": 0.7944,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018892893340988914,
+      "loss": 0.276,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018890219716527422,
+      "loss": 0.882,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188875430572514,
+      "loss": 0.883,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888486336407457,
+      "loss": 0.904,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018882180637911687,
+      "loss": 0.8361,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018879494879678536,
+      "loss": 0.814,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018876806090291947,
+      "loss": 0.8247,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887411427066978,
+      "loss": 0.8131,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018871419421730926,
+      "loss": 0.8728,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018868721544395317,
+      "loss": 0.8489,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018866020639583917,
+      "loss": 0.8063,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018863316708218718,
+      "loss": 0.8384,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018860609751222748,
+      "loss": 0.862,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885789976952007,
+      "loss": 0.8065,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885518676403578,
+      "loss": 0.8093,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018852470735696003,
+      "loss": 0.8418,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018849751685427897,
+      "loss": 0.7745,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018847029614159656,
+      "loss": 0.8936,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018844304522820497,
+      "loss": 0.2443,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018841576412340674,
+      "loss": 0.8241,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018838845283651468,
+      "loss": 0.8171,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018836111137685192,
+      "loss": 0.886,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018833373975375194,
+      "loss": 0.8325,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018830633797655842,
+      "loss": 0.7899,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018827890605462543,
+      "loss": 0.2322,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018825144399731722,
+      "loss": 0.8895,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018822395181400845,
+      "loss": 0.8204,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018819642951408393,
+      "loss": 0.8937,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881688771069389,
+      "loss": 0.8283,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018814129460197876,
+      "loss": 0.8599,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018811368200861924,
+      "loss": 0.8399,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880860393362863,
+      "loss": 0.8629,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018805836659441626,
+      "loss": 0.8379,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018803066379245557,
+      "loss": 0.7974,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018800293093986105,
+      "loss": 0.8796,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018797516804609974,
+      "loss": 0.8374,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879473751206489,
+      "loss": 0.7718,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018791955217299612,
+      "loss": 0.7772,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878916992126392,
+      "loss": 0.8448,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018786381624908614,
+      "loss": 0.847,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018783590329185523,
+      "loss": 0.869,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018780796035047502,
+      "loss": 0.8391,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777998743448423,
+      "loss": 0.8389,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018775198455343186,
+      "loss": 0.8268,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018772395171687715,
+      "loss": 0.8606,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018769588893438954,
+      "loss": 0.8513,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018766779621554868,
+      "loss": 0.8114,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018763967356994447,
+      "loss": 0.2363,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018761152100717698,
+      "loss": 0.831,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018758333853685657,
+      "loss": 0.8459,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018755512616860372,
+      "loss": 0.7709,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875268839120492,
+      "loss": 0.8918,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018749861177683394,
+      "loss": 0.8313,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018747030977260903,
+      "loss": 0.8711,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018744197790903584,
+      "loss": 0.8331,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018741361619578587,
+      "loss": 0.8392,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018738522464254085,
+      "loss": 0.8368,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018735680325899265,
+      "loss": 0.8346,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018732835205484335,
+      "loss": 0.8029,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018729987103980526,
+      "loss": 0.8381,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872713602236008,
+      "loss": 0.8346,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018724281961596254,
+      "loss": 0.831,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018721424922663325,
+      "loss": 0.8234,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018718564906536594,
+      "loss": 0.7445,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871570191419237,
+      "loss": 0.8308,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018712835946607976,
+      "loss": 0.8196,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018709967004761757,
+      "loss": 0.7853,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707095089633073,
+      "loss": 0.8688,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018704220202202297,
+      "loss": 0.7763,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018701342343450809,
+      "loss": 0.8516,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869846151436102,
+      "loss": 0.8131,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018695577715916343,
+      "loss": 0.8738,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186926909491012,
+      "loss": 0.8487,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018689801214901043,
+      "loss": 0.9276,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018686908514302326,
+      "loss": 0.8535,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018684012848292517,
+      "loss": 0.8593,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018681114217860092,
+      "loss": 0.8162,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867821262399455,
+      "loss": 0.8025,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867530806768639,
+      "loss": 0.8619,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867240054992713,
+      "loss": 0.8229,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018669490071709292,
+      "loss": 0.8926,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001866657663402642,
+      "loss": 0.8358,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018663660237873056,
+      "loss": 0.8284,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660740884244762,
+      "loss": 0.8724,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186578185741381,
+      "loss": 0.7849,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018654893308550646,
+      "loss": 0.823,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865196508848099,
+      "loss": 0.8131,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018649033914928717,
+      "loss": 0.8411,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018646099788894436,
+      "loss": 0.822,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018643162711379753,
+      "loss": 0.8261,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018640222683387285,
+      "loss": 0.8261,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018637279705920658,
+      "loss": 0.8465,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018634333779984506,
+      "loss": 0.815,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018631384906584456,
+      "loss": 0.8432,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018628433086727162,
+      "loss": 0.8806,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018625478321420268,
+      "loss": 0.8777,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018622520611672432,
+      "loss": 0.7652,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861955995849331,
+      "loss": 0.8274,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018616596362893578,
+      "loss": 0.7962,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018613629825884887,
+      "loss": 0.9205,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018610660348479928,
+      "loss": 0.8402,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860768793169237,
+      "loss": 0.9056,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001860471257653689,
+      "loss": 0.8427,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018601734284029182,
+      "loss": 0.7752,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018598753055185923,
+      "loss": 0.8333,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018595768891024807,
+      "loss": 0.2614,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018592781792564522,
+      "loss": 0.7673,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018589791760824763,
+      "loss": 0.7494,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018586798796826223,
+      "loss": 0.7962,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018583802901590593,
+      "loss": 0.7938,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018580804076140573,
+      "loss": 0.7837,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857780232149986,
+      "loss": 0.8486,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018574797638693145,
+      "loss": 0.8608,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018571790028746124,
+      "loss": 0.8212,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018568779492685492,
+      "loss": 0.8945,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018565766031538943,
+      "loss": 0.8528,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018562749646335166,
+      "loss": 0.7988,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018559730338103854,
+      "loss": 0.8542,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018556708107875694,
+      "loss": 0.8692,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018553682956682374,
+      "loss": 0.8242,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855065488555657,
+      "loss": 0.826,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018547623895531965,
+      "loss": 0.8817,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018544589987643236,
+      "loss": 0.8128,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001854155316292605,
+      "loss": 0.9219,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018538513422417077,
+      "loss": 0.8099,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018535470767153976,
+      "loss": 0.8725,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018532425198175415,
+      "loss": 0.7938,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018529376716521035,
+      "loss": 0.8465,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018526325323231486,
+      "loss": 0.8754,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852327101934841,
+      "loss": 0.9057,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018520213805914445,
+      "loss": 0.8687,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018517153683973207,
+      "loss": 0.8118,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018514090654569326,
+      "loss": 0.8057,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851102471874841,
+      "loss": 0.7991,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018507955877557065,
+      "loss": 0.8644,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850488413204289,
+      "loss": 0.8559,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018501809483254466,
+      "loss": 0.7872,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018498731932241382,
+      "loss": 0.8111,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000184956514800542,
+      "loss": 0.8078,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018492568127744484,
+      "loss": 0.8048,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848948187636478,
+      "loss": 0.8437,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018486392726968636,
+      "loss": 0.8345,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018483300680610572,
+      "loss": 0.7824,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848020573834611,
+      "loss": 0.8461,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847710790123176,
+      "loss": 0.827,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847400717032501,
+      "loss": 0.8853,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847090354668435,
+      "loss": 0.8497,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846779703136924,
+      "loss": 0.8187,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018464687625440146,
+      "loss": 0.8797,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018461575329958512,
+      "loss": 0.8381,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018458460145986765,
+      "loss": 0.8314,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018455342074588318,
+      "loss": 0.801,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018452221116827582,
+      "loss": 0.8539,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018449097273769938,
+      "loss": 0.8717,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844597054648176,
+      "loss": 0.8493,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000184428409360304,
+      "loss": 0.8348,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843970844348421,
+      "loss": 0.8105,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001843657306991251,
+      "loss": 0.8057,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018433434816385604,
+      "loss": 0.8503,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018430293683974788,
+      "loss": 0.8325,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018427149673752333,
+      "loss": 0.7978,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018424002786791502,
+      "loss": 0.7943,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018420853024166527,
+      "loss": 0.7839,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018417700386952633,
+      "loss": 0.8401,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018414544876226016,
+      "loss": 0.8245,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018411386493063866,
+      "loss": 0.7882,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018408225238544343,
+      "loss": 0.8758,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018405061113746586,
+      "loss": 0.8083,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018401894119750723,
+      "loss": 0.8564,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839872425763786,
+      "loss": 0.8145,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018395551528490068,
+      "loss": 0.8222,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018392375933390417,
+      "loss": 0.8507,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838919747342294,
+      "loss": 0.8115,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018386016149672657,
+      "loss": 0.8739,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018382831963225562,
+      "loss": 0.7822,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379644915168623,
+      "loss": 0.8049,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837645500658979,
+      "loss": 0.8425,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018373262238577994,
+      "loss": 0.8009,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018370066612223126,
+      "loss": 0.8705,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836686812861607,
+      "loss": 0.8329,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018363666788848675,
+      "loss": 0.2559,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018360462594013767,
+      "loss": 0.8155,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835725554520515,
+      "loss": 0.8176,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018354045643517597,
+      "loss": 0.8691,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835083289004686,
+      "loss": 0.8472,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018347617285889665,
+      "loss": 0.8335,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018344398832143704,
+      "loss": 0.8139,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018341177529907645,
+      "loss": 0.8327,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018337953380281134,
+      "loss": 0.8787,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018334726384364785,
+      "loss": 0.7597,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018331496543260177,
+      "loss": 0.7448,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832826385806987,
+      "loss": 0.8953,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018325028329897394,
+      "loss": 0.8032,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018321789959847243,
+      "loss": 0.8636,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018318548749024887,
+      "loss": 0.7852,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018315304698536764,
+      "loss": 0.765,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018312057809490275,
+      "loss": 0.7996,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018308808082993805,
+      "loss": 0.7786,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018305555520156693,
+      "loss": 0.7914,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018302300122089254,
+      "loss": 0.8204,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018299041889902767,
+      "loss": 0.2975,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829578082470948,
+      "loss": 0.8378,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829251692762261,
+      "loss": 0.8466,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018289250199756337,
+      "loss": 0.7601,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018285980642225813,
+      "loss": 0.8575,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018282708256147148,
+      "loss": 0.9209,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018279433042637424,
+      "loss": 0.8067,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018276155002814684,
+      "loss": 0.8427,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827287413779794,
+      "loss": 0.8197,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826959044870717,
+      "loss": 0.8842,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182663039366633,
+      "loss": 0.8134,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018263014602788244,
+      "loss": 0.8269,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018259722448204865,
+      "loss": 0.8119,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018256427474036983,
+      "loss": 0.8557,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018253129681409396,
+      "loss": 0.8389,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824982907144786,
+      "loss": 0.8132,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824652564527908,
+      "loss": 0.8786,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018243219404030735,
+      "loss": 0.8724,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018239910348831465,
+      "loss": 0.2844,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018236598480810868,
+      "loss": 0.7825,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000182332838010995,
+      "loss": 0.864,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822996631082887,
+      "loss": 0.8056,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822664601113147,
+      "loss": 0.8319,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018223322903140724,
+      "loss": 0.812,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018219996987991035,
+      "loss": 0.8009,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018216668266817746,
+      "loss": 0.8092,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018213336740757177,
+      "loss": 0.8434,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018210002410946592,
+      "loss": 0.8081,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018206665278524216,
+      "loss": 0.839,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018203325344629228,
+      "loss": 0.7939,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018199982610401776,
+      "loss": 0.8288,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018196637076982946,
+      "loss": 0.8712,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018193288745514788,
+      "loss": 0.7603,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818993761714031,
+      "loss": 0.8001,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018186583693003468,
+      "loss": 0.8208,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018183226974249181,
+      "loss": 0.2846,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018179867462023315,
+      "loss": 0.846,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817650515747269,
+      "loss": 0.8256,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018173140061745083,
+      "loss": 0.8833,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018169772175989217,
+      "loss": 0.8688,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001816640150135478,
+      "loss": 0.8512,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018163028038992401,
+      "loss": 0.8655,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018159651790053657,
+      "loss": 0.8907,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018156272755691095,
+      "loss": 0.2515,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815289093705819,
+      "loss": 0.2631,
+      "step": 1224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018149506335309384,
+      "loss": 0.8266,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018146118951600063,
+      "loss": 0.8386,
+      "step": 1226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018142728787086564,
+      "loss": 0.8001,
+      "step": 1227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018139335842926166,
+      "loss": 0.7995,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018135940120277112,
+      "loss": 0.8077,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813254162029858,
+      "loss": 0.8493,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018129140344150696,
+      "loss": 0.8583,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018125736292994547,
+      "loss": 0.8149,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018122329467992157,
+      "loss": 0.8138,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018118919870306494,
+      "loss": 0.8731,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018115507501101482,
+      "loss": 0.8623,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018112092361541982,
+      "loss": 0.8136,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810867445279381,
+      "loss": 0.8177,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018105253776023718,
+      "loss": 0.7938,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018101830332399406,
+      "loss": 0.8609,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018098404123089524,
+      "loss": 0.7864,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809497514926366,
+      "loss": 0.8506,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018091543412092348,
+      "loss": 0.2665,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018088108912747063,
+      "loss": 0.8063,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018084671652400224,
+      "loss": 0.2472,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018081231632225198,
+      "loss": 0.8502,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018077788853396282,
+      "loss": 0.8301,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074343317088732,
+      "loss": 0.8491,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018070895024478723,
+      "loss": 0.8025,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018067443976743395,
+      "loss": 0.8363,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018063990175060807,
+      "loss": 0.8216,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018060533620609974,
+      "loss": 0.8042,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018057074314570842,
+      "loss": 0.8597,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018053612258124303,
+      "loss": 0.8664,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018050147452452174,
+      "loss": 0.8473,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018046679898737233,
+      "loss": 0.8307,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018043209598163172,
+      "loss": 0.2596,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018039736551914642,
+      "loss": 0.8392,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803626076117721,
+      "loss": 0.8134,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018032782227137402,
+      "loss": 0.8934,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018029300950982662,
+      "loss": 0.8569,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018025816933901382,
+      "loss": 0.8251,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018022330177082884,
+      "loss": 0.8161,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018018840681717423,
+      "loss": 0.847,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000180153484489962,
+      "loss": 0.8237,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018011853480111336,
+      "loss": 0.8985,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018008355776255898,
+      "loss": 0.8409,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018004855338623875,
+      "loss": 0.8661,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018001352168410203,
+      "loss": 0.8902,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799784626681074,
+      "loss": 0.8477,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017994337635022278,
+      "loss": 0.8634,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017990826274242547,
+      "loss": 0.8366,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017987312185670205,
+      "loss": 0.8276,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017983795370504836,
+      "loss": 0.8504,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798027582994696,
+      "loss": 0.7518,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797675356519803,
+      "loss": 0.7601,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017973228577460425,
+      "loss": 0.8724,
+      "step": 1276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017969700867937453,
+      "loss": 0.824,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017966170437833356,
+      "loss": 0.8041,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017962637288353295,
+      "loss": 0.9022,
+      "step": 1279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795910142070337,
+      "loss": 0.8613,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017955562836090603,
+      "loss": 0.236,
+      "step": 1281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017952021535722943,
+      "loss": 0.8197,
+      "step": 1282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017948477520809272,
+      "loss": 0.879,
+      "step": 1283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794493079255939,
+      "loss": 0.8547,
+      "step": 1284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794138135218403,
+      "loss": 0.7926,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017937829200894852,
+      "loss": 0.7648,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017934274339904431,
+      "loss": 0.8262,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793071677042628,
+      "loss": 0.8349,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017927156493674826,
+      "loss": 0.821,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017923593510865427,
+      "loss": 0.2899,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017920027823214362,
+      "loss": 0.8102,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017916459431938837,
+      "loss": 0.8106,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017912888338256973,
+      "loss": 0.8221,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017909314543387817,
+      "loss": 0.8152,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017905738048551347,
+      "loss": 0.876,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790215885496845,
+      "loss": 0.7509,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789857696386094,
+      "loss": 0.7877,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789499237645155,
+      "loss": 0.7796,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 0.8795,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017887815117622672,
+      "loss": 0.8471,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017884222448653255,
+      "loss": 0.8391,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017880627088282092,
+      "loss": 0.8345,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001787702903773652,
+      "loss": 0.8866,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017873428298244789,
+      "loss": 0.8036,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017869824871036066,
+      "loss": 0.8559,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017866218757340435,
+      "loss": 0.8379,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017862609958388902,
+      "loss": 0.8023,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785899847541339,
+      "loss": 0.823,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017855384309646728,
+      "loss": 0.849,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785176746232267,
+      "loss": 0.8579,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001784814793467588,
+      "loss": 0.7916,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017844525727941948,
+      "loss": 0.8506,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017840900843357362,
+      "loss": 0.8364,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783727328215954,
+      "loss": 0.8091,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178336430455868,
+      "loss": 0.8517,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783001013487838,
+      "loss": 0.8424,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017826374551274435,
+      "loss": 0.8478,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017822736296016026,
+      "loss": 0.8605,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017819095370345128,
+      "loss": 0.8476,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017815451775504628,
+      "loss": 0.8696,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017811805512738319,
+      "loss": 0.7904,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001780815658329092,
+      "loss": 0.2503,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017804504988408034,
+      "loss": 0.8155,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017800850729336203,
+      "loss": 0.783,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779719380732286,
+      "loss": 0.8679,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017793534223616353,
+      "loss": 0.8513,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017789871979465938,
+      "loss": 0.8537,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017786207076121775,
+      "loss": 0.8342,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017782539514834942,
+      "loss": 0.7928,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017778869296857414,
+      "loss": 0.8233,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017775196423442074,
+      "loss": 0.8017,
+      "step": 1331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017771520895842725,
+      "loss": 0.8257,
+      "step": 1332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017767842715314057,
+      "loss": 0.2438,
+      "step": 1333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017764161883111672,
+      "loss": 0.8052,
+      "step": 1334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017760478400492084,
+      "loss": 0.832,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017756792268712706,
+      "loss": 0.8163,
+      "step": 1336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017753103489031856,
+      "loss": 0.8751,
+      "step": 1337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017749412062708757,
+      "loss": 0.8819,
+      "step": 1338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017745717991003533,
+      "loss": 0.7803,
+      "step": 1339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017742021275177216,
+      "loss": 0.8751,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773832191649173,
+      "loss": 0.8434,
+      "step": 1341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017734619916209919,
+      "loss": 0.8003,
+      "step": 1342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773091527559551,
+      "loss": 0.8525,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017727207995913143,
+      "loss": 0.8063,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723498078428356,
+      "loss": 0.8381,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017719785524407584,
+      "loss": 0.7754,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017716070335118168,
+      "loss": 0.8749,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017712352511828344,
+      "loss": 0.7987,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017708632055807248,
+      "loss": 0.7849,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001770490896832492,
+      "loss": 0.8777,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017701183250652287,
+      "loss": 0.7972,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017697454904061185,
+      "loss": 0.8148,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769372392982434,
+      "loss": 0.8503,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768999032921538,
+      "loss": 0.8013,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017686254103508827,
+      "loss": 0.8367,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017682515253980102,
+      "loss": 0.8056,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017678773781905517,
+      "loss": 0.8618,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017675029688562282,
+      "loss": 0.8271,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000176712829752285,
+      "loss": 0.8534,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017667533643183175,
+      "loss": 0.8022,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017663781693706189,
+      "loss": 0.8477,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766002712807834,
+      "loss": 0.8557,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000176562699475813,
+      "loss": 0.8692,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017652510153497644,
+      "loss": 0.8544,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017648747747110833,
+      "loss": 0.8203,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017644982729705228,
+      "loss": 0.8434,
+      "step": 1366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764121510256607,
+      "loss": 0.8669,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017637444866979503,
+      "loss": 0.812,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017633672024232552,
+      "loss": 0.826,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017629896575613137,
+      "loss": 0.8216,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762611852241006,
+      "loss": 0.8578,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762233786591303,
+      "loss": 0.8388,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017618554607412622,
+      "loss": 0.8039,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017614768748200313,
+      "loss": 0.8243,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017610980289568467,
+      "loss": 0.8357,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017607189232810334,
+      "loss": 0.8219,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017603395579220049,
+      "loss": 0.845,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017599599330092632,
+      "loss": 0.8408,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759580048672399,
+      "loss": 0.7919,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017591999050410923,
+      "loss": 0.8013,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017588195022451107,
+      "loss": 0.8136,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001758438840414311,
+      "loss": 0.8005,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017580579196786373,
+      "loss": 0.8471,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017576767401681235,
+      "loss": 0.2579,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017572953020128905,
+      "loss": 0.8472,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017569136053431487,
+      "loss": 0.7836,
+      "step": 1386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756531650289196,
+      "loss": 0.8003,
+      "step": 1387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017561494369814184,
+      "loss": 0.2697,
+      "step": 1388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017557669655502908,
+      "loss": 0.8246,
+      "step": 1389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017553842361263757,
+      "loss": 0.816,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017550012488403235,
+      "loss": 0.8962,
+      "step": 1391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001754618003822873,
+      "loss": 0.772,
+      "step": 1392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017542345012048504,
+      "loss": 0.8418,
+      "step": 1393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017538507411171713,
+      "loss": 0.8645,
+      "step": 1394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534667236908372,
+      "loss": 0.7816,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017530824490569384,
+      "loss": 0.8371,
+      "step": 1396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017526979173466538,
+      "loss": 0.7701,
+      "step": 1397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001752313128691249,
+      "loss": 0.7735,
+      "step": 1398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017519280832220768,
+      "loss": 0.8618,
+      "step": 1399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017515427810705793,
+      "loss": 0.8331,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017511572223682852,
+      "loss": 0.7623,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017507714072468109,
+      "loss": 0.833,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017503853358378604,
+      "loss": 0.2779,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017499990082732247,
+      "loss": 0.8607,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017496124246847833,
+      "loss": 0.8785,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017492255852045028,
+      "loss": 0.8284,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748838489964436,
+      "loss": 0.8306,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017484511390967244,
+      "loss": 0.8426,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748063532733596,
+      "loss": 0.8522,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017476756710073664,
+      "loss": 0.8589,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017472875540504383,
+      "loss": 0.8721,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017468991819953012,
+      "loss": 0.8241,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001746510554974532,
+      "loss": 0.2461,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001746121673120795,
+      "loss": 0.7853,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017457325365668403,
+      "loss": 0.8365,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001745343145445506,
+      "loss": 0.8717,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744953499889717,
+      "loss": 0.2722,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744563600032485,
+      "loss": 0.8454,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017441734460069082,
+      "loss": 0.8278,
+      "step": 1419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017437830379461714,
+      "loss": 0.7254,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017433923759835467,
+      "loss": 0.8478,
+      "step": 1421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017430014602523927,
+      "loss": 0.8027,
+      "step": 1422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017426102908861543,
+      "loss": 0.845,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017422188680183638,
+      "loss": 0.8503,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741827191782639,
+      "loss": 0.7894,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017414352623126847,
+      "loss": 0.2296,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741043079742292,
+      "loss": 0.8232,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740650644205339,
+      "loss": 0.8054,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740257955835789,
+      "loss": 0.8137,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739865014767693,
+      "loss": 0.8419,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739471821135187,
+      "loss": 0.8277,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017390783750724937,
+      "loss": 0.8566,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017386846767139224,
+      "loss": 0.8539,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738290726193868,
+      "loss": 0.7909,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017378965236468114,
+      "loss": 0.7923,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017375020692073204,
+      "loss": 0.8221,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017371073630100475,
+      "loss": 0.8045,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736712405189732,
+      "loss": 0.8172,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017363171958811985,
+      "loss": 0.8437,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017359217352193586,
+      "loss": 0.8153,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017355260233392085,
+      "loss": 0.8187,
+      "step": 1441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017351300603758306,
+      "loss": 0.8259,
+      "step": 1442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017347338464643932,
+      "loss": 0.858,
+      "step": 1443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173433738174015,
+      "loss": 0.8461,
+      "step": 1444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017339406663384398,
+      "loss": 0.2796,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017335437003946887,
+      "loss": 0.8832,
+      "step": 1446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017331464840444061,
+      "loss": 0.8754,
+      "step": 1447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017327490174231884,
+      "loss": 0.7838,
+      "step": 1448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017323513006667172,
+      "loss": 0.8723,
+      "step": 1449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017319533339107593,
+      "loss": 0.7955,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017315551172911666,
+      "loss": 0.8374,
+      "step": 1451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017311566509438763,
+      "loss": 0.8442,
+      "step": 1452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017307579350049114,
+      "loss": 0.8231,
+      "step": 1453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017303589696103798,
+      "loss": 0.8106,
+      "step": 1454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017299597548964748,
+      "loss": 0.8042,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729560290999474,
+      "loss": 0.7969,
+      "step": 1456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017291605780557406,
+      "loss": 0.8653,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017287606162017235,
+      "loss": 0.7688,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001728360405573955,
+      "loss": 0.7803,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017279599463090534,
+      "loss": 0.7961,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727559238543722,
+      "loss": 0.7617,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017271582824147488,
+      "loss": 0.8332,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017267570780590062,
+      "loss": 0.8109,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726355625613451,
+      "loss": 0.9028,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017259539252151263,
+      "loss": 0.8289,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017255519770011576,
+      "loss": 0.7989,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725149781108757,
+      "loss": 0.8735,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017247473376752202,
+      "loss": 0.2611,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017243446468379273,
+      "loss": 0.8383,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017239417087343427,
+      "loss": 0.8854,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017235385235020163,
+      "loss": 0.8815,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017231350912785812,
+      "loss": 0.8753,
+      "step": 1472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017227314122017555,
+      "loss": 0.7796,
+      "step": 1473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017223274864093413,
+      "loss": 0.8438,
+      "step": 1474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017219233140392245,
+      "loss": 0.8338,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017215188952293762,
+      "loss": 0.8282,
+      "step": 1476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001721114230117851,
+      "loss": 0.8674,
+      "step": 1477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001720709318842787,
+      "loss": 0.8275,
+      "step": 1478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017203041615424075,
+      "loss": 0.8226,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719898758355019,
+      "loss": 0.2567,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017194931094190129,
+      "loss": 0.8494,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017190872148728625,
+      "loss": 0.8164,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017186810748551272,
+      "loss": 0.8415,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017182746895044485,
+      "loss": 0.842,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017178680589595527,
+      "loss": 0.8747,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017174611833592498,
+      "loss": 0.8118,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017170540628424322,
+      "loss": 0.8517,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716646697548078,
+      "loss": 0.7983,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017162390876152467,
+      "loss": 0.8726,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017158312331830826,
+      "loss": 0.7807,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001715423134390813,
+      "loss": 0.8259,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017150147913777493,
+      "loss": 0.864,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714606204283285,
+      "loss": 0.8355,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017141973732468982,
+      "loss": 0.7997,
+      "step": 1494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017137882984081496,
+      "loss": 0.8062,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017133789799066826,
+      "loss": 0.8614,
+      "step": 1496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017129694178822256,
+      "loss": 0.8235,
+      "step": 1497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712559612474588,
+      "loss": 0.7835,
+      "step": 1498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017121495638236636,
+      "loss": 0.8072,
+      "step": 1499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017117392720694287,
+      "loss": 0.7996,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001711328737351943,
+      "loss": 0.7797,
+      "step": 1501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001710917959811349,
+      "loss": 0.7948,
+      "step": 1502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017105069395878717,
+      "loss": 0.8323,
+      "step": 1503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001710095676821819,
+      "loss": 0.8051,
+      "step": 1504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017096841716535824,
+      "loss": 0.8128,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017092724242236354,
+      "loss": 0.8018,
+      "step": 1506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017088604346725345,
+      "loss": 0.8285,
+      "step": 1507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017084482031409185,
+      "loss": 0.817,
+      "step": 1508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017080357297695088,
+      "loss": 0.7824,
+      "step": 1509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170762301469911,
+      "loss": 0.2849,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017072100580706084,
+      "loss": 0.7837,
+      "step": 1511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017067968600249737,
+      "loss": 0.2549,
+      "step": 1512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706383420703257,
+      "loss": 0.9255,
+      "step": 1513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017059697402465926,
+      "loss": 0.8418,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017055558187961965,
+      "loss": 0.7728,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017051416564933675,
+      "loss": 0.8467,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704727253479486,
+      "loss": 0.8177,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017043126098960148,
+      "loss": 0.8712,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017038977258844993,
+      "loss": 0.2638,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017034826015865665,
+      "loss": 0.8148,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017030672371439254,
+      "loss": 0.8333,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702651632698367,
+      "loss": 0.793,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702235788391765,
+      "loss": 0.7913,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001701819704366073,
+      "loss": 0.83,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017014033807633293,
+      "loss": 0.8429,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017009868177256512,
+      "loss": 0.8442,
+      "step": 1526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017005700153952398,
+      "loss": 0.812,
+      "step": 1527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017001529739143765,
+      "loss": 0.8148,
+      "step": 1528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001699735693425425,
+      "loss": 0.8353,
+      "step": 1529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016993181740708306,
+      "loss": 0.8854,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016989004159931198,
+      "loss": 0.8164,
+      "step": 1531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016984824193349015,
+      "loss": 0.8209,
+      "step": 1532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698064184238864,
+      "loss": 0.8126,
+      "step": 1533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016976457108477797,
+      "loss": 0.881,
+      "step": 1534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972269993045002,
+      "loss": 0.8377,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016968080497519595,
+      "loss": 0.813,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016963888623331726,
+      "loss": 0.8853,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016959694371912354,
+      "loss": 0.7907,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695549774469325,
+      "loss": 0.8268,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016951298743107004,
+      "loss": 0.8364,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001694709736858701,
+      "loss": 0.7974,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016942893622567463,
+      "loss": 0.838,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016938687506483385,
+      "loss": 0.8064,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169344790217706,
+      "loss": 0.8166,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016930268169865738,
+      "loss": 0.7902,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016926054952206234,
+      "loss": 0.8451,
+      "step": 1546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692183937023034,
+      "loss": 0.8403,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691762142537711,
+      "loss": 0.8038,
+      "step": 1548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016913401119086403,
+      "loss": 0.248,
+      "step": 1549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690917845279889,
+      "loss": 0.2648,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016904953427956037,
+      "loss": 0.8099,
+      "step": 1551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690072604600013,
+      "loss": 0.8273,
+      "step": 1552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016896496308374244,
+      "loss": 0.803,
+      "step": 1553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016892264216522272,
+      "loss": 0.8024,
+      "step": 1554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168880297718889,
+      "loss": 0.7826,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016883792975919622,
+      "loss": 0.8154,
+      "step": 1556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016879553830060736,
+      "loss": 0.8517,
+      "step": 1557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016875312335759337,
+      "loss": 0.8119,
+      "step": 1558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687106849446333,
+      "loss": 0.8944,
+      "step": 1559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016866822307621412,
+      "loss": 0.872,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016862573776683083,
+      "loss": 0.7776,
+      "step": 1561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685832290309865,
+      "loss": 0.7614,
+      "step": 1562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016854069688319214,
+      "loss": 0.8331,
+      "step": 1563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016849814133796672,
+      "loss": 0.8927,
+      "step": 1564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016845556240983726,
+      "loss": 0.8506,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016841296011333877,
+      "loss": 0.8324,
+      "step": 1566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016837033446301413,
+      "loss": 0.7731,
+      "step": 1567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016832768547341435,
+      "loss": 0.8203,
+      "step": 1568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016828501315909827,
+      "loss": 0.8295,
+      "step": 1569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016824231753463276,
+      "loss": 0.8031,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016819959861459264,
+      "loss": 0.7687,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016815685641356068,
+      "loss": 0.8481,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016811409094612758,
+      "loss": 0.8311,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016807130222689203,
+      "loss": 0.7637,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016802849027046057,
+      "loss": 0.755,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016798565509144777,
+      "loss": 0.7999,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016794279670447605,
+      "loss": 0.8525,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678999151241758,
+      "loss": 0.8373,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016785701036518534,
+      "loss": 0.75,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016781408244215083,
+      "loss": 0.8493,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016777113136972642,
+      "loss": 0.8748,
+      "step": 1581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016772815716257412,
+      "loss": 0.7306,
+      "step": 1582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016768515983536385,
+      "loss": 0.8766,
+      "step": 1583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016764213940277336,
+      "loss": 0.8464,
+      "step": 1584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016759909587948837,
+      "loss": 0.8319,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001675560292802025,
+      "loss": 0.8533,
+      "step": 1586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016751293961961714,
+      "loss": 0.837,
+      "step": 1587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016746982691244163,
+      "loss": 0.8211,
+      "step": 1588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016742669117339324,
+      "loss": 0.854,
+      "step": 1589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016738353241719693,
+      "loss": 0.8128,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016734035065858562,
+      "loss": 0.7615,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672971459123001,
+      "loss": 0.7678,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000167253918193089,
+      "loss": 0.8898,
+      "step": 1593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016721066751570872,
+      "loss": 0.8741,
+      "step": 1594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671673938949236,
+      "loss": 0.8826,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016712409734550573,
+      "loss": 0.8159,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016708077788223509,
+      "loss": 0.8051,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016703743551989942,
+      "loss": 0.8781,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669940702732943,
+      "loss": 0.8603,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016695068215722317,
+      "loss": 0.8247,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016690727118649721,
+      "loss": 0.8491,
+      "step": 1601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016686383737593542,
+      "loss": 0.8526,
+      "step": 1602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016682038074036467,
+      "loss": 0.2605,
+      "step": 1603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016677690129461947,
+      "loss": 0.7659,
+      "step": 1604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016673339905354226,
+      "loss": 0.83,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016668987403198319,
+      "loss": 0.7934,
+      "step": 1606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666463262448002,
+      "loss": 0.8851,
+      "step": 1607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016660275570685898,
+      "loss": 0.7903,
+      "step": 1608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016655916243303308,
+      "loss": 0.8337,
+      "step": 1609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016651554643820363,
+      "loss": 0.83,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016647190773725972,
+      "loss": 0.8186,
+      "step": 1611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664282463450981,
+      "loss": 0.8056,
+      "step": 1612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016638456227662318,
+      "loss": 0.8517,
+      "step": 1613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663408555467473,
+      "loss": 0.8057,
+      "step": 1614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016629712617039036,
+      "loss": 0.8159,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016625337416248005,
+      "loss": 0.897,
+      "step": 1616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016620959953795187,
+      "loss": 0.8468,
+      "step": 1617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661658023117489,
+      "loss": 0.8352,
+      "step": 1618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016612198249882205,
+      "loss": 0.7801,
+      "step": 1619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016607814011412986,
+      "loss": 0.8537,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016603427517263863,
+      "loss": 0.7523,
+      "step": 1621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001659903876893223,
+      "loss": 0.8342,
+      "step": 1622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001659464776791626,
+      "loss": 0.8031,
+      "step": 1623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016590254515714886,
+      "loss": 0.8414,
+      "step": 1624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016585859013827813,
+      "loss": 0.8101,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016581461263755518,
+      "loss": 0.797,
+      "step": 1626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016577061266999233,
+      "loss": 0.8353,
+      "step": 1627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016572659025060972,
+      "loss": 0.8182,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016568254539443508,
+      "loss": 0.8459,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016563847811650376,
+      "loss": 0.8547,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016559438843185884,
+      "loss": 0.8535,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016555027635555103,
+      "loss": 0.8374,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016550614190263867,
+      "loss": 0.891,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016546198508818772,
+      "loss": 0.8911,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016541780592727178,
+      "loss": 0.8189,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016537360443497217,
+      "loss": 0.8112,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001653293806263777,
+      "loss": 0.7776,
+      "step": 1637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016528513451658482,
+      "loss": 0.8191,
+      "step": 1638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016524086612069766,
+      "loss": 0.7498,
+      "step": 1639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016519657545382795,
+      "loss": 0.8258,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000165152262531095,
+      "loss": 0.8347,
+      "step": 1641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016510792736762565,
+      "loss": 0.7879,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016506356997855447,
+      "loss": 0.8546,
+      "step": 1643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016501919037902347,
+      "loss": 0.8106,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016497478858418235,
+      "loss": 0.8039,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016493036460918842,
+      "loss": 0.8669,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016488591846920635,
+      "loss": 0.7905,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016484145017940863,
+      "loss": 0.7815,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016479695975497519,
+      "loss": 0.7643,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016475244721109349,
+      "loss": 0.8217,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016470791256295862,
+      "loss": 0.7652,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016466335582577312,
+      "loss": 0.8318,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016461877701474714,
+      "loss": 0.8303,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001645741761450984,
+      "loss": 0.8717,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016452955323205206,
+      "loss": 0.823,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016448490829084082,
+      "loss": 0.8724,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016444024133670498,
+      "loss": 0.7762,
+      "step": 1657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016439555238489226,
+      "loss": 0.8317,
+      "step": 1658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016435084145065791,
+      "loss": 0.8389,
+      "step": 1659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016430610854926473,
+      "loss": 0.2659,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000164261353695983,
+      "loss": 0.8164,
+      "step": 1661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001642165769060905,
+      "loss": 0.8271,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016417177819487243,
+      "loss": 0.7868,
+      "step": 1663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016412695757762156,
+      "loss": 0.8538,
+      "step": 1664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016408211506963805,
+      "loss": 0.2598,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016403725068622968,
+      "loss": 0.7996,
+      "step": 1666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016399236444271153,
+      "loss": 0.7656,
+      "step": 1667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016394745635440626,
+      "loss": 0.7966,
+      "step": 1668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016390252643664393,
+      "loss": 0.8467,
+      "step": 1669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016385757470476202,
+      "loss": 0.8192,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016381260117410556,
+      "loss": 0.8418,
+      "step": 1671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016376760586002696,
+      "loss": 0.7908,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016372258877788603,
+      "loss": 0.7737,
+      "step": 1673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016367754994305007,
+      "loss": 0.2711,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016363248937089378,
+      "loss": 0.8104,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635874070767993,
+      "loss": 0.8236,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016354230307615616,
+      "loss": 0.8134,
+      "step": 1677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016349717738436135,
+      "loss": 0.8172,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016345203001681916,
+      "loss": 0.8378,
+      "step": 1679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016340686098894138,
+      "loss": 0.8711,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016336167031614712,
+      "loss": 0.2655,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016331645801386302,
+      "loss": 0.7607,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632712240975229,
+      "loss": 0.8109,
+      "step": 1683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016322596858256807,
+      "loss": 0.8243,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016318069148444725,
+      "loss": 0.7676,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016313539281861646,
+      "loss": 0.8311,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630900726005391,
+      "loss": 0.8138,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630447308456859,
+      "loss": 0.8611,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016299936756953502,
+      "loss": 0.7822,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016295398278757191,
+      "loss": 0.8411,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016290857651528936,
+      "loss": 0.838,
+      "step": 1691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016286314876818748,
+      "loss": 0.8023,
+      "step": 1692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628176995617738,
+      "loss": 0.7905,
+      "step": 1693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016277222891156306,
+      "loss": 0.8183,
+      "step": 1694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016272673683307738,
+      "loss": 0.8142,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016268122334184618,
+      "loss": 0.8155,
+      "step": 1696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016263568845340624,
+      "loss": 0.856,
+      "step": 1697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016259013218330158,
+      "loss": 0.8276,
+      "step": 1698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016254455454708347,
+      "loss": 0.7295,
+      "step": 1699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016249895556031066,
+      "loss": 0.798,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162453335238549,
+      "loss": 0.8109,
+      "step": 1701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016240769359737172,
+      "loss": 0.7418,
+      "step": 1702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623620306523593,
+      "loss": 0.809,
+      "step": 1703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016231634641909948,
+      "loss": 0.8781,
+      "step": 1704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016227064091318733,
+      "loss": 0.2518,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016222491415022507,
+      "loss": 0.8145,
+      "step": 1706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016217916614582226,
+      "loss": 0.8168,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621333969155957,
+      "loss": 0.2464,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016208760647516942,
+      "loss": 0.7579,
+      "step": 1709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016204179484017475,
+      "loss": 0.8759,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016199596202625012,
+      "loss": 0.8283,
+      "step": 1711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016195010804904132,
+      "loss": 0.751,
+      "step": 1712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016190423292420132,
+      "loss": 0.8064,
+      "step": 1713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016185833666739028,
+      "loss": 0.775,
+      "step": 1714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618124192942756,
+      "loss": 0.7889,
+      "step": 1715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016176648082053192,
+      "loss": 0.8381,
+      "step": 1716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016172052126184103,
+      "loss": 0.8068,
+      "step": 1717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016167454063389195,
+      "loss": 0.8363,
+      "step": 1718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016162853895238088,
+      "loss": 0.8337,
+      "step": 1719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016158251623301119,
+      "loss": 0.7949,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016153647249149343,
+      "loss": 0.2455,
+      "step": 1721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016149040774354542,
+      "loss": 0.7752,
+      "step": 1722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016144432200489202,
+      "loss": 0.807,
+      "step": 1723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613982152912653,
+      "loss": 0.8282,
+      "step": 1724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016135208761840458,
+      "loss": 0.7465,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016130593900205617,
+      "loss": 0.8163,
+      "step": 1726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016125976945797368,
+      "loss": 0.8115,
+      "step": 1727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016121357900191776,
+      "loss": 0.8151,
+      "step": 1728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016116736764965625,
+      "loss": 0.7681,
+      "step": 1729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016112113541696412,
+      "loss": 0.7811,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016107488231962343,
+      "loss": 0.8345,
+      "step": 1731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016102860837342346,
+      "loss": 0.8395,
+      "step": 1732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016098231359416048,
+      "loss": 0.8496,
+      "step": 1733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016093599799763794,
+      "loss": 0.832,
+      "step": 1734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608896615996664,
+      "loss": 0.8537,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016084330441606354,
+      "loss": 0.8015,
+      "step": 1736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016079692646265404,
+      "loss": 0.8336,
+      "step": 1737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016075052775526977,
+      "loss": 0.8523,
+      "step": 1738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016070410830974966,
+      "loss": 0.7961,
+      "step": 1739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606576681419397,
+      "loss": 0.8193,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016061120726769296,
+      "loss": 0.2513,
+      "step": 1741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016056472570286955,
+      "loss": 0.826,
+      "step": 1742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016051822346333669,
+      "loss": 0.8428,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604717005649687,
+      "loss": 0.8141,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604251570236468,
+      "loss": 0.8401,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001603785928552594,
+      "loss": 0.7952,
+      "step": 1746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001603320080757019,
+      "loss": 0.8537,
+      "step": 1747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016028540270087674,
+      "loss": 0.779,
+      "step": 1748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602387767466934,
+      "loss": 0.8276,
+      "step": 1749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016019213022906836,
+      "loss": 0.8255,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016014546316392512,
+      "loss": 0.822,
+      "step": 1751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016009877556719425,
+      "loss": 0.8471,
+      "step": 1752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600520674548132,
+      "loss": 0.7904,
+      "step": 1753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600053388427266,
+      "loss": 0.8432,
+      "step": 1754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599585897468859,
+      "loss": 0.8335,
+      "step": 1755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015991182018324976,
+      "loss": 0.8675,
+      "step": 1756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015986503016778353,
+      "loss": 0.8313,
+      "step": 1757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015981821971645984,
+      "loss": 0.8149,
+      "step": 1758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015977138884525814,
+      "loss": 0.8037,
+      "step": 1759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015972453757016485,
+      "loss": 0.8178,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015967766590717334,
+      "loss": 0.8921,
+      "step": 1761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015963077387228406,
+      "loss": 0.8255,
+      "step": 1762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595838614815043,
+      "loss": 0.2794,
+      "step": 1763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595369287508483,
+      "loss": 0.8438,
+      "step": 1764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015948997569633735,
+      "loss": 0.8217,
+      "step": 1765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015944300233399953,
+      "loss": 0.2336,
+      "step": 1766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015939600867987,
+      "loss": 0.8158,
+      "step": 1767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593489947499907,
+      "loss": 0.8432,
+      "step": 1768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015930196056041058,
+      "loss": 0.7258,
+      "step": 1769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015925490612718552,
+      "loss": 0.7994,
+      "step": 1770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015920783146637827,
+      "loss": 0.8217,
+      "step": 1771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591607365940585,
+      "loss": 0.7548,
+      "step": 1772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591136215263028,
+      "loss": 0.7853,
+      "step": 1773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015906648627919454,
+      "loss": 0.827,
+      "step": 1774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590193308688242,
+      "loss": 0.8348,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015897215531128888,
+      "loss": 0.7897,
+      "step": 1776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589249596226928,
+      "loss": 0.8237,
+      "step": 1777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015887774381914685,
+      "loss": 0.7761,
+      "step": 1778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015883050791676891,
+      "loss": 0.7416,
+      "step": 1779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015878325193168373,
+      "loss": 0.8172,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015873597588002286,
+      "loss": 0.8454,
+      "step": 1781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015868867977792466,
+      "loss": 0.8064,
+      "step": 1782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015864136364153442,
+      "loss": 0.8351,
+      "step": 1783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015859402748700426,
+      "loss": 0.2709,
+      "step": 1784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585466713304931,
+      "loss": 0.859,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015849929518816665,
+      "loss": 0.8234,
+      "step": 1786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015845189907619757,
+      "loss": 0.7721,
+      "step": 1787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015840448301076524,
+      "loss": 0.7984,
+      "step": 1788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015835704700805582,
+      "loss": 0.827,
+      "step": 1789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015830959108426238,
+      "loss": 0.8176,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015826211525558472,
+      "loss": 0.8057,
+      "step": 1791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015821461953822942,
+      "loss": 0.8522,
+      "step": 1792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015816710394840993,
+      "loss": 0.864,
+      "step": 1793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581195685023464,
+      "loss": 0.8189,
+      "step": 1794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015807201321626582,
+      "loss": 0.8388,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580244381064019,
+      "loss": 0.7704,
+      "step": 1796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015797684318899515,
+      "loss": 0.8861,
+      "step": 1797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001579292284802928,
+      "loss": 0.7938,
+      "step": 1798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015788159399654896,
+      "loss": 0.8064,
+      "step": 1799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015783393975402435,
+      "loss": 0.8525,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015778626576898645,
+      "loss": 0.234,
+      "step": 1801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015773857205770957,
+      "loss": 0.7646,
+      "step": 1802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015769085863647465,
+      "loss": 0.8549,
+      "step": 1803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015764312552156947,
+      "loss": 0.8069,
+      "step": 1804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575953727292884,
+      "loss": 0.7819,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015754760027593263,
+      "loss": 0.2529,
+      "step": 1806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015749980817781004,
+      "loss": 0.8245,
+      "step": 1807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574519964512352,
+      "loss": 0.7547,
+      "step": 1808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015740416511252937,
+      "loss": 0.8243,
+      "step": 1809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015735631417802056,
+      "loss": 0.7825,
+      "step": 1810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015730844366404335,
+      "loss": 0.8062,
+      "step": 1811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015726055358693914,
+      "loss": 0.7591,
+      "step": 1812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015721264396305595,
+      "loss": 0.7983,
+      "step": 1813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015716471480874852,
+      "loss": 0.8313,
+      "step": 1814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015711676614037806,
+      "loss": 0.8493,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015706879797431275,
+      "loss": 0.8664,
+      "step": 1816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001570208103269272,
+      "loss": 0.8445,
+      "step": 1817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015697280321460275,
+      "loss": 0.8193,
+      "step": 1818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015692477665372735,
+      "loss": 0.8133,
+      "step": 1819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015687673066069565,
+      "loss": 0.8131,
+      "step": 1820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015682866525190887,
+      "loss": 0.7921,
+      "step": 1821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567805804437749,
+      "loss": 0.7852,
+      "step": 1822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015673247625270823,
+      "loss": 0.8034,
+      "step": 1823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015668435269512994,
+      "loss": 0.7662,
+      "step": 1824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015663620978746782,
+      "loss": 0.8132,
+      "step": 1825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015658804754615618,
+      "loss": 0.7885,
+      "step": 1826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015653986598763585,
+      "loss": 0.8169,
+      "step": 1827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015649166512835452,
+      "loss": 0.8404,
+      "step": 1828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015644344498476616,
+      "loss": 0.8509,
+      "step": 1829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015639520557333153,
+      "loss": 0.8559,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563469469105179,
+      "loss": 0.2568,
+      "step": 1831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015629866901279913,
+      "loss": 0.7919,
+      "step": 1832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015625037189665556,
+      "loss": 0.8243,
+      "step": 1833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015620205557857425,
+      "loss": 0.8375,
+      "step": 1834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015615372007504865,
+      "loss": 0.8378,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015610536540257888,
+      "loss": 0.8069,
+      "step": 1836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015605699157767155,
+      "loss": 0.7994,
+      "step": 1837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015600859861683981,
+      "loss": 0.8032,
+      "step": 1838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015596018653660335,
+      "loss": 0.2549,
+      "step": 1839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015591175535348837,
+      "loss": 0.7509,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015586330508402766,
+      "loss": 0.8106,
+      "step": 1841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015581483574476037,
+      "loss": 0.8782,
+      "step": 1842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015576634735223234,
+      "loss": 0.7682,
+      "step": 1843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015571783992299583,
+      "loss": 0.8359,
+      "step": 1844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015566931347360964,
+      "loss": 0.3047,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015562076802063893,
+      "loss": 0.8172,
+      "step": 1846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555722035806555,
+      "loss": 0.773,
+      "step": 1847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015552362017023756,
+      "loss": 0.7877,
+      "step": 1848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015547501780596986,
+      "loss": 0.8393,
+      "step": 1849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001554263965044435,
+      "loss": 0.8224,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015537775628225617,
+      "loss": 0.7709,
+      "step": 1851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015532909715601192,
+      "loss": 0.8389,
+      "step": 1852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015528041914232135,
+      "loss": 0.8205,
+      "step": 1853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015523172225780142,
+      "loss": 0.8108,
+      "step": 1854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015518300651907557,
+      "loss": 0.7792,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001551342719427737,
+      "loss": 0.7719,
+      "step": 1856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015508551854553207,
+      "loss": 0.8441,
+      "step": 1857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015503674634399347,
+      "loss": 0.7618,
+      "step": 1858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015498795535480694,
+      "loss": 0.8291,
+      "step": 1859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015493914559462817,
+      "loss": 0.8238,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015489031708011908,
+      "loss": 0.8338,
+      "step": 1861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000154841469827948,
+      "loss": 0.8288,
+      "step": 1862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015479260385478974,
+      "loss": 0.8347,
+      "step": 1863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015474371917732546,
+      "loss": 0.8584,
+      "step": 1864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015469481581224272,
+      "loss": 0.8379,
+      "step": 1865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015464589377623536,
+      "loss": 0.836,
+      "step": 1866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015459695308600376,
+      "loss": 0.8055,
+      "step": 1867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015454799375825458,
+      "loss": 0.8236,
+      "step": 1868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544990158097008,
+      "loss": 0.8691,
+      "step": 1869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015445001925706183,
+      "loss": 0.7847,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544010041170634,
+      "loss": 0.8469,
+      "step": 1871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001543519704064376,
+      "loss": 0.8533,
+      "step": 1872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015430291814192286,
+      "loss": 0.8631,
+      "step": 1873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001542538473402639,
+      "loss": 0.8228,
+      "step": 1874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001542047580182118,
+      "loss": 0.7948,
+      "step": 1875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015415565019252396,
+      "loss": 0.8247,
+      "step": 1876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015410652387996414,
+      "loss": 0.8067,
+      "step": 1877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015405737909730233,
+      "loss": 0.8446,
+      "step": 1878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015400821586131487,
+      "loss": 0.7931,
+      "step": 1879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015395903418878443,
+      "loss": 0.8243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015390983409649986,
+      "loss": 0.2715,
+      "step": 1881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015386061560125643,
+      "loss": 0.7832,
+      "step": 1882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538113787198557,
+      "loss": 0.8325,
+      "step": 1883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001537621234691053,
+      "loss": 0.8805,
+      "step": 1884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001537128498658194,
+      "loss": 0.8008,
+      "step": 1885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015366355792681823,
+      "loss": 0.8548,
+      "step": 1886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001536142476689284,
+      "loss": 0.7783,
+      "step": 1887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535649191089827,
+      "loss": 0.789,
+      "step": 1888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015351557226382024,
+      "loss": 0.7985,
+      "step": 1889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015346620715028632,
+      "loss": 0.2683,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534168237852325,
+      "loss": 0.8147,
+      "step": 1891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015336742218551646,
+      "loss": 0.8464,
+      "step": 1892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533180023680023,
+      "loss": 0.8194,
+      "step": 1893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015326856434956027,
+      "loss": 0.8527,
+      "step": 1894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015321910814706673,
+      "loss": 0.8095,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015316963377740437,
+      "loss": 0.8398,
+      "step": 1896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015312014125746202,
+      "loss": 0.7514,
+      "step": 1897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015307063060413466,
+      "loss": 0.858,
+      "step": 1898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015302110183432365,
+      "loss": 0.845,
+      "step": 1899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015297155496493624,
+      "loss": 0.743,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015292199001288621,
+      "loss": 0.8282,
+      "step": 1901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015287240699509317,
+      "loss": 0.7786,
+      "step": 1902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015282280592848312,
+      "loss": 0.7861,
+      "step": 1903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015277318682998815,
+      "loss": 0.8551,
+      "step": 1904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001527235497165465,
+      "loss": 0.8281,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015267389460510257,
+      "loss": 0.7517,
+      "step": 1906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015262422151260692,
+      "loss": 0.8233,
+      "step": 1907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001525745304560162,
+      "loss": 0.7783,
+      "step": 1908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015252482145229328,
+      "loss": 0.8248,
+      "step": 1909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015247509451840708,
+      "loss": 0.8108,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015242534967133258,
+      "loss": 0.808,
+      "step": 1911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015237558692805106,
+      "loss": 0.8555,
+      "step": 1912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015232580630554976,
+      "loss": 0.8051,
+      "step": 1913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015227600782082202,
+      "loss": 0.779,
+      "step": 1914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015222619149086745,
+      "loss": 0.8076,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015217635733269153,
+      "loss": 0.7702,
+      "step": 1916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015212650536330596,
+      "loss": 0.815,
+      "step": 1917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015207663559972846,
+      "loss": 0.851,
+      "step": 1918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520267480589829,
+      "loss": 0.8106,
+      "step": 1919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015197684275809908,
+      "loss": 0.825,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015192691971411304,
+      "loss": 0.8128,
+      "step": 1921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015187697894406669,
+      "loss": 0.7762,
+      "step": 1922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015182702046500818,
+      "loss": 0.7808,
+      "step": 1923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517770442939916,
+      "loss": 0.8275,
+      "step": 1924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000151727050448077,
+      "loss": 0.8604,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015167703894433066,
+      "loss": 0.2675,
+      "step": 1926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015162700979982476,
+      "loss": 0.8579,
+      "step": 1927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015157696303163748,
+      "loss": 0.2517,
+      "step": 1928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015152689865685311,
+      "loss": 0.823,
+      "step": 1929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015147681669256188,
+      "loss": 0.7529,
+      "step": 1930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015142671715586006,
+      "loss": 0.7297,
+      "step": 1931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015137660006384987,
+      "loss": 0.8169,
+      "step": 1932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513264654336396,
+      "loss": 0.8978,
+      "step": 1933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015127631328234347,
+      "loss": 0.7883,
+      "step": 1934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015122614362708168,
+      "loss": 0.7467,
+      "step": 1935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015117595648498039,
+      "loss": 0.812,
+      "step": 1936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015112575187317185,
+      "loss": 0.8157,
+      "step": 1937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015107552980879417,
+      "loss": 0.8724,
+      "step": 1938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015102529030899134,
+      "loss": 0.7844,
+      "step": 1939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015097503339091346,
+      "loss": 0.8601,
+      "step": 1940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001509247590717165,
+      "loss": 0.8061,
+      "step": 1941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508744673685624,
+      "loss": 0.8711,
+      "step": 1942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015082415829861896,
+      "loss": 0.7991,
+      "step": 1943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015077383187906004,
+      "loss": 0.8734,
+      "step": 1944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015072348812706526,
+      "loss": 0.8075,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001506731270598203,
+      "loss": 0.8003,
+      "step": 1946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015062274869451667,
+      "loss": 0.7891,
+      "step": 1947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505723530483518,
+      "loss": 0.7762,
+      "step": 1948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015052194013852907,
+      "loss": 0.8112,
+      "step": 1949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015047150998225763,
+      "loss": 0.2682,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015042106259675268,
+      "loss": 0.8277,
+      "step": 1951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015037059799923516,
+      "loss": 0.7672,
+      "step": 1952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015032011620693197,
+      "loss": 0.8111,
+      "step": 1953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015026961723707584,
+      "loss": 0.8591,
+      "step": 1954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001502191011069054,
+      "loss": 0.7904,
+      "step": 1955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501685678336651,
+      "loss": 0.8033,
+      "step": 1956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015011801743460525,
+      "loss": 0.8075,
+      "step": 1957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015006744992698205,
+      "loss": 0.7665,
+      "step": 1958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015001686532805747,
+      "loss": 0.2314,
+      "step": 1959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014996626365509935,
+      "loss": 0.7421,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014991564492538133,
+      "loss": 0.7705,
+      "step": 1961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014986500915618292,
+      "loss": 0.8071,
+      "step": 1962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001498143563647895,
+      "loss": 0.8196,
+      "step": 1963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014976368656849208,
+      "loss": 0.8265,
+      "step": 1964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014971299978458767,
+      "loss": 0.8431,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014966229603037887,
+      "loss": 0.8835,
+      "step": 1966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014961157532317434,
+      "loss": 0.8727,
+      "step": 1967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014956083768028825,
+      "loss": 0.7598,
+      "step": 1968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001495100831190408,
+      "loss": 0.7914,
+      "step": 1969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014945931165675775,
+      "loss": 0.813,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494085233107708,
+      "loss": 0.7817,
+      "step": 1971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014935771809841726,
+      "loss": 0.8038,
+      "step": 1972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014930689603704037,
+      "loss": 0.8176,
+      "step": 1973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014925605714398899,
+      "loss": 0.8169,
+      "step": 1974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014920520143661773,
+      "loss": 0.7763,
+      "step": 1975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014915432893228704,
+      "loss": 0.7799,
+      "step": 1976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014910343964836295,
+      "loss": 0.7472,
+      "step": 1977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001490525336022174,
+      "loss": 0.8519,
+      "step": 1978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001490016108112279,
+      "loss": 0.8001,
+      "step": 1979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014895067129277776,
+      "loss": 0.7989,
+      "step": 1980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148899715064256,
+      "loss": 0.8141,
+      "step": 1981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014884874214305725,
+      "loss": 0.7818,
+      "step": 1982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014879775254658198,
+      "loss": 0.7927,
+      "step": 1983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014874674629223624,
+      "loss": 0.8163,
+      "step": 1984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014869572339743184,
+      "loss": 0.7768,
+      "step": 1985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014864468387958618,
+      "loss": 0.2452,
+      "step": 1986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014859362775612245,
+      "loss": 0.7701,
+      "step": 1987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014854255504446944,
+      "loss": 0.7936,
+      "step": 1988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014849146576206157,
+      "loss": 0.7594,
+      "step": 1989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148440359926339,
+      "loss": 0.8198,
+      "step": 1990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014838923755474746,
+      "loss": 0.7906,
+      "step": 1991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014833809866473844,
+      "loss": 0.8601,
+      "step": 1992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014828694327376893,
+      "loss": 0.8609,
+      "step": 1993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014823577139930165,
+      "loss": 0.8014,
+      "step": 1994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014818458305880487,
+      "loss": 0.7845,
+      "step": 1995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014813337826975257,
+      "loss": 0.8296,
+      "step": 1996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001480821570496243,
+      "loss": 0.8627,
+      "step": 1997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001480309194159052,
+      "loss": 0.8478,
+      "step": 1998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014797966538608603,
+      "loss": 0.7461,
+      "step": 1999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014792839497766312,
+      "loss": 0.8032,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001478771082081385,
+      "loss": 0.7774,
+      "step": 2001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014782580509501965,
+      "loss": 0.9235,
+      "step": 2002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001477744856558197,
+      "loss": 0.8377,
+      "step": 2003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001477231499080573,
+      "loss": 0.8348,
+      "step": 2004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014767179786925674,
+      "loss": 0.8295,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014762042955694785,
+      "loss": 0.2633,
+      "step": 2006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014756904498866596,
+      "loss": 0.7761,
+      "step": 2007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014751764418195206,
+      "loss": 0.8139,
+      "step": 2008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014746622715435253,
+      "loss": 0.8169,
+      "step": 2009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014741479392341942,
+      "loss": 0.83,
+      "step": 2010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014736334450671023,
+      "loss": 0.7992,
+      "step": 2011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014731187892178804,
+      "loss": 0.795,
+      "step": 2012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472603971862214,
+      "loss": 0.7799,
+      "step": 2013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014720889931758445,
+      "loss": 0.7628,
+      "step": 2014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014715738533345677,
+      "loss": 0.7756,
+      "step": 2015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014710585525142341,
+      "loss": 0.7857,
+      "step": 2016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014705430908907503,
+      "loss": 0.7977,
+      "step": 2017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014700274686400765,
+      "loss": 0.7962,
+      "step": 2018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001469511685938229,
+      "loss": 0.2617,
+      "step": 2019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001468995742961278,
+      "loss": 0.2819,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014684796398853486,
+      "loss": 0.8225,
+      "step": 2021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014679633768866204,
+      "loss": 0.8198,
+      "step": 2022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014674469541413278,
+      "loss": 0.8391,
+      "step": 2023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014669303718257601,
+      "loss": 0.2782,
+      "step": 2024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014664136301162603,
+      "loss": 0.8271,
+      "step": 2025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014658967291892265,
+      "loss": 0.8584,
+      "step": 2026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014653796692211106,
+      "loss": 0.8214,
+      "step": 2027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014648624503884195,
+      "loss": 0.7942,
+      "step": 2028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001464345072867713,
+      "loss": 0.7658,
+      "step": 2029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014638275368356068,
+      "loss": 0.7728,
+      "step": 2030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000146330984246877,
+      "loss": 0.8052,
+      "step": 2031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014627919899439247,
+      "loss": 0.7986,
+      "step": 2032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014622739794378485,
+      "loss": 0.8181,
+      "step": 2033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001461755811127372,
+      "loss": 0.8112,
+      "step": 2034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014612374851893806,
+      "loss": 0.2576,
+      "step": 2035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014607190018008125,
+      "loss": 0.8126,
+      "step": 2036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014602003611386598,
+      "loss": 0.7631,
+      "step": 2037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459681563379969,
+      "loss": 0.8223,
+      "step": 2038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014591626087018394,
+      "loss": 0.8791,
+      "step": 2039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014586434972814244,
+      "loss": 0.8292,
+      "step": 2040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014581242292959309,
+      "loss": 0.7896,
+      "step": 2041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001457604804922619,
+      "loss": 0.2597,
+      "step": 2042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001457085224338802,
+      "loss": 0.8218,
+      "step": 2043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014565654877218473,
+      "loss": 0.7369,
+      "step": 2044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014560455952491744,
+      "loss": 0.7715,
+      "step": 2045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014555255470982573,
+      "loss": 0.8092,
+      "step": 2046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014550053434466225,
+      "loss": 0.7948,
+      "step": 2047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454484984471849,
+      "loss": 0.7781,
+      "step": 2048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014539644703515702,
+      "loss": 0.7925,
+      "step": 2049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001453443801263471,
+      "loss": 0.8307,
+      "step": 2050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014529229773852897,
+      "loss": 0.7888,
+      "step": 2051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014524019988948185,
+      "loss": 0.8295,
+      "step": 2052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451880865969901,
+      "loss": 0.7478,
+      "step": 2053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451359578788434,
+      "loss": 0.7614,
+      "step": 2054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001450838137528367,
+      "loss": 0.8246,
+      "step": 2055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001450316542367702,
+      "loss": 0.8441,
+      "step": 2056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001449794793484494,
+      "loss": 0.2623,
+      "step": 2057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014492728910568497,
+      "loss": 0.7538,
+      "step": 2058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014487508352629285,
+      "loss": 0.7323,
+      "step": 2059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014482286262809427,
+      "loss": 0.7213,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001447706264289156,
+      "loss": 0.8248,
+      "step": 2061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014471837494658849,
+      "loss": 0.772,
+      "step": 2062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446661081989498,
+      "loss": 0.7592,
+      "step": 2063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446138262038416,
+      "loss": 0.774,
+      "step": 2064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014456152897911114,
+      "loss": 0.8446,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014450921654261093,
+      "loss": 0.7491,
+      "step": 2066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014445688891219862,
+      "loss": 0.8323,
+      "step": 2067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014440454610573702,
+      "loss": 0.8296,
+      "step": 2068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001443521881410942,
+      "loss": 0.7521,
+      "step": 2069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014429981503614336,
+      "loss": 0.7989,
+      "step": 2070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014424742680876286,
+      "loss": 0.7999,
+      "step": 2071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001441950234768362,
+      "loss": 0.7676,
+      "step": 2072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014414260505825216,
+      "loss": 0.8208,
+      "step": 2073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014409017157090452,
+      "loss": 0.7834,
+      "step": 2074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014403772303269227,
+      "loss": 0.7884,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014398525946151953,
+      "loss": 0.7772,
+      "step": 2076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014393278087529557,
+      "loss": 0.7699,
+      "step": 2077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014388028729193475,
+      "loss": 0.7977,
+      "step": 2078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001438277787293566,
+      "loss": 0.7888,
+      "step": 2079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014377525520548572,
+      "loss": 0.7981,
+      "step": 2080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014372271673825183,
+      "loss": 0.8237,
+      "step": 2081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014367016334558972,
+      "loss": 0.8598,
+      "step": 2082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014361759504543935,
+      "loss": 0.8236,
+      "step": 2083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001435650118557457,
+      "loss": 0.7963,
+      "step": 2084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014351241379445887,
+      "loss": 0.8211,
+      "step": 2085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014345980087953405,
+      "loss": 0.7573,
+      "step": 2086
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014340717312893143,
+      "loss": 0.7942,
+      "step": 2087
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014335453056061635,
+      "loss": 0.8128,
+      "step": 2088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433018731925591,
+      "loss": 0.7821,
+      "step": 2089
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014324920104273516,
+      "loss": 0.7376,
+      "step": 2090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014319651412912494,
+      "loss": 0.7882,
+      "step": 2091
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143143812469714,
+      "loss": 0.8456,
+      "step": 2092
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430910960824928,
+      "loss": 0.7886,
+      "step": 2093
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014303836498545693,
+      "loss": 0.8127,
+      "step": 2094
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014298561919660698,
+      "loss": 0.8109,
+      "step": 2095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001429328587339485,
+      "loss": 0.7562,
+      "step": 2096
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014288008361549215,
+      "loss": 0.8414,
+      "step": 2097
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001428272938592535,
+      "loss": 0.8078,
+      "step": 2098
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014277448948325314,
+      "loss": 0.7817,
+      "step": 2099
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001427216705055167,
+      "loss": 0.7639,
+      "step": 2100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014266883694407474,
+      "loss": 0.7982,
+      "step": 2101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001426159888169628,
+      "loss": 0.7883,
+      "step": 2102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014256312614222143,
+      "loss": 0.8307,
+      "step": 2103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014251024893789616,
+      "loss": 0.8342,
+      "step": 2104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014245735722203735,
+      "loss": 0.7834,
+      "step": 2105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001424044510127005,
+      "loss": 0.79,
+      "step": 2106
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014235153032794593,
+      "loss": 0.7746,
+      "step": 2107
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014229859518583894,
+      "loss": 0.8053,
+      "step": 2108
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014224564560444975,
+      "loss": 0.8238,
+      "step": 2109
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014219268160185354,
+      "loss": 0.8474,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014213970319613037,
+      "loss": 0.8723,
+      "step": 2111
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014208671040536527,
+      "loss": 0.8003,
+      "step": 2112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014203370324764814,
+      "loss": 0.8586,
+      "step": 2113
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014198068174107377,
+      "loss": 0.8702,
+      "step": 2114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014192764590374198,
+      "loss": 0.7979,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001418745957537572,
+      "loss": 0.8149,
+      "step": 2116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014182153130922907,
+      "loss": 0.716,
+      "step": 2117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014176845258827192,
+      "loss": 0.8082,
+      "step": 2118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014171535960900498,
+      "loss": 0.8649,
+      "step": 2119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014166225238955235,
+      "loss": 0.7966,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014160913094804305,
+      "loss": 0.8481,
+      "step": 2121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415559953026109,
+      "loss": 0.8358,
+      "step": 2122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014150284547139448,
+      "loss": 0.7594,
+      "step": 2123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014144968147253742,
+      "loss": 0.8049,
+      "step": 2124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014139650332418811,
+      "loss": 0.8728,
+      "step": 2125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001413433110444996,
+      "loss": 0.8459,
+      "step": 2126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014129010465163,
+      "loss": 0.809,
+      "step": 2127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412368841637421,
+      "loss": 0.2631,
+      "step": 2128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001411836495990036,
+      "loss": 0.8462,
+      "step": 2129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014113040097558685,
+      "loss": 0.8745,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014107713831166917,
+      "loss": 0.8179,
+      "step": 2131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014102386162543256,
+      "loss": 0.8095,
+      "step": 2132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014097057093506382,
+      "loss": 0.8472,
+      "step": 2133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014091726625875463,
+      "loss": 0.8153,
+      "step": 2134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001408639476147013,
+      "loss": 0.2717,
+      "step": 2135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000140810615021105,
+      "loss": 0.8141,
+      "step": 2136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014075726849617167,
+      "loss": 0.8113,
+      "step": 2137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014070390805811194,
+      "loss": 0.2689,
+      "step": 2138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001406505337251412,
+      "loss": 0.7875,
+      "step": 2139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014059714551547967,
+      "loss": 0.7749,
+      "step": 2140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014054374344735218,
+      "loss": 0.778,
+      "step": 2141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014049032753898842,
+      "loss": 0.8094,
+      "step": 2142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014043689780862272,
+      "loss": 0.7546,
+      "step": 2143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014038345427449408,
+      "loss": 0.8416,
+      "step": 2144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014032999695484639,
+      "loss": 0.8124,
+      "step": 2145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014027652586792803,
+      "loss": 0.7701,
+      "step": 2146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001402230410319923,
+      "loss": 0.7998,
+      "step": 2147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014016954246529696,
+      "loss": 0.819,
+      "step": 2148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014011603018610464,
+      "loss": 0.7606,
+      "step": 2149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014006250421268258,
+      "loss": 0.8229,
+      "step": 2150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014000896456330272,
+      "loss": 0.8264,
+      "step": 2151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013995541125624164,
+      "loss": 0.7669,
+      "step": 2152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013990184430978054,
+      "loss": 0.7995,
+      "step": 2153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013984826374220544,
+      "loss": 0.8311,
+      "step": 2154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013979466957180677,
+      "loss": 0.8505,
+      "step": 2155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013974106181687982,
+      "loss": 0.7606,
+      "step": 2156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396874404957244,
+      "loss": 0.8069,
+      "step": 2157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139633805626645,
+      "loss": 0.7769,
+      "step": 2158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013958015722795067,
+      "loss": 0.8073,
+      "step": 2159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395264953179552,
+      "loss": 0.857,
+      "step": 2160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013947281991497684,
+      "loss": 0.8047,
+      "step": 2161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394191310373385,
+      "loss": 0.8157,
+      "step": 2162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013936542870336783,
+      "loss": 0.7983,
+      "step": 2163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013931171293139687,
+      "loss": 0.8131,
+      "step": 2164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013925798373976236,
+      "loss": 0.8283,
+      "step": 2165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392042411468056,
+      "loss": 0.7969,
+      "step": 2166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001391504851708724,
+      "loss": 0.7966,
+      "step": 2167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390967158303133,
+      "loss": 0.7355,
+      "step": 2168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013904293314348317,
+      "loss": 0.8319,
+      "step": 2169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013898913712874172,
+      "loss": 0.8065,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001389353278044529,
+      "loss": 0.785,
+      "step": 2171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001388815051889855,
+      "loss": 0.8385,
+      "step": 2172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013882766930071257,
+      "loss": 0.8387,
+      "step": 2173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001387738201580119,
+      "loss": 0.7968,
+      "step": 2174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013871995777926569,
+      "loss": 0.8008,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001386660821828608,
+      "loss": 0.7823,
+      "step": 2176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001386121933871884,
+      "loss": 0.8046,
+      "step": 2177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013855829141064428,
+      "loss": 0.8292,
+      "step": 2178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013850437627162878,
+      "loss": 0.8052,
+      "step": 2179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013845044798854657,
+      "loss": 0.7788,
+      "step": 2180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013839650657980704,
+      "loss": 0.7688,
+      "step": 2181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013834255206382378,
+      "loss": 0.7765,
+      "step": 2182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001382885844590151,
+      "loss": 0.7854,
+      "step": 2183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013823460378380365,
+      "loss": 0.7924,
+      "step": 2184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381806100566166,
+      "loss": 0.8096,
+      "step": 2185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013812660329588547,
+      "loss": 0.7744,
+      "step": 2186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013807258352004634,
+      "loss": 0.8001,
+      "step": 2187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013801855074753978,
+      "loss": 0.7849,
+      "step": 2188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013796450499681059,
+      "loss": 0.879,
+      "step": 2189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013791044628630824,
+      "loss": 0.7646,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013785637463448644,
+      "loss": 0.7175,
+      "step": 2191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013780229005980337,
+      "loss": 0.8318,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013774819258072172,
+      "loss": 0.8054,
+      "step": 2193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013769408221570844,
+      "loss": 0.781,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013763995898323497,
+      "loss": 0.8199,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001375858229017771,
+      "loss": 0.8293,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013753167398981506,
+      "loss": 0.8264,
+      "step": 2197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013747751226583335,
+      "loss": 0.8126,
+      "step": 2198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013742333774832097,
+      "loss": 0.793,
+      "step": 2199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013736915045577122,
+      "loss": 0.7804,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013731495040668177,
+      "loss": 0.8397,
+      "step": 2201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013726073761955465,
+      "loss": 0.819,
+      "step": 2202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013720651211289624,
+      "loss": 0.8163,
+      "step": 2203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013715227390521728,
+      "loss": 0.8243,
+      "step": 2204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013709802301503277,
+      "loss": 0.875,
+      "step": 2205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013704375946086214,
+      "loss": 0.7967,
+      "step": 2206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013698948326122906,
+      "loss": 0.8868,
+      "step": 2207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369351944346616,
+      "loss": 0.7873,
+      "step": 2208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013688089299969207,
+      "loss": 0.7497,
+      "step": 2209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368265789748571,
+      "loss": 0.2635,
+      "step": 2210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001367722523786976,
+      "loss": 0.8518,
+      "step": 2211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013671791322975884,
+      "loss": 0.2996,
+      "step": 2212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013666356154659037,
+      "loss": 0.8488,
+      "step": 2213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013660919734774588,
+      "loss": 0.8353,
+      "step": 2214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013655482065178353,
+      "loss": 0.7757,
+      "step": 2215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013650043147726558,
+      "loss": 0.8366,
+      "step": 2216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001364460298427586,
+      "loss": 0.7848,
+      "step": 2217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013639161576683355,
+      "loss": 0.2643,
+      "step": 2218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001363371892680654,
+      "loss": 0.7999,
+      "step": 2219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013628275036503356,
+      "loss": 0.7872,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013622829907632155,
+      "loss": 0.8316,
+      "step": 2221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013617383542051715,
+      "loss": 0.849,
+      "step": 2222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013611935941621242,
+      "loss": 0.7794,
+      "step": 2223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013606487108200357,
+      "loss": 0.8153,
+      "step": 2224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013601037043649102,
+      "loss": 0.8211,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013595585749827944,
+      "loss": 0.7976,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359013322859777,
+      "loss": 0.852,
+      "step": 2227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358467948181987,
+      "loss": 0.8572,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013579224511355982,
+      "loss": 0.7863,
+      "step": 2229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001357376831906823,
+      "loss": 0.8154,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013568310906819178,
+      "loss": 0.808,
+      "step": 2231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013562852276471798,
+      "loss": 0.8619,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013557392429889474,
+      "loss": 0.8243,
+      "step": 2233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013551931368936017,
+      "loss": 0.8242,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013546469095475637,
+      "loss": 0.7942,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013541005611372972,
+      "loss": 0.8225,
+      "step": 2236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001353554091849306,
+      "loss": 0.7886,
+      "step": 2237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001353007501870137,
+      "loss": 0.8689,
+      "step": 2238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001352460791386376,
+      "loss": 0.2362,
+      "step": 2239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013519139605846516,
+      "loss": 0.8328,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351367009651633,
+      "loss": 0.2435,
+      "step": 2241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013508199387740307,
+      "loss": 0.8057,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013502727481385951,
+      "loss": 0.7958,
+      "step": 2243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013497254379321187,
+      "loss": 0.7666,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013491780083414346,
+      "loss": 0.7937,
+      "step": 2245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013486304595534157,
+      "loss": 0.8224,
+      "step": 2246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013480827917549765,
+      "loss": 0.767,
+      "step": 2247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013475350051330725,
+      "loss": 0.7725,
+      "step": 2248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013469870998746985,
+      "loss": 0.845,
+      "step": 2249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013464390761668905,
+      "loss": 0.8341,
+      "step": 2250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001345890934196725,
+      "loss": 0.8326,
+      "step": 2251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013453426741513196,
+      "loss": 0.8426,
+      "step": 2252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013447942962178298,
+      "loss": 0.7725,
+      "step": 2253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013442458005834544,
+      "loss": 0.7978,
+      "step": 2254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000134369718743543,
+      "loss": 0.7984,
+      "step": 2255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001343148456961034,
+      "loss": 0.2803,
+      "step": 2256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342599609347585,
+      "loss": 0.8217,
+      "step": 2257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013420506447824404,
+      "loss": 0.7959,
+      "step": 2258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013415015634529972,
+      "loss": 0.2807,
+      "step": 2259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013409523655466934,
+      "loss": 0.8314,
+      "step": 2260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340403051251006,
+      "loss": 0.7843,
+      "step": 2261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013398536207534517,
+      "loss": 0.8094,
+      "step": 2262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001339304074241588,
+      "loss": 0.8135,
+      "step": 2263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000133875441190301,
+      "loss": 0.7938,
+      "step": 2264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013382046339253543,
+      "loss": 0.8125,
+      "step": 2265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001337654740496296,
+      "loss": 0.79,
+      "step": 2266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013371047318035492,
+      "loss": 0.7911,
+      "step": 2267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336554608034869,
+      "loss": 0.8007,
+      "step": 2268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013360043693780475,
+      "loss": 0.8704,
+      "step": 2269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001335454016020918,
+      "loss": 0.8199,
+      "step": 2270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001334903548151352,
+      "loss": 0.7703,
+      "step": 2271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013343529659572598,
+      "loss": 0.779,
+      "step": 2272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013338022696265917,
+      "loss": 0.8374,
+      "step": 2273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013332514593473365,
+      "loss": 0.8086,
+      "step": 2274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013327005353075215,
+      "loss": 0.7985,
+      "step": 2275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013321494976952132,
+      "loss": 0.7761,
+      "step": 2276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013315983466985173,
+      "loss": 0.8072,
+      "step": 2277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001331047082505577,
+      "loss": 0.8076,
+      "step": 2278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013304957053045753,
+      "loss": 0.8301,
+      "step": 2279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013299442152837337,
+      "loss": 0.7379,
+      "step": 2280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013293926126313113,
+      "loss": 0.2575,
+      "step": 2281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013288408975356065,
+      "loss": 0.8,
+      "step": 2282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013282890701849559,
+      "loss": 0.7966,
+      "step": 2283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327737130767734,
+      "loss": 0.8046,
+      "step": 2284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327185079472354,
+      "loss": 0.8132,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013266329164872675,
+      "loss": 0.7763,
+      "step": 2286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013260806420009637,
+      "loss": 0.7553,
+      "step": 2287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013255282562019702,
+      "loss": 0.8061,
+      "step": 2288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013249757592788522,
+      "loss": 0.8697,
+      "step": 2289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001324423151420213,
+      "loss": 0.8003,
+      "step": 2290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013238704328146945,
+      "loss": 0.8617,
+      "step": 2291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001323317603650975,
+      "loss": 0.7833,
+      "step": 2292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013227646641177722,
+      "loss": 0.8337,
+      "step": 2293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013222116144038398,
+      "loss": 0.8469,
+      "step": 2294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013216584546979702,
+      "loss": 0.8287,
+      "step": 2295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013211051851889932,
+      "loss": 0.7712,
+      "step": 2296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013205518060657755,
+      "loss": 0.7826,
+      "step": 2297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001319998317517222,
+      "loss": 0.794,
+      "step": 2298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013194447197322748,
+      "loss": 0.7724,
+      "step": 2299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013188910128999128,
+      "loss": 0.7536,
+      "step": 2300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013183371972091523,
+      "loss": 0.803,
+      "step": 2301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013177832728490473,
+      "loss": 0.8022,
+      "step": 2302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013172292400086884,
+      "loss": 0.7154,
+      "step": 2303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316675098877203,
+      "loss": 0.8141,
+      "step": 2304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013161208496437557,
+      "loss": 0.8074,
+      "step": 2305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013155664924975483,
+      "loss": 0.7839,
+      "step": 2306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013150120276278195,
+      "loss": 0.8597,
+      "step": 2307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001314457455223844,
+      "loss": 0.7703,
+      "step": 2308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013139027754749345,
+      "loss": 0.7895,
+      "step": 2309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013133479885704382,
+      "loss": 0.8439,
+      "step": 2310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013127930946997416,
+      "loss": 0.8079,
+      "step": 2311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013122380940522655,
+      "loss": 0.8188,
+      "step": 2312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001311682986817468,
+      "loss": 0.776,
+      "step": 2313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013111277731848442,
+      "loss": 0.8232,
+      "step": 2314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001310572453343924,
+      "loss": 0.8081,
+      "step": 2315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013100170274842757,
+      "loss": 0.8468,
+      "step": 2316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013094614957955008,
+      "loss": 0.7942,
+      "step": 2317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013089058584672404,
+      "loss": 0.842,
+      "step": 2318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013083501156891688,
+      "loss": 0.8769,
+      "step": 2319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013077942676509982,
+      "loss": 0.809,
+      "step": 2320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013072383145424754,
+      "loss": 0.7931,
+      "step": 2321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013066822565533837,
+      "loss": 0.8171,
+      "step": 2322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001306126093873542,
+      "loss": 0.2618,
+      "step": 2323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013055698266928054,
+      "loss": 0.7698,
+      "step": 2324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013050134552010644,
+      "loss": 0.8313,
+      "step": 2325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001304456979588245,
+      "loss": 0.7716,
+      "step": 2326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013039004000443084,
+      "loss": 0.8422,
+      "step": 2327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001303343716759252,
+      "loss": 0.772,
+      "step": 2328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013027869299231078,
+      "loss": 0.818,
+      "step": 2329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013022300397259445,
+      "loss": 0.8118,
+      "step": 2330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013016730463578647,
+      "loss": 0.7545,
+      "step": 2331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011159500090068,
+      "loss": 0.8073,
+      "step": 2332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013005587508695444,
+      "loss": 0.7803,
+      "step": 2333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013000014491296855,
+      "loss": 0.8549,
+      "step": 2334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012994440449796746,
+      "loss": 0.7225,
+      "step": 2335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012988865386097897,
+      "loss": 0.7969,
+      "step": 2336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298328930210344,
+      "loss": 0.7975,
+      "step": 2337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012977712199716866,
+      "loss": 0.8263,
+      "step": 2338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012972134080842,
+      "loss": 0.2631,
+      "step": 2339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001296655494738302,
+      "loss": 0.8505,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012960974801244453,
+      "loss": 0.8164,
+      "step": 2341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012955393644331162,
+      "loss": 0.8321,
+      "step": 2342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001294981147854837,
+      "loss": 0.8391,
+      "step": 2343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012944228305801625,
+      "loss": 0.2872,
+      "step": 2344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293864412799684,
+      "loss": 0.8056,
+      "step": 2345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012933058947040257,
+      "loss": 0.82,
+      "step": 2346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012927472764838463,
+      "loss": 0.787,
+      "step": 2347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012921885583298386,
+      "loss": 0.8436,
+      "step": 2348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012916297404327303,
+      "loss": 0.7944,
+      "step": 2349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012910708229832828,
+      "loss": 0.7974,
+      "step": 2350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129051180617229,
+      "loss": 0.8679,
+      "step": 2351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012899526901905822,
+      "loss": 0.8115,
+      "step": 2352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012893934752290215,
+      "loss": 0.818,
+      "step": 2353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012888341614785052,
+      "loss": 0.8554,
+      "step": 2354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012882747491299633,
+      "loss": 0.7279,
+      "step": 2355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012877152383743604,
+      "loss": 0.8127,
+      "step": 2356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012871556294026935,
+      "loss": 0.7709,
+      "step": 2357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012865959224059943,
+      "loss": 0.7732,
+      "step": 2358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012860361175753274,
+      "loss": 0.7394,
+      "step": 2359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012854762151017903,
+      "loss": 0.8197,
+      "step": 2360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012849162151765155,
+      "loss": 0.8105,
+      "step": 2361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001284356117990667,
+      "loss": 0.8316,
+      "step": 2362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012837959237354423,
+      "loss": 0.8165,
+      "step": 2363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012832356326020734,
+      "loss": 0.8033,
+      "step": 2364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012826752447818234,
+      "loss": 0.7802,
+      "step": 2365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012821147604659901,
+      "loss": 0.7673,
+      "step": 2366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012815541798459032,
+      "loss": 0.2618,
+      "step": 2367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012809935031129257,
+      "loss": 0.778,
+      "step": 2368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001280432730458453,
+      "loss": 0.8014,
+      "step": 2369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012798718620739142,
+      "loss": 0.7867,
+      "step": 2370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012793108981507694,
+      "loss": 0.8294,
+      "step": 2371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012787498388805132,
+      "loss": 0.8265,
+      "step": 2372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001278188684454672,
+      "loss": 0.787,
+      "step": 2373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012776274350648038,
+      "loss": 0.7729,
+      "step": 2374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012770660909025008,
+      "loss": 0.8005,
+      "step": 2375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012765046521593854,
+      "loss": 0.829,
+      "step": 2376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012759431190271142,
+      "loss": 0.7831,
+      "step": 2377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012753814916973753,
+      "loss": 0.7454,
+      "step": 2378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012748197703618883,
+      "loss": 0.8754,
+      "step": 2379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012742579552124062,
+      "loss": 0.831,
+      "step": 2380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012736960464407125,
+      "loss": 0.8454,
+      "step": 2381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012731340442386245,
+      "loss": 0.7437,
+      "step": 2382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012725719487979894,
+      "loss": 0.8296,
+      "step": 2383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012720097603106882,
+      "loss": 0.7101,
+      "step": 2384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012714474789686316,
+      "loss": 0.7951,
+      "step": 2385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001270885104963764,
+      "loss": 0.8218,
+      "step": 2386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000127032263848806,
+      "loss": 0.8112,
+      "step": 2387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001269760079733526,
+      "loss": 0.7972,
+      "step": 2388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001269197428892201,
+      "loss": 0.779,
+      "step": 2389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012686346861561537,
+      "loss": 0.8068,
+      "step": 2390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012680718517174862,
+      "loss": 0.808,
+      "step": 2391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012675089257683294,
+      "loss": 0.854,
+      "step": 2392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012669459085008476,
+      "loss": 0.7886,
+      "step": 2393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012663828001072355,
+      "loss": 0.7885,
+      "step": 2394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012658196007797186,
+      "loss": 0.796,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012652563107105538,
+      "loss": 0.7969,
+      "step": 2396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264692930092029,
+      "loss": 0.7668,
+      "step": 2397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012641294591164633,
+      "loss": 0.7824,
+      "step": 2398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012635658979762057,
+      "loss": 0.8118,
+      "step": 2399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012630022468636368,
+      "loss": 0.8125,
+      "step": 2400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012624385059711675,
+      "loss": 0.8703,
+      "step": 2401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012618746754912397,
+      "loss": 0.8509,
+      "step": 2402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012613107556163256,
+      "loss": 0.747,
+      "step": 2403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012607467465389276,
+      "loss": 0.797,
+      "step": 2404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012601826484515797,
+      "loss": 0.8216,
+      "step": 2405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012596184615468452,
+      "loss": 0.7735,
+      "step": 2406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001259054186017318,
+      "loss": 0.2316,
+      "step": 2407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001258489822055622,
+      "loss": 0.7933,
+      "step": 2408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012579253698544125,
+      "loss": 0.2735,
+      "step": 2409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012573608296063732,
+      "loss": 0.7932,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012567962015042191,
+      "loss": 0.7591,
+      "step": 2411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012562314857406944,
+      "loss": 0.7866,
+      "step": 2412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255666682508574,
+      "loss": 0.8034,
+      "step": 2413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012551017920006623,
+      "loss": 0.849,
+      "step": 2414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012545368144097928,
+      "loss": 0.7779,
+      "step": 2415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012539717499288302,
+      "loss": 0.2753,
+      "step": 2416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012534065987506674,
+      "loss": 0.7098,
+      "step": 2417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012528413610682275,
+      "loss": 0.7858,
+      "step": 2418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012522760370744636,
+      "loss": 0.8265,
+      "step": 2419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012517106269623577,
+      "loss": 0.7733,
+      "step": 2420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012511451309249214,
+      "loss": 0.8169,
+      "step": 2421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012505795491551952,
+      "loss": 0.8152,
+      "step": 2422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012500138818462498,
+      "loss": 0.8079,
+      "step": 2423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012494481291911837,
+      "loss": 0.7889,
+      "step": 2424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248882291383126,
+      "loss": 0.7898,
+      "step": 2425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248316368615234,
+      "loss": 0.8231,
+      "step": 2426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001247750361080694,
+      "loss": 0.8028,
+      "step": 2427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001247184268972722,
+      "loss": 0.8505,
+      "step": 2428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001246618092484562,
+      "loss": 0.8526,
+      "step": 2429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012460518318094873,
+      "loss": 0.8321,
+      "step": 2430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012454854871407994,
+      "loss": 0.8108,
+      "step": 2431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244919058671829,
+      "loss": 0.7656,
+      "step": 2432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012443525465959354,
+      "loss": 0.8127,
+      "step": 2433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012437859511065064,
+      "loss": 0.838,
+      "step": 2434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012432192723969577,
+      "loss": 0.7563,
+      "step": 2435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012426525106607345,
+      "loss": 0.8143,
+      "step": 2436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012420856660913093,
+      "loss": 0.8004,
+      "step": 2437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012415187388821835,
+      "loss": 0.8271,
+      "step": 2438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012409517292268863,
+      "loss": 0.7526,
+      "step": 2439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001240384637318975,
+      "loss": 0.7367,
+      "step": 2440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012398174633520363,
+      "loss": 0.7555,
+      "step": 2441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012392502075196825,
+      "loss": 0.776,
+      "step": 2442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001238682870015556,
+      "loss": 0.8112,
+      "step": 2443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012381154510333263,
+      "loss": 0.8199,
+      "step": 2444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012375479507666895,
+      "loss": 0.7863,
+      "step": 2445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012369803694093725,
+      "loss": 0.8327,
+      "step": 2446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001236412707155127,
+      "loss": 0.7489,
+      "step": 2447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012358449641977333,
+      "loss": 0.8155,
+      "step": 2448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001235277140730999,
+      "loss": 0.7671,
+      "step": 2449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012347092369487602,
+      "loss": 0.7917,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012341412530448792,
+      "loss": 0.7549,
+      "step": 2451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012335731892132462,
+      "loss": 0.794,
+      "step": 2452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001233005045647779,
+      "loss": 0.8174,
+      "step": 2453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012324368225424213,
+      "loss": 0.7743,
+      "step": 2454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012318685200911458,
+      "loss": 0.8766,
+      "step": 2455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012313001384879507,
+      "loss": 0.8332,
+      "step": 2456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012307316779268623,
+      "loss": 0.2972,
+      "step": 2457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012301631386019333,
+      "loss": 0.8245,
+      "step": 2458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012295945207072437,
+      "loss": 0.2613,
+      "step": 2459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012290258244368997,
+      "loss": 0.8058,
+      "step": 2460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012284570499850346,
+      "loss": 0.8314,
+      "step": 2461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012278881975458084,
+      "loss": 0.8312,
+      "step": 2462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012273192673134078,
+      "loss": 0.8109,
+      "step": 2463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012267502594820464,
+      "loss": 0.2534,
+      "step": 2464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012261811742459632,
+      "loss": 0.8736,
+      "step": 2465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012256120117994245,
+      "loss": 0.7315,
+      "step": 2466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012250427723367225,
+      "loss": 0.8193,
+      "step": 2467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012244734560521765,
+      "loss": 0.773,
+      "step": 2468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001223904063140131,
+      "loss": 0.7823,
+      "step": 2469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012233345937949573,
+      "loss": 0.269,
+      "step": 2470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012227650482110524,
+      "loss": 0.7815,
+      "step": 2471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012221954265828397,
+      "loss": 0.2514,
+      "step": 2472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012216257291047686,
+      "loss": 0.7869,
+      "step": 2473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012210559559713136,
+      "loss": 0.813,
+      "step": 2474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012204861073769762,
+      "loss": 0.7741,
+      "step": 2475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012199161835162826,
+      "loss": 0.8107,
+      "step": 2476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012193461845837855,
+      "loss": 0.8068,
+      "step": 2477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012187761107740627,
+      "loss": 0.7857,
+      "step": 2478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001218205962281718,
+      "loss": 0.7943,
+      "step": 2479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000121763573930138,
+      "loss": 0.8054,
+      "step": 2480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012170654420277035,
+      "loss": 0.8222,
+      "step": 2481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012164950706553679,
+      "loss": 0.8281,
+      "step": 2482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001215924625379079,
+      "loss": 0.8317,
+      "step": 2483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001215354106393567,
+      "loss": 0.8109,
+      "step": 2484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012147835138935868,
+      "loss": 0.8985,
+      "step": 2485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012142128480739195,
+      "loss": 0.7737,
+      "step": 2486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012136421091293708,
+      "loss": 0.8227,
+      "step": 2487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012130712972547711,
+      "loss": 0.8265,
+      "step": 2488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012125004126449762,
+      "loss": 0.7939,
+      "step": 2489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001211929455494866,
+      "loss": 0.8366,
+      "step": 2490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012113584259993458,
+      "loss": 0.7872,
+      "step": 2491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012107873243533452,
+      "loss": 0.7706,
+      "step": 2492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012102161507518189,
+      "loss": 0.7504,
+      "step": 2493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012096449053897457,
+      "loss": 0.8388,
+      "step": 2494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012090735884621289,
+      "loss": 0.7577,
+      "step": 2495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012085022001639972,
+      "loss": 0.8123,
+      "step": 2496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012079307406904019,
+      "loss": 0.8463,
+      "step": 2497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012073592102364201,
+      "loss": 0.8911,
+      "step": 2498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012067876089971524,
+      "loss": 0.7199,
+      "step": 2499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001206215937167724,
+      "loss": 0.7998,
+      "step": 2500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012056441949432833,
+      "loss": 0.2796,
+      "step": 2501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001205072382519004,
+      "loss": 0.8202,
+      "step": 2502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012045005000900833,
+      "loss": 0.8279,
+      "step": 2503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012039285478517417,
+      "loss": 0.2644,
+      "step": 2504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001203356525999224,
+      "loss": 0.7745,
+      "step": 2505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012027844347277988,
+      "loss": 0.8174,
+      "step": 2506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012022122742327586,
+      "loss": 0.827,
+      "step": 2507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012016400447094191,
+      "loss": 0.7455,
+      "step": 2508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012010677463531197,
+      "loss": 0.7844,
+      "step": 2509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012004953793592233,
+      "loss": 0.8003,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011999229439231161,
+      "loss": 0.7893,
+      "step": 2511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011993504402402082,
+      "loss": 0.8081,
+      "step": 2512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011987778685059323,
+      "loss": 0.7656,
+      "step": 2513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011982052289157451,
+      "loss": 0.7614,
+      "step": 2514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011976325216651254,
+      "loss": 0.7622,
+      "step": 2515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011970597469495759,
+      "loss": 0.8581,
+      "step": 2516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011964869049646226,
+      "loss": 0.7813,
+      "step": 2517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011959139959058134,
+      "loss": 0.8139,
+      "step": 2518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000119534101996872,
+      "loss": 0.7876,
+      "step": 2519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011947679773489367,
+      "loss": 0.7752,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011941948682420802,
+      "loss": 0.2615,
+      "step": 2521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011936216928437905,
+      "loss": 0.7866,
+      "step": 2522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000119304845134973,
+      "loss": 0.7921,
+      "step": 2523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001192475143955583,
+      "loss": 0.8372,
+      "step": 2524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011919017708570575,
+      "loss": 0.8427,
+      "step": 2525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011913283322498831,
+      "loss": 0.7982,
+      "step": 2526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011907548283298118,
+      "loss": 0.7915,
+      "step": 2527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011901812592926184,
+      "loss": 0.2924,
+      "step": 2528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011896076253340994,
+      "loss": 0.8208,
+      "step": 2529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011890339266500736,
+      "loss": 0.2844,
+      "step": 2530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011884601634363816,
+      "loss": 0.7886,
+      "step": 2531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011878863358888874,
+      "loss": 0.7658,
+      "step": 2532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011873124442034751,
+      "loss": 0.7556,
+      "step": 2533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011867384885760517,
+      "loss": 0.7939,
+      "step": 2534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011861644692025462,
+      "loss": 0.7888,
+      "step": 2535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011855903862789086,
+      "loss": 0.7511,
+      "step": 2536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011850162400011113,
+      "loss": 0.8342,
+      "step": 2537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011844420305651478,
+      "loss": 0.824,
+      "step": 2538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011838677581670339,
+      "loss": 0.8623,
+      "step": 2539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011832934230028057,
+      "loss": 0.899,
+      "step": 2540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011827190252685217,
+      "loss": 0.7862,
+      "step": 2541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011821445651602616,
+      "loss": 0.8366,
+      "step": 2542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011815700428741262,
+      "loss": 0.8138,
+      "step": 2543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011809954586062377,
+      "loss": 0.7939,
+      "step": 2544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011804208125527392,
+      "loss": 0.7591,
+      "step": 2545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011798461049097955,
+      "loss": 0.7487,
+      "step": 2546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011792713358735912,
+      "loss": 0.7868,
+      "step": 2547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011786965056403334,
+      "loss": 0.7778,
+      "step": 2548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011781216144062488,
+      "loss": 0.8163,
+      "step": 2549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011775466623675855,
+      "loss": 0.7574,
+      "step": 2550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011769716497206128,
+      "loss": 0.7702,
+      "step": 2551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011763965766616194,
+      "loss": 0.8178,
+      "step": 2552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011758214433869163,
+      "loss": 0.7779,
+      "step": 2553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011752462500928334,
+      "loss": 0.7456,
+      "step": 2554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011746709969757221,
+      "loss": 0.782,
+      "step": 2555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011740956842319542,
+      "loss": 0.8018,
+      "step": 2556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011735203120579217,
+      "loss": 0.7818,
+      "step": 2557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011729448806500363,
+      "loss": 0.2564,
+      "step": 2558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011723693902047307,
+      "loss": 0.7724,
+      "step": 2559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001171793840918458,
+      "loss": 0.8287,
+      "step": 2560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011712182329876902,
+      "loss": 0.7595,
+      "step": 2561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011706425666089202,
+      "loss": 0.7764,
+      "step": 2562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011700668419786609,
+      "loss": 0.7998,
+      "step": 2563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011694910592934447,
+      "loss": 0.7649,
+      "step": 2564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011689152187498236,
+      "loss": 0.7506,
+      "step": 2565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000116833932054437,
+      "loss": 0.7995,
+      "step": 2566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011677633648736761,
+      "loss": 0.793,
+      "step": 2567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011671873519343524,
+      "loss": 0.8133,
+      "step": 2568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011666112819230309,
+      "loss": 0.7914,
+      "step": 2569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011660351550363612,
+      "loss": 0.7733,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011654589714710138,
+      "loss": 0.8406,
+      "step": 2571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011648827314236772,
+      "loss": 0.2589,
+      "step": 2572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011643064350910609,
+      "loss": 0.8314,
+      "step": 2573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011637300826698914,
+      "loss": 0.7792,
+      "step": 2574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011631536743569163,
+      "loss": 0.7705,
+      "step": 2575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011625772103489014,
+      "loss": 0.7741,
+      "step": 2576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011620006908426315,
+      "loss": 0.7727,
+      "step": 2577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011614241160349106,
+      "loss": 0.7924,
+      "step": 2578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011608474861225615,
+      "loss": 0.8174,
+      "step": 2579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011602708013024255,
+      "loss": 0.7699,
+      "step": 2580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011596940617713632,
+      "loss": 0.2497,
+      "step": 2581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011591172677262538,
+      "loss": 0.7705,
+      "step": 2582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011585404193639944,
+      "loss": 0.7553,
+      "step": 2583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011579635168815013,
+      "loss": 0.7864,
+      "step": 2584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011573865604757095,
+      "loss": 0.8488,
+      "step": 2585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011568095503435717,
+      "loss": 0.8003,
+      "step": 2586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011562324866820591,
+      "loss": 0.7979,
+      "step": 2587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011556553696881614,
+      "loss": 0.7634,
+      "step": 2588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011550781995588869,
+      "loss": 0.7763,
+      "step": 2589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001154500976491261,
+      "loss": 0.808,
+      "step": 2590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011539237006823281,
+      "loss": 0.7669,
+      "step": 2591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000115334637232915,
+      "loss": 0.813,
+      "step": 2592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001152768991628807,
+      "loss": 0.831,
+      "step": 2593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011521915587783968,
+      "loss": 0.7777,
+      "step": 2594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011516140739750351,
+      "loss": 0.83,
+      "step": 2595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011510365374158556,
+      "loss": 0.7437,
+      "step": 2596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011504589492980087,
+      "loss": 0.7827,
+      "step": 2597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011498813098186638,
+      "loss": 0.811,
+      "step": 2598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011493036191750066,
+      "loss": 0.8373,
+      "step": 2599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011487258775642406,
+      "loss": 0.2642,
+      "step": 2600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011481480851835878,
+      "loss": 0.7798,
+      "step": 2601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011475702422302854,
+      "loss": 0.8265,
+      "step": 2602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011469923489015901,
+      "loss": 0.762,
+      "step": 2603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011464144053947739,
+      "loss": 0.7667,
+      "step": 2604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011458364119071276,
+      "loss": 0.7919,
+      "step": 2605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011452583686359577,
+      "loss": 0.7503,
+      "step": 2606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011446802757785884,
+      "loss": 0.8375,
+      "step": 2607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011441021335323608,
+      "loss": 0.714,
+      "step": 2608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011435239420946321,
+      "loss": 0.8152,
+      "step": 2609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011429457016627783,
+      "loss": 0.8331,
+      "step": 2610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011423674124341892,
+      "loss": 0.883,
+      "step": 2611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011417890746062738,
+      "loss": 0.8214,
+      "step": 2612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011412106883764561,
+      "loss": 0.7761,
+      "step": 2613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011406322539421777,
+      "loss": 0.8295,
+      "step": 2614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001140053771500896,
+      "loss": 0.7612,
+      "step": 2615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011394752412500849,
+      "loss": 0.7943,
+      "step": 2616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011388966633872346,
+      "loss": 0.7515,
+      "step": 2617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011383180381098517,
+      "loss": 0.7399,
+      "step": 2618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011377393656154594,
+      "loss": 0.7632,
+      "step": 2619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011371606461015958,
+      "loss": 0.8311,
+      "step": 2620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011365818797658164,
+      "loss": 0.8414,
+      "step": 2621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011360030668056913,
+      "loss": 0.7923,
+      "step": 2622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011354242074188082,
+      "loss": 0.7968,
+      "step": 2623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011348453018027692,
+      "loss": 0.778,
+      "step": 2624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011342663501551926,
+      "loss": 0.8355,
+      "step": 2625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011336873526737132,
+      "loss": 0.8219,
+      "step": 2626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113310830955598,
+      "loss": 0.844,
+      "step": 2627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011325292209996592,
+      "loss": 0.8289,
+      "step": 2628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011319500872024305,
+      "loss": 0.7545,
+      "step": 2629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011313709083619913,
+      "loss": 0.8102,
+      "step": 2630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011307916846760527,
+      "loss": 0.8461,
+      "step": 2631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001130212416342342,
+      "loss": 0.7804,
+      "step": 2632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011296331035586015,
+      "loss": 0.7822,
+      "step": 2633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001129053746522588,
+      "loss": 0.8023,
+      "step": 2634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011284743454320747,
+      "loss": 0.7722,
+      "step": 2635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011278949004848489,
+      "loss": 0.7527,
+      "step": 2636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011273154118787132,
+      "loss": 0.8129,
+      "step": 2637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011267358798114848,
+      "loss": 0.7214,
+      "step": 2638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011261563044809964,
+      "loss": 0.8217,
+      "step": 2639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011255766860850947,
+      "loss": 0.8103,
+      "step": 2640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011249970248216411,
+      "loss": 0.8163,
+      "step": 2641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001124417320888513,
+      "loss": 0.8549,
+      "step": 2642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011238375744836003,
+      "loss": 0.7936,
+      "step": 2643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011232577858048092,
+      "loss": 0.85,
+      "step": 2644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001122677955050059,
+      "loss": 0.8369,
+      "step": 2645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011220980824172842,
+      "loss": 0.7887,
+      "step": 2646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011215181681044333,
+      "loss": 0.7854,
+      "step": 2647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011209382123094689,
+      "loss": 0.7988,
+      "step": 2648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011203582152303685,
+      "loss": 0.7966,
+      "step": 2649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011197781770651222,
+      "loss": 0.7694,
+      "step": 2650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011191980980117358,
+      "loss": 0.7616,
+      "step": 2651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011186179782682277,
+      "loss": 0.8194,
+      "step": 2652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011180378180326312,
+      "loss": 0.8557,
+      "step": 2653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011174576175029927,
+      "loss": 0.8255,
+      "step": 2654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011168773768773733,
+      "loss": 0.8157,
+      "step": 2655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011162970963538462,
+      "loss": 0.7422,
+      "step": 2656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011157167761304999,
+      "loss": 0.2556,
+      "step": 2657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011151364164054354,
+      "loss": 0.8753,
+      "step": 2658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011145560173767675,
+      "loss": 0.7617,
+      "step": 2659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011139755792426247,
+      "loss": 0.787,
+      "step": 2660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011133951022011481,
+      "loss": 0.797,
+      "step": 2661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001112814586450493,
+      "loss": 0.7932,
+      "step": 2662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011122340321888271,
+      "loss": 0.7665,
+      "step": 2663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011116534396143319,
+      "loss": 0.8022,
+      "step": 2664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011110728089252019,
+      "loss": 0.783,
+      "step": 2665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011104921403196434,
+      "loss": 0.8307,
+      "step": 2666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011099114339958776,
+      "loss": 0.297,
+      "step": 2667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011093306901521371,
+      "loss": 0.8013,
+      "step": 2668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011087499089866687,
+      "loss": 0.8053,
+      "step": 2669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000110816909069773,
+      "loss": 0.8326,
+      "step": 2670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001107588235483593,
+      "loss": 0.7605,
+      "step": 2671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001107007343542541,
+      "loss": 0.7716,
+      "step": 2672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011064264150728711,
+      "loss": 0.8354,
+      "step": 2673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011058454502728923,
+      "loss": 0.7808,
+      "step": 2674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011052644493409255,
+      "loss": 0.2504,
+      "step": 2675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011046834124753045,
+      "loss": 0.8193,
+      "step": 2676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011041023398743755,
+      "loss": 0.7691,
+      "step": 2677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011035212317364963,
+      "loss": 0.8448,
+      "step": 2678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011029400882600374,
+      "loss": 0.2541,
+      "step": 2679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011023589096433815,
+      "loss": 0.8068,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011017776960849222,
+      "loss": 0.8514,
+      "step": 2681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011011964477830663,
+      "loss": 0.8189,
+      "step": 2682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001100615164936232,
+      "loss": 0.7487,
+      "step": 2683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011000338477428487,
+      "loss": 0.8558,
+      "step": 2684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010994524964013584,
+      "loss": 0.7638,
+      "step": 2685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010988711111102144,
+      "loss": 0.8034,
+      "step": 2686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010982896920678818,
+      "loss": 0.8309,
+      "step": 2687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010977082394728364,
+      "loss": 0.8223,
+      "step": 2688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010971267535235665,
+      "loss": 0.7918,
+      "step": 2689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010965452344185714,
+      "loss": 0.8057,
+      "step": 2690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010959636823563614,
+      "loss": 0.7649,
+      "step": 2691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010953820975354589,
+      "loss": 0.8126,
+      "step": 2692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010948004801543964,
+      "loss": 0.782,
+      "step": 2693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010942188304117182,
+      "loss": 0.7684,
+      "step": 2694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010936371485059791,
+      "loss": 0.8233,
+      "step": 2695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001093055434635746,
+      "loss": 0.7656,
+      "step": 2696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001092473688999595,
+      "loss": 0.7592,
+      "step": 2697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010918919117961148,
+      "loss": 0.7929,
+      "step": 2698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010913101032239039,
+      "loss": 0.7401,
+      "step": 2699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010907282634815714,
+      "loss": 0.7671,
+      "step": 2700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010901463927677378,
+      "loss": 0.7366,
+      "step": 2701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001089564491281033,
+      "loss": 0.8032,
+      "step": 2702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010889825592200989,
+      "loss": 0.8449,
+      "step": 2703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010884005967835864,
+      "loss": 0.7262,
+      "step": 2704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001087818604170158,
+      "loss": 0.8477,
+      "step": 2705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010872365815784853,
+      "loss": 0.7847,
+      "step": 2706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001086654529207251,
+      "loss": 0.7689,
+      "step": 2707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010860724472551482,
+      "loss": 0.8035,
+      "step": 2708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010854903359208787,
+      "loss": 0.8517,
+      "step": 2709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010849081954031562,
+      "loss": 0.2575,
+      "step": 2710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010843260259007027,
+      "loss": 0.7653,
+      "step": 2711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010837438276122509,
+      "loss": 0.7965,
+      "step": 2712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010831616007365435,
+      "loss": 0.8042,
+      "step": 2713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 0.7848,
+      "step": 2714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010819970620183797,
+      "loss": 0.795,
+      "step": 2715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010814147505734567,
+      "loss": 0.8103,
+      "step": 2716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010808324113363448,
+      "loss": 0.7808,
+      "step": 2717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010802500445058341,
+      "loss": 0.8502,
+      "step": 2718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010796676502807248,
+      "loss": 0.7739,
+      "step": 2719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010790852288598261,
+      "loss": 0.8262,
+      "step": 2720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010785027804419568,
+      "loss": 0.7458,
+      "step": 2721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010779203052259446,
+      "loss": 0.7304,
+      "step": 2722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010773378034106263,
+      "loss": 0.8221,
+      "step": 2723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010767552751948482,
+      "loss": 0.7698,
+      "step": 2724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001076172720777465,
+      "loss": 0.8127,
+      "step": 2725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010755901403573411,
+      "loss": 0.7998,
+      "step": 2726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001075007534133349,
+      "loss": 0.7482,
+      "step": 2727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010744249023043706,
+      "loss": 0.8095,
+      "step": 2728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010738422450692965,
+      "loss": 0.7718,
+      "step": 2729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010732595626270252,
+      "loss": 0.7279,
+      "step": 2730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010726768551764652,
+      "loss": 0.2563,
+      "step": 2731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010720941229165322,
+      "loss": 0.7868,
+      "step": 2732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010715113660461512,
+      "loss": 0.809,
+      "step": 2733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010709285847642547,
+      "loss": 0.8595,
+      "step": 2734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010703457792697848,
+      "loss": 0.7647,
+      "step": 2735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010697629497616907,
+      "loss": 0.7751,
+      "step": 2736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010691800964389309,
+      "loss": 0.7871,
+      "step": 2737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010685972195004712,
+      "loss": 0.8315,
+      "step": 2738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001068014319145285,
+      "loss": 0.7979,
+      "step": 2739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010674313955723554,
+      "loss": 0.7812,
+      "step": 2740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010668484489806718,
+      "loss": 0.8326,
+      "step": 2741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010662654795692326,
+      "loss": 0.7943,
+      "step": 2742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010656824875370428,
+      "loss": 0.7766,
+      "step": 2743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010650994730831165,
+      "loss": 0.2673,
+      "step": 2744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010645164364064739,
+      "loss": 0.7045,
+      "step": 2745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010639333777061443,
+      "loss": 0.8502,
+      "step": 2746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010633502971811636,
+      "loss": 0.8415,
+      "step": 2747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010627671950305751,
+      "loss": 0.7983,
+      "step": 2748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010621840714534304,
+      "loss": 0.2604,
+      "step": 2749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010616009266487871,
+      "loss": 0.7569,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010610177608157117,
+      "loss": 0.7805,
+      "step": 2751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010604345741532755,
+      "loss": 0.7807,
+      "step": 2752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010598513668605597,
+      "loss": 0.7963,
+      "step": 2753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010592681391366501,
+      "loss": 0.8303,
+      "step": 2754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010586848911806414,
+      "loss": 0.8039,
+      "step": 2755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010581016231916339,
+      "loss": 0.7978,
+      "step": 2756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001057518335368735,
+      "loss": 0.8351,
+      "step": 2757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010569350279110597,
+      "loss": 0.8212,
+      "step": 2758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010563517010177284,
+      "loss": 0.7766,
+      "step": 2759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010557683548878691,
+      "loss": 0.2481,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010551849897206161,
+      "loss": 0.7861,
+      "step": 2761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010546016057151099,
+      "loss": 0.8056,
+      "step": 2762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010540182030704979,
+      "loss": 0.7826,
+      "step": 2763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010534347819859334,
+      "loss": 0.7783,
+      "step": 2764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010528513426605768,
+      "loss": 0.7872,
+      "step": 2765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010522678852935936,
+      "loss": 0.7397,
+      "step": 2766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010516844100841565,
+      "loss": 0.8016,
+      "step": 2767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010511009172314434,
+      "loss": 0.7292,
+      "step": 2768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010505174069346389,
+      "loss": 0.7011,
+      "step": 2769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001049933879392933,
+      "loss": 0.8029,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010493503348055223,
+      "loss": 0.7622,
+      "step": 2771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010487667733716086,
+      "loss": 0.8171,
+      "step": 2772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010481831952903996,
+      "loss": 0.7875,
+      "step": 2773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010475996007611089,
+      "loss": 0.784,
+      "step": 2774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010470159899829553,
+      "loss": 0.767,
+      "step": 2775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010464323631551636,
+      "loss": 0.7713,
+      "step": 2776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010458487204769638,
+      "loss": 0.7848,
+      "step": 2777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010452650621475915,
+      "loss": 0.7496,
+      "step": 2778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010446813883662873,
+      "loss": 0.3096,
+      "step": 2779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010440976993322976,
+      "loss": 0.7618,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010435139952448738,
+      "loss": 0.7536,
+      "step": 2781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010429302763032718,
+      "loss": 0.7812,
+      "step": 2782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010423465427067536,
+      "loss": 0.7904,
+      "step": 2783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010417627946545856,
+      "loss": 0.8351,
+      "step": 2784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010411790323460392,
+      "loss": 0.8043,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010405952559803906,
+      "loss": 0.7554,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010400114657569213,
+      "loss": 0.7726,
+      "step": 2787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010394276618749168,
+      "loss": 0.8152,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010388438445336676,
+      "loss": 0.7261,
+      "step": 2789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010382600139324696,
+      "loss": 0.8005,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010376761702706212,
+      "loss": 0.8002,
+      "step": 2791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010370923137474278,
+      "loss": 0.8037,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010365084445621967,
+      "loss": 0.8147,
+      "step": 2793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010359245629142414,
+      "loss": 0.7965,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010353406690028787,
+      "loss": 0.7961,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010347567630274297,
+      "loss": 0.7881,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010341728451872206,
+      "loss": 0.7731,
+      "step": 2797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010335889156815797,
+      "loss": 0.8207,
+      "step": 2798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010330049747098414,
+      "loss": 0.8031,
+      "step": 2799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010324210224713422,
+      "loss": 0.7844,
+      "step": 2800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010318370591654236,
+      "loss": 0.8098,
+      "step": 2801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010312530849914306,
+      "loss": 0.7953,
+      "step": 2802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010306691001487119,
+      "loss": 0.8024,
+      "step": 2803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010300851048366196,
+      "loss": 0.7962,
+      "step": 2804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010295010992545095,
+      "loss": 0.7866,
+      "step": 2805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010289170836017416,
+      "loss": 0.7597,
+      "step": 2806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001028333058077678,
+      "loss": 0.2624,
+      "step": 2807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010277490228816853,
+      "loss": 0.7941,
+      "step": 2808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010271649782131325,
+      "loss": 0.7543,
+      "step": 2809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010265809242713928,
+      "loss": 0.7847,
+      "step": 2810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010259968612558419,
+      "loss": 0.2552,
+      "step": 2811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010254127893658587,
+      "loss": 0.7919,
+      "step": 2812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010248287088008255,
+      "loss": 0.8511,
+      "step": 2813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010242446197601268,
+      "loss": 0.2287,
+      "step": 2814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001023660522443151,
+      "loss": 0.7651,
+      "step": 2815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010230764170492884,
+      "loss": 0.2594,
+      "step": 2816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010224923037779326,
+      "loss": 0.7998,
+      "step": 2817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010219081828284796,
+      "loss": 0.8363,
+      "step": 2818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010213240544003281,
+      "loss": 0.8043,
+      "step": 2819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000102073991869288,
+      "loss": 0.853,
+      "step": 2820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010201557759055383,
+      "loss": 0.7447,
+      "step": 2821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010195716262377098,
+      "loss": 0.8359,
+      "step": 2822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010189874698888025,
+      "loss": 0.7947,
+      "step": 2823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010184033070582277,
+      "loss": 0.8061,
+      "step": 2824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010178191379453981,
+      "loss": 0.7904,
+      "step": 2825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010172349627497295,
+      "loss": 0.814,
+      "step": 2826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010166507816706386,
+      "loss": 0.802,
+      "step": 2827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010160665949075448,
+      "loss": 0.7797,
+      "step": 2828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010154824026598695,
+      "loss": 0.7934,
+      "step": 2829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010148982051270355,
+      "loss": 0.7565,
+      "step": 2830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010143140025084681,
+      "loss": 0.8284,
+      "step": 2831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010137297950035935,
+      "loss": 0.8101,
+      "step": 2832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010131455828118403,
+      "loss": 0.793,
+      "step": 2833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010125613661326383,
+      "loss": 0.7856,
+      "step": 2834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001011977145165419,
+      "loss": 0.8148,
+      "step": 2835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010113929201096152,
+      "loss": 0.7913,
+      "step": 2836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010108086911646612,
+      "loss": 0.2529,
+      "step": 2837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010102244585299926,
+      "loss": 0.7149,
+      "step": 2838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010096402224050465,
+      "loss": 0.7832,
+      "step": 2839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010090559829892608,
+      "loss": 0.7708,
+      "step": 2840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010084717404820745,
+      "loss": 0.7797,
+      "step": 2841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010078874950829284,
+      "loss": 0.7669,
+      "step": 2842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010073032469912632,
+      "loss": 0.8121,
+      "step": 2843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010067189964065211,
+      "loss": 0.8102,
+      "step": 2844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010061347435281456,
+      "loss": 0.7736,
+      "step": 2845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010055504885555797,
+      "loss": 0.2409,
+      "step": 2846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010049662316882689,
+      "loss": 0.7334,
+      "step": 2847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010043819731256574,
+      "loss": 0.7778,
+      "step": 2848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010037977130671916,
+      "loss": 0.775,
+      "step": 2849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001003213451712317,
+      "loss": 0.8298,
+      "step": 2850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010026291892604814,
+      "loss": 0.7869,
+      "step": 2851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010020449259111306,
+      "loss": 0.7767,
+      "step": 2852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001001460661863713,
+      "loss": 0.8257,
+      "step": 2853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010008763973176758,
+      "loss": 0.7699,
+      "step": 2854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010002921324724666,
+      "loss": 0.7814,
+      "step": 2855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997078675275334e-05,
+      "loss": 0.7459,
+      "step": 2856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991236026823242e-05,
+      "loss": 0.8168,
+      "step": 2857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985393381362871e-05,
+      "loss": 0.8313,
+      "step": 2858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979550740888694e-05,
+      "loss": 0.7959,
+      "step": 2859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97370810739519e-05,
+      "loss": 0.819,
+      "step": 2860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967865482876831e-05,
+      "loss": 0.2962,
+      "step": 2861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962022869328086e-05,
+      "loss": 0.7855,
+      "step": 2862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956180268743428e-05,
+      "loss": 0.8099,
+      "step": 2863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950337683117313e-05,
+      "loss": 0.2763,
+      "step": 2864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944495114444205e-05,
+      "loss": 0.7704,
+      "step": 2865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938652564718547e-05,
+      "loss": 0.7857,
+      "step": 2866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93281003593479e-05,
+      "loss": 0.81,
+      "step": 2867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92696753008737e-05,
+      "loss": 0.7442,
+      "step": 2868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921125049170718e-05,
+      "loss": 0.8099,
+      "step": 2869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915282595179256e-05,
+      "loss": 0.8048,
+      "step": 2870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909440170107393e-05,
+      "loss": 0.845,
+      "step": 2871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903597775949537e-05,
+      "loss": 0.77,
+      "step": 2872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897755414700075e-05,
+      "loss": 0.8229,
+      "step": 2873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891913088353392e-05,
+      "loss": 0.7823,
+      "step": 2874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886070798903851e-05,
+      "loss": 0.8041,
+      "step": 2875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880228548345813e-05,
+      "loss": 0.7996,
+      "step": 2876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87438633867362e-05,
+      "loss": 0.7753,
+      "step": 2877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868544171881599e-05,
+      "loss": 0.8123,
+      "step": 2878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862702049964068e-05,
+      "loss": 0.8446,
+      "step": 2879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856859974915324e-05,
+      "loss": 0.2828,
+      "step": 2880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851017948729645e-05,
+      "loss": 0.8245,
+      "step": 2881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845175973401306e-05,
+      "loss": 0.8085,
+      "step": 2882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839334050924553e-05,
+      "loss": 0.8182,
+      "step": 2883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833492183293615e-05,
+      "loss": 0.7418,
+      "step": 2884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827650372502708e-05,
+      "loss": 0.7432,
+      "step": 2885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82180862054602e-05,
+      "loss": 0.7764,
+      "step": 2886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815966929417724e-05,
+      "loss": 0.7939,
+      "step": 2887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810125301111977e-05,
+      "loss": 0.7669,
+      "step": 2888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804283737622906e-05,
+      "loss": 0.7872,
+      "step": 2889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798442240944621e-05,
+      "loss": 0.7386,
+      "step": 2890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792600813071203e-05,
+      "loss": 0.7909,
+      "step": 2891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786759455996721e-05,
+      "loss": 0.7308,
+      "step": 2892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780918171715208e-05,
+      "loss": 0.8494,
+      "step": 2893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775076962220678e-05,
+      "loss": 0.783,
+      "step": 2894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76923582950712e-05,
+      "loss": 0.2414,
+      "step": 2895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763394775568493e-05,
+      "loss": 0.8385,
+      "step": 2896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757553802398734e-05,
+      "loss": 0.2609,
+      "step": 2897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751712911991748e-05,
+      "loss": 0.8344,
+      "step": 2898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.745872106341414e-05,
+      "loss": 0.2632,
+      "step": 2899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740031387441583e-05,
+      "loss": 0.7848,
+      "step": 2900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734190757286075e-05,
+      "loss": 0.7528,
+      "step": 2901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728350217868679e-05,
+      "loss": 0.787,
+      "step": 2902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722509771183151e-05,
+      "loss": 0.7665,
+      "step": 2903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716669419223225e-05,
+      "loss": 0.2458,
+      "step": 2904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710829163982583e-05,
+      "loss": 0.7812,
+      "step": 2905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704989007454903e-05,
+      "loss": 0.8464,
+      "step": 2906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699148951633803e-05,
+      "loss": 0.7861,
+      "step": 2907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693308998512883e-05,
+      "loss": 0.7476,
+      "step": 2908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687469150085695e-05,
+      "loss": 0.7628,
+      "step": 2909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681629408345768e-05,
+      "loss": 0.7962,
+      "step": 2910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675789775286582e-05,
+      "loss": 0.7839,
+      "step": 2911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66995025290159e-05,
+      "loss": 0.7665,
+      "step": 2912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664110843184205e-05,
+      "loss": 0.7754,
+      "step": 2913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658271548127798e-05,
+      "loss": 0.7998,
+      "step": 2914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652432369725704e-05,
+      "loss": 0.7555,
+      "step": 2915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646593309971215e-05,
+      "loss": 0.8377,
+      "step": 2916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640754370857588e-05,
+      "loss": 0.8089,
+      "step": 2917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634915554378037e-05,
+      "loss": 0.7314,
+      "step": 2918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629076862525726e-05,
+      "loss": 0.2845,
+      "step": 2919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623238297293789e-05,
+      "loss": 0.8079,
+      "step": 2920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617399860675306e-05,
+      "loss": 0.7673,
+      "step": 2921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611561554663325e-05,
+      "loss": 0.2703,
+      "step": 2922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605723381250834e-05,
+      "loss": 0.7964,
+      "step": 2923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59988534243079e-05,
+      "loss": 0.799,
+      "step": 2924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594047440196097e-05,
+      "loss": 0.7793,
+      "step": 2925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588209676539613e-05,
+      "loss": 0.7637,
+      "step": 2926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582372053454149e-05,
+      "loss": 0.2404,
+      "step": 2927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576534572932467e-05,
+      "loss": 0.709,
+      "step": 2928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570697236967287e-05,
+      "loss": 0.8224,
+      "step": 2929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564860047551263e-05,
+      "loss": 0.7838,
+      "step": 2930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559023006677024e-05,
+      "loss": 0.7425,
+      "step": 2931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553186116337126e-05,
+      "loss": 0.7575,
+      "step": 2932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547349378524086e-05,
+      "loss": 0.7954,
+      "step": 2933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541512795230363e-05,
+      "loss": 0.7969,
+      "step": 2934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535676368448364e-05,
+      "loss": 0.8024,
+      "step": 2935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529840100170449e-05,
+      "loss": 0.7955,
+      "step": 2936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524003992388914e-05,
+      "loss": 0.7838,
+      "step": 2937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518168047096007e-05,
+      "loss": 0.7751,
+      "step": 2938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512332266283916e-05,
+      "loss": 0.7838,
+      "step": 2939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506496651944779e-05,
+      "loss": 0.7482,
+      "step": 2940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500661206070671e-05,
+      "loss": 0.7609,
+      "step": 2941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494825930653612e-05,
+      "loss": 0.8043,
+      "step": 2942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488990827685568e-05,
+      "loss": 0.7161,
+      "step": 2943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483155899158437e-05,
+      "loss": 0.8207,
+      "step": 2944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477321147064065e-05,
+      "loss": 0.7971,
+      "step": 2945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471486573394234e-05,
+      "loss": 0.6677,
+      "step": 2946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465652180140669e-05,
+      "loss": 0.831,
+      "step": 2947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459817969295024e-05,
+      "loss": 0.2738,
+      "step": 2948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453983942848904e-05,
+      "loss": 0.7739,
+      "step": 2949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448150102793844e-05,
+      "loss": 0.7651,
+      "step": 2950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442316451121311e-05,
+      "loss": 0.7832,
+      "step": 2951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43648298982272e-05,
+      "loss": 0.7803,
+      "step": 2952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430649720889407e-05,
+      "loss": 0.7389,
+      "step": 2953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424816646312648e-05,
+      "loss": 0.7696,
+      "step": 2954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418983768083662e-05,
+      "loss": 0.769,
+      "step": 2955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413151088193587e-05,
+      "loss": 0.7594,
+      "step": 2956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407318608633497e-05,
+      "loss": 0.7872,
+      "step": 2957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401486331394405e-05,
+      "loss": 0.8183,
+      "step": 2958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395654258467244e-05,
+      "loss": 0.7995,
+      "step": 2959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389822391842886e-05,
+      "loss": 0.7494,
+      "step": 2960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38399073351213e-05,
+      "loss": 0.7,
+      "step": 2961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378159285465697e-05,
+      "loss": 0.7922,
+      "step": 2962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372328049694251e-05,
+      "loss": 0.7437,
+      "step": 2963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366497028188366e-05,
+      "loss": 0.7808,
+      "step": 2964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360666222938561e-05,
+      "loss": 0.8063,
+      "step": 2965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354835635935262e-05,
+      "loss": 0.7534,
+      "step": 2966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349005269168838e-05,
+      "loss": 0.7892,
+      "step": 2967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343175124629573e-05,
+      "loss": 0.2683,
+      "step": 2968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337345204307676e-05,
+      "loss": 0.8122,
+      "step": 2969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331515510193283e-05,
+      "loss": 0.784,
+      "step": 2970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325686044276447e-05,
+      "loss": 0.7863,
+      "step": 2971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319856808547152e-05,
+      "loss": 0.7842,
+      "step": 2972
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314027804995293e-05,
+      "loss": 0.8281,
+      "step": 2973
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308199035610694e-05,
+      "loss": 0.7419,
+      "step": 2974
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302370502383095e-05,
+      "loss": 0.7609,
+      "step": 2975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296542207302155e-05,
+      "loss": 0.7273,
+      "step": 2976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290714152357458e-05,
+      "loss": 0.7729,
+      "step": 2977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284886339538495e-05,
+      "loss": 0.7604,
+      "step": 2978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279058770834679e-05,
+      "loss": 0.7481,
+      "step": 2979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273231448235348e-05,
+      "loss": 0.8067,
+      "step": 2980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267404373729746e-05,
+      "loss": 0.7993,
+      "step": 2981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261577549307036e-05,
+      "loss": 0.7831,
+      "step": 2982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255750976956295e-05,
+      "loss": 0.7886,
+      "step": 2983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24992465866651e-05,
+      "loss": 0.7715,
+      "step": 2984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244098596426591e-05,
+      "loss": 0.7141,
+      "step": 2985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238272792225352e-05,
+      "loss": 0.7744,
+      "step": 2986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232447248051521e-05,
+      "loss": 0.8196,
+      "step": 2987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22662196589374e-05,
+      "loss": 0.7853,
+      "step": 2988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220796947740556e-05,
+      "loss": 0.7731,
+      "step": 2989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214972195580434e-05,
+      "loss": 0.7882,
+      "step": 2990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20914771140174e-05,
+      "loss": 0.7866,
+      "step": 2991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203323497192753e-05,
+      "loss": 0.7692,
+      "step": 2992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197499554941662e-05,
+      "loss": 0.7423,
+      "step": 2993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.191675886636554e-05,
+      "loss": 0.8148,
+      "step": 2994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185852494265435e-05,
+      "loss": 0.8143,
+      "step": 2995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180029379816205e-05,
+      "loss": 0.7321,
+      "step": 2996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174206545276677e-05,
+      "loss": 0.7622,
+      "step": 2997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168383992634569e-05,
+      "loss": 0.7944,
+      "step": 2998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162561723877495e-05,
+      "loss": 0.7653,
+      "step": 2999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156739740992978e-05,
+      "loss": 0.7456,
+      "step": 3000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150918045968442e-05,
+      "loss": 0.7944,
+      "step": 3001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145096640791216e-05,
+      "loss": 0.7929,
+      "step": 3002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139275527448519e-05,
+      "loss": 0.8122,
+      "step": 3003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133454707927488e-05,
+      "loss": 0.7598,
+      "step": 3004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127634184215148e-05,
+      "loss": 0.8165,
+      "step": 3005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121813958298423e-05,
+      "loss": 0.7783,
+      "step": 3006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115994032164137e-05,
+      "loss": 0.8231,
+      "step": 3007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110174407799015e-05,
+      "loss": 0.7723,
+      "step": 3008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104355087189671e-05,
+      "loss": 0.7706,
+      "step": 3009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098536072322626e-05,
+      "loss": 0.7737,
+      "step": 3010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092717365184288e-05,
+      "loss": 0.7088,
+      "step": 3011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086898967760962e-05,
+      "loss": 0.776,
+      "step": 3012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081080882038855e-05,
+      "loss": 0.7443,
+      "step": 3013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075263110004051e-05,
+      "loss": 0.285,
+      "step": 3014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069445653642543e-05,
+      "loss": 0.8251,
+      "step": 3015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063628514940211e-05,
+      "loss": 0.2688,
+      "step": 3016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05781169588282e-05,
+      "loss": 0.7621,
+      "step": 3017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051995198456039e-05,
+      "loss": 0.7971,
+      "step": 3018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046179024645413e-05,
+      "loss": 0.7707,
+      "step": 3019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040363176436387e-05,
+      "loss": 0.8092,
+      "step": 3020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034547655814289e-05,
+      "loss": 0.7423,
+      "step": 3021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028732464764336e-05,
+      "loss": 0.8061,
+      "step": 3022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02291760527164e-05,
+      "loss": 0.8188,
+      "step": 3023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017103079321187e-05,
+      "loss": 0.7373,
+      "step": 3024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01128888889786e-05,
+      "loss": 0.7975,
+      "step": 3025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00547503598642e-05,
+      "loss": 0.7644,
+      "step": 3026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999661522571519e-05,
+      "loss": 0.778,
+      "step": 3027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993848350637681e-05,
+      "loss": 0.8099,
+      "step": 3028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988035522169338e-05,
+      "loss": 0.7711,
+      "step": 3029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982223039150777e-05,
+      "loss": 0.8781,
+      "step": 3030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976410903566188e-05,
+      "loss": 0.2847,
+      "step": 3031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970599117399625e-05,
+      "loss": 0.7597,
+      "step": 3032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964787682635037e-05,
+      "loss": 0.8088,
+      "step": 3033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958976601256248e-05,
+      "loss": 0.7746,
+      "step": 3034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953165875246956e-05,
+      "loss": 0.8394,
+      "step": 3035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947355506590749e-05,
+      "loss": 0.82,
+      "step": 3036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94154549727108e-05,
+      "loss": 0.8271,
+      "step": 3037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935735849271291e-05,
+      "loss": 0.2681,
+      "step": 3038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92992656457459e-05,
+      "loss": 0.7312,
+      "step": 3039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924117645164073e-05,
+      "loss": 0.7594,
+      "step": 3040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918309093022703e-05,
+      "loss": 0.7858,
+      "step": 3041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912500910133316e-05,
+      "loss": 0.7442,
+      "step": 3042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90669309847863e-05,
+      "loss": 0.691,
+      "step": 3043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900885660041226e-05,
+      "loss": 0.7939,
+      "step": 3044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89507859680357e-05,
+      "loss": 0.816,
+      "step": 3045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889271910747988e-05,
+      "loss": 0.7823,
+      "step": 3046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883465603856682e-05,
+      "loss": 0.809,
+      "step": 3047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877659678111731e-05,
+      "loss": 0.7969,
+      "step": 3048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871854135495072e-05,
+      "loss": 0.7273,
+      "step": 3049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866048977988522e-05,
+      "loss": 0.8036,
+      "step": 3050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860244207573756e-05,
+      "loss": 0.749,
+      "step": 3051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854439826232325e-05,
+      "loss": 0.7322,
+      "step": 3052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848635835945645e-05,
+      "loss": 0.772,
+      "step": 3053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842832238695002e-05,
+      "loss": 0.7686,
+      "step": 3054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837029036461538e-05,
+      "loss": 0.7468,
+      "step": 3055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83122623122627e-05,
+      "loss": 0.7636,
+      "step": 3056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825423824970073e-05,
+      "loss": 0.7922,
+      "step": 3057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819621819673689e-05,
+      "loss": 0.7627,
+      "step": 3058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813820217317725e-05,
+      "loss": 0.7875,
+      "step": 3059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808019019882645e-05,
+      "loss": 0.8159,
+      "step": 3060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80221822934878e-05,
+      "loss": 0.7575,
+      "step": 3061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796417847696319e-05,
+      "loss": 0.7534,
+      "step": 3062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790617876905313e-05,
+      "loss": 0.8272,
+      "step": 3063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78481831895567e-05,
+      "loss": 0.7525,
+      "step": 3064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779019175827159e-05,
+      "loss": 0.7795,
+      "step": 3065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773220449499413e-05,
+      "loss": 0.733,
+      "step": 3066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767422141951911e-05,
+      "loss": 0.7781,
+      "step": 3067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761624255164e-05,
+      "loss": 0.7463,
+      "step": 3068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755826791114873e-05,
+      "loss": 0.7951,
+      "step": 3069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75002975178359e-05,
+      "loss": 0.7996,
+      "step": 3070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744233139149058e-05,
+      "loss": 0.8056,
+      "step": 3071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73843695519004e-05,
+      "loss": 0.713,
+      "step": 3072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732641201885154e-05,
+      "loss": 0.7786,
+      "step": 3073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72684588121287e-05,
+      "loss": 0.7763,
+      "step": 3074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721050995151514e-05,
+      "loss": 0.8481,
+      "step": 3075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715256545679255e-05,
+      "loss": 0.7915,
+      "step": 3076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.70946253477412e-05,
+      "loss": 0.7846,
+      "step": 3077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703668964413986e-05,
+      "loss": 0.7974,
+      "step": 3078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697875836576581e-05,
+      "loss": 0.2713,
+      "step": 3079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692083153239473e-05,
+      "loss": 0.8106,
+      "step": 3080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686290916380088e-05,
+      "loss": 0.795,
+      "step": 3081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680499127975696e-05,
+      "loss": 0.8196,
+      "step": 3082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674707790003411e-05,
+      "loss": 0.8084,
+      "step": 3083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668916904440201e-05,
+      "loss": 0.8184,
+      "step": 3084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663126473262869e-05,
+      "loss": 0.7877,
+      "step": 3085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657336498448075e-05,
+      "loss": 0.785,
+      "step": 3086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651546981972309e-05,
+      "loss": 0.7994,
+      "step": 3087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645757925811922e-05,
+      "loss": 0.7821,
+      "step": 3088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639969331943088e-05,
+      "loss": 0.7388,
+      "step": 3089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634181202341839e-05,
+      "loss": 0.8134,
+      "step": 3090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628393538984044e-05,
+      "loss": 0.7868,
+      "step": 3091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622606343845409e-05,
+      "loss": 0.798,
+      "step": 3092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616819618901484e-05,
+      "loss": 0.2549,
+      "step": 3093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611033366127656e-05,
+      "loss": 0.8136,
+      "step": 3094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605247587499153e-05,
+      "loss": 0.7954,
+      "step": 3095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599462284991045e-05,
+      "loss": 0.7781,
+      "step": 3096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593677460578225e-05,
+      "loss": 0.8342,
+      "step": 3097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587893116235442e-05,
+      "loss": 0.8616,
+      "step": 3098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582109253937267e-05,
+      "loss": 0.7746,
+      "step": 3099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576325875658113e-05,
+      "loss": 0.8322,
+      "step": 3100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570542983372222e-05,
+      "loss": 0.7738,
+      "step": 3101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.564760579053677e-05,
+      "loss": 0.7659,
+      "step": 3102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558978664676393e-05,
+      "loss": 0.7546,
+      "step": 3103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553197242214117e-05,
+      "loss": 0.7812,
+      "step": 3104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547416313640424e-05,
+      "loss": 0.7891,
+      "step": 3105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541635880928726e-05,
+      "loss": 0.7877,
+      "step": 3106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53585594605226e-05,
+      "loss": 0.7624,
+      "step": 3107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5300765109841e-05,
+      "loss": 0.7666,
+      "step": 3108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524297577697147e-05,
+      "loss": 0.7603,
+      "step": 3109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518519148164124e-05,
+      "loss": 0.7958,
+      "step": 3110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512741224357595e-05,
+      "loss": 0.7158,
+      "step": 3111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506963808249938e-05,
+      "loss": 0.7549,
+      "step": 3112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501186901813363e-05,
+      "loss": 0.7423,
+      "step": 3113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495410507019914e-05,
+      "loss": 0.7807,
+      "step": 3114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489634625841446e-05,
+      "loss": 0.8212,
+      "step": 3115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48385926024965e-05,
+      "loss": 0.7712,
+      "step": 3116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478084412216033e-05,
+      "loss": 0.7675,
+      "step": 3117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472310083711933e-05,
+      "loss": 0.8453,
+      "step": 3118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466536276708502e-05,
+      "loss": 0.7812,
+      "step": 3119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460762993176722e-05,
+      "loss": 0.7721,
+      "step": 3120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454990235087395e-05,
+      "loss": 0.8617,
+      "step": 3121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449218004411135e-05,
+      "loss": 0.7922,
+      "step": 3122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44344630311839e-05,
+      "loss": 0.8436,
+      "step": 3123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437675133179414e-05,
+      "loss": 0.8016,
+      "step": 3124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43190449656429e-05,
+      "loss": 0.7476,
+      "step": 3125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426134395242904e-05,
+      "loss": 0.8115,
+      "step": 3126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.420364831184986e-05,
+      "loss": 0.8223,
+      "step": 3127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414595806360056e-05,
+      "loss": 0.744,
+      "step": 3128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408827322737464e-05,
+      "loss": 0.7759,
+      "step": 3129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403059382286366e-05,
+      "loss": 0.7736,
+      "step": 3130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397291986975743e-05,
+      "loss": 0.7937,
+      "step": 3131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391525138774387e-05,
+      "loss": 0.8139,
+      "step": 3132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385758839650896e-05,
+      "loss": 0.7537,
+      "step": 3133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379993091573687e-05,
+      "loss": 0.8477,
+      "step": 3134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374227896510988e-05,
+      "loss": 0.7692,
+      "step": 3135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368463256430841e-05,
+      "loss": 0.7622,
+      "step": 3136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36269917330109e-05,
+      "loss": 0.8035,
+      "step": 3137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356935649089395e-05,
+      "loss": 0.7968,
+      "step": 3138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351172685763229e-05,
+      "loss": 0.7855,
+      "step": 3139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345410285289865e-05,
+      "loss": 0.7904,
+      "step": 3140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33964844963639e-05,
+      "loss": 0.8267,
+      "step": 3141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333887180769694e-05,
+      "loss": 0.814,
+      "step": 3142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328126480656477e-05,
+      "loss": 0.7881,
+      "step": 3143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322366351263244e-05,
+      "loss": 0.7844,
+      "step": 3144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3166067945563e-05,
+      "loss": 0.8104,
+      "step": 3145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310847812501769e-05,
+      "loss": 0.7657,
+      "step": 3146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305089407065558e-05,
+      "loss": 0.2659,
+      "step": 3147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299331580213396e-05,
+      "loss": 0.769,
+      "step": 3148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2935743339108e-05,
+      "loss": 0.7682,
+      "step": 3149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287817670123101e-05,
+      "loss": 0.6833,
+      "step": 3150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28206159081542e-05,
+      "loss": 0.8211,
+      "step": 3151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276306097952694e-05,
+      "loss": 0.7466,
+      "step": 3152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270551193499638e-05,
+      "loss": 0.7389,
+      "step": 3153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264796879420786e-05,
+      "loss": 0.7836,
+      "step": 3154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259043157680459e-05,
+      "loss": 0.7675,
+      "step": 3155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253290030242778e-05,
+      "loss": 0.8203,
+      "step": 3156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247537499071669e-05,
+      "loss": 0.7748,
+      "step": 3157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24178556613084e-05,
+      "loss": 0.2864,
+      "step": 3158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236034233383807e-05,
+      "loss": 0.7717,
+      "step": 3159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230283502793873e-05,
+      "loss": 0.7331,
+      "step": 3160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224533376324146e-05,
+      "loss": 0.8112,
+      "step": 3161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218783855937514e-05,
+      "loss": 0.7835,
+      "step": 3162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213034943596667e-05,
+      "loss": 0.7564,
+      "step": 3163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207286641264089e-05,
+      "loss": 0.7585,
+      "step": 3164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201538950902048e-05,
+      "loss": 0.7922,
+      "step": 3165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195791874472609e-05,
+      "loss": 0.8042,
+      "step": 3166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190045413937625e-05,
+      "loss": 0.8321,
+      "step": 3167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184299571258741e-05,
+      "loss": 0.8018,
+      "step": 3168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178554348397387e-05,
+      "loss": 0.811,
+      "step": 3169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172809747314785e-05,
+      "loss": 0.7403,
+      "step": 3170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167065769971948e-05,
+      "loss": 0.2385,
+      "step": 3171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161322418329666e-05,
+      "loss": 0.7609,
+      "step": 3172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155579694348526e-05,
+      "loss": 0.8412,
+      "step": 3173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149837599988891e-05,
+      "loss": 0.7884,
+      "step": 3174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144096137210914e-05,
+      "loss": 0.2682,
+      "step": 3175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138355307974538e-05,
+      "loss": 0.7993,
+      "step": 3176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132615114239482e-05,
+      "loss": 0.8065,
+      "step": 3177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126875557965248e-05,
+      "loss": 0.7619,
+      "step": 3178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121136641111128e-05,
+      "loss": 0.8335,
+      "step": 3179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115398365636184e-05,
+      "loss": 0.7238,
+      "step": 3180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109660733499265e-05,
+      "loss": 0.7967,
+      "step": 3181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103923746659009e-05,
+      "loss": 0.7851,
+      "step": 3182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098187407073817e-05,
+      "loss": 0.7489,
+      "step": 3183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092451716701883e-05,
+      "loss": 0.8101,
+      "step": 3184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08671667750117e-05,
+      "loss": 0.831,
+      "step": 3185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080982291429428e-05,
+      "loss": 0.8106,
+      "step": 3186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075248560444171e-05,
+      "loss": 0.7897,
+      "step": 3187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.069515486502702e-05,
+      "loss": 0.7625,
+      "step": 3188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063783071562096e-05,
+      "loss": 0.7544,
+      "step": 3189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0580513175792e-05,
+      "loss": 0.7724,
+      "step": 3190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052320226510637e-05,
+      "loss": 0.7931,
+      "step": 3191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046589800312803e-05,
+      "loss": 0.7848,
+      "step": 3192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040860040941868e-05,
+      "loss": 0.7359,
+      "step": 3193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035130950353779e-05,
+      "loss": 0.8066,
+      "step": 3194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029402530504243e-05,
+      "loss": 0.789,
+      "step": 3195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023674783348751e-05,
+      "loss": 0.8029,
+      "step": 3196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017947710842554e-05,
+      "loss": 0.8122,
+      "step": 3197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01222131494068e-05,
+      "loss": 0.7645,
+      "step": 3198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006495597597922e-05,
+      "loss": 0.7414,
+      "step": 3199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00077056076884e-05,
+      "loss": 0.7503,
+      "step": 3200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995046206407768e-05,
+      "loss": 0.7464,
+      "step": 3201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989322536468805e-05,
+      "loss": 0.7691,
+      "step": 3202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98359955290581e-05,
+      "loss": 0.7837,
+      "step": 3203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977877257672415e-05,
+      "loss": 0.7342,
+      "step": 3204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97215565272201e-05,
+      "loss": 0.2597,
+      "step": 3205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96643474000776e-05,
+      "loss": 0.8258,
+      "step": 3206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960714521482587e-05,
+      "loss": 0.7788,
+      "step": 3207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954994999099168e-05,
+      "loss": 0.2356,
+      "step": 3208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94927617480996e-05,
+      "loss": 0.8408,
+      "step": 3209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943558050567168e-05,
+      "loss": 0.7813,
+      "step": 3210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937840628322764e-05,
+      "loss": 0.739,
+      "step": 3211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93212391002848e-05,
+      "loss": 0.724,
+      "step": 3212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926407897635801e-05,
+      "loss": 0.7694,
+      "step": 3213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920692593095984e-05,
+      "loss": 0.8124,
+      "step": 3214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914977998360031e-05,
+      "loss": 0.7779,
+      "step": 3215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909264115378713e-05,
+      "loss": 0.7302,
+      "step": 3216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903550946102547e-05,
+      "loss": 0.7474,
+      "step": 3217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897838492481814e-05,
+      "loss": 0.7867,
+      "step": 3218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892126756466552e-05,
+      "loss": 0.7727,
+      "step": 3219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886415740006547e-05,
+      "loss": 0.836,
+      "step": 3220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880705445051347e-05,
+      "loss": 0.7713,
+      "step": 3221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874995873550243e-05,
+      "loss": 0.7566,
+      "step": 3222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869287027452294e-05,
+      "loss": 0.798,
+      "step": 3223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863578908706291e-05,
+      "loss": 0.7504,
+      "step": 3224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857871519260805e-05,
+      "loss": 0.7792,
+      "step": 3225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852164861064133e-05,
+      "loss": 0.8013,
+      "step": 3226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846458936064334e-05,
+      "loss": 0.7914,
+      "step": 3227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840753746209211e-05,
+      "loss": 0.7685,
+      "step": 3228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835049293446319e-05,
+      "loss": 0.7616,
+      "step": 3229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829345579722969e-05,
+      "loss": 0.8077,
+      "step": 3230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823642606986201e-05,
+      "loss": 0.7619,
+      "step": 3231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817940377182824e-05,
+      "loss": 0.796,
+      "step": 3232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812238892259374e-05,
+      "loss": 0.8178,
+      "step": 3233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806538154162148e-05,
+      "loss": 0.7956,
+      "step": 3234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800838164837175e-05,
+      "loss": 0.7786,
+      "step": 3235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79513892623024e-05,
+      "loss": 0.7918,
+      "step": 3236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789440440286866e-05,
+      "loss": 0.734,
+      "step": 3237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783742708952316e-05,
+      "loss": 0.7694,
+      "step": 3238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778045734171606e-05,
+      "loss": 0.26,
+      "step": 3239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772349517889478e-05,
+      "loss": 0.7867,
+      "step": 3240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766654062050431e-05,
+      "loss": 0.7137,
+      "step": 3241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760959368598692e-05,
+      "loss": 0.8047,
+      "step": 3242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755265439478237e-05,
+      "loss": 0.8112,
+      "step": 3243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.749572276632777e-05,
+      "loss": 0.8195,
+      "step": 3244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743879882005759e-05,
+      "loss": 0.7911,
+      "step": 3245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738188257540373e-05,
+      "loss": 0.7158,
+      "step": 3246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732497405179539e-05,
+      "loss": 0.8152,
+      "step": 3247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726807326865921e-05,
+      "loss": 0.7686,
+      "step": 3248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721118024541914e-05,
+      "loss": 0.7536,
+      "step": 3249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715429500149655e-05,
+      "loss": 0.7937,
+      "step": 3250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709741755631004e-05,
+      "loss": 0.7923,
+      "step": 3251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704054792927564e-05,
+      "loss": 0.2491,
+      "step": 3252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698368613980667e-05,
+      "loss": 0.7812,
+      "step": 3253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692683220731376e-05,
+      "loss": 0.7856,
+      "step": 3254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686998615120494e-05,
+      "loss": 0.7639,
+      "step": 3255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681314799088544e-05,
+      "loss": 0.8184,
+      "step": 3256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675631774575788e-05,
+      "loss": 0.8021,
+      "step": 3257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669949543522214e-05,
+      "loss": 0.7666,
+      "step": 3258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66426810786754e-05,
+      "loss": 0.7408,
+      "step": 3259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658587469551209e-05,
+      "loss": 0.8082,
+      "step": 3260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652907630512397e-05,
+      "loss": 0.7539,
+      "step": 3261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647228592690011e-05,
+      "loss": 0.7753,
+      "step": 3262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64155035802267e-05,
+      "loss": 0.724,
+      "step": 3263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635872928448734e-05,
+      "loss": 0.7843,
+      "step": 3264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630196305906277e-05,
+      "loss": 0.8532,
+      "step": 3265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624520492333104e-05,
+      "loss": 0.8138,
+      "step": 3266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618845489666742e-05,
+      "loss": 0.7873,
+      "step": 3267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613171299844443e-05,
+      "loss": 0.7881,
+      "step": 3268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607497924803179e-05,
+      "loss": 0.7264,
+      "step": 3269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601825366479642e-05,
+      "loss": 0.7185,
+      "step": 3270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596153626810252e-05,
+      "loss": 0.7451,
+      "step": 3271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590482707731141e-05,
+      "loss": 0.2448,
+      "step": 3272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584812611178166e-05,
+      "loss": 0.779,
+      "step": 3273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579143339086906e-05,
+      "loss": 0.7599,
+      "step": 3274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573474893392654e-05,
+      "loss": 0.8335,
+      "step": 3275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567807276030421e-05,
+      "loss": 0.8454,
+      "step": 3276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562140488934938e-05,
+      "loss": 0.8022,
+      "step": 3277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556474534040646e-05,
+      "loss": 0.7984,
+      "step": 3278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55080941328171e-05,
+      "loss": 0.7832,
+      "step": 3279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54514512859201e-05,
+      "loss": 0.7957,
+      "step": 3280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53948168190513e-05,
+      "loss": 0.6792,
+      "step": 3281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533819075154382e-05,
+      "loss": 0.7572,
+      "step": 3282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52815731027278e-05,
+      "loss": 0.8043,
+      "step": 3283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522496389193062e-05,
+      "loss": 0.7624,
+      "step": 3284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516836313847664e-05,
+      "loss": 0.7383,
+      "step": 3285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511177086168742e-05,
+      "loss": 0.8058,
+      "step": 3286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505518708088166e-05,
+      "loss": 0.8151,
+      "step": 3287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499861181537506e-05,
+      "loss": 0.7273,
+      "step": 3288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49420450844805e-05,
+      "loss": 0.8222,
+      "step": 3289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488548690750788e-05,
+      "loss": 0.8202,
+      "step": 3290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.482893730376424e-05,
+      "loss": 0.6973,
+      "step": 3291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.477239629255366e-05,
+      "loss": 0.7238,
+      "step": 3292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471586389317727e-05,
+      "loss": 0.7254,
+      "step": 3293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465934012493332e-05,
+      "loss": 0.8006,
+      "step": 3294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460282500711703e-05,
+      "loss": 0.727,
+      "step": 3295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454631855902077e-05,
+      "loss": 0.8254,
+      "step": 3296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448982079993382e-05,
+      "loss": 0.7612,
+      "step": 3297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44333317491426e-05,
+      "loss": 0.7872,
+      "step": 3298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.437685142593055e-05,
+      "loss": 0.8167,
+      "step": 3299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432037984957811e-05,
+      "loss": 0.2682,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426391703936269e-05,
+      "loss": 0.7453,
+      "step": 3301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420746301455877e-05,
+      "loss": 0.7232,
+      "step": 3302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41510177944378e-05,
+      "loss": 0.7858,
+      "step": 3303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409458139826822e-05,
+      "loss": 0.8303,
+      "step": 3304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40381538453155e-05,
+      "loss": 0.787,
+      "step": 3305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398173515484204e-05,
+      "loss": 0.7776,
+      "step": 3306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392532534610725e-05,
+      "loss": 0.7951,
+      "step": 3307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386892443836748e-05,
+      "loss": 0.763,
+      "step": 3308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381253245087605e-05,
+      "loss": 0.8166,
+      "step": 3309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375614940288327e-05,
+      "loss": 0.7596,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369977531363634e-05,
+      "loss": 0.8015,
+      "step": 3311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364341020237946e-05,
+      "loss": 0.7955,
+      "step": 3312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35870540883537e-05,
+      "loss": 0.8066,
+      "step": 3313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35307069907971e-05,
+      "loss": 0.8154,
+      "step": 3314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347436892894464e-05,
+      "loss": 0.7749,
+      "step": 3315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341803992202816e-05,
+      "loss": 0.7827,
+      "step": 3316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33617199892765e-05,
+      "loss": 0.7713,
+      "step": 3317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330540914991527e-05,
+      "loss": 0.7684,
+      "step": 3318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324910742316711e-05,
+      "loss": 0.7501,
+      "step": 3319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319281482825145e-05,
+      "loss": 0.7552,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313653138438465e-05,
+      "loss": 0.7678,
+      "step": 3321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30802571107799e-05,
+      "loss": 0.7434,
+      "step": 3322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302399202664739e-05,
+      "loss": 0.7629,
+      "step": 3323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2967736151194e-05,
+      "loss": 0.7911,
+      "step": 3324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291148950362361e-05,
+      "loss": 0.7895,
+      "step": 3325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285525210313683e-05,
+      "loss": 0.7973,
+      "step": 3326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279902396893119e-05,
+      "loss": 0.8247,
+      "step": 3327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274280512020105e-05,
+      "loss": 0.2637,
+      "step": 3328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.268659557613756e-05,
+      "loss": 0.2855,
+      "step": 3329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263039535592875e-05,
+      "loss": 0.7929,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25742044787594e-05,
+      "loss": 0.7459,
+      "step": 3331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251802296381119e-05,
+      "loss": 0.7605,
+      "step": 3332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24618508302625e-05,
+      "loss": 0.7008,
+      "step": 3333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240568809728858e-05,
+      "loss": 0.7816,
+      "step": 3334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234953478406148e-05,
+      "loss": 0.7593,
+      "step": 3335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229339090974996e-05,
+      "loss": 0.8611,
+      "step": 3336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223725649351964e-05,
+      "loss": 0.8138,
+      "step": 3337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218113155453283e-05,
+      "loss": 0.2392,
+      "step": 3338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21250161119487e-05,
+      "loss": 0.737,
+      "step": 3339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206891018492308e-05,
+      "loss": 0.7329,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201281379260862e-05,
+      "loss": 0.7645,
+      "step": 3341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195672695415472e-05,
+      "loss": 0.7792,
+      "step": 3342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190064968870747e-05,
+      "loss": 0.7838,
+      "step": 3343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184458201540972e-05,
+      "loss": 0.8014,
+      "step": 3344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178852395340101e-05,
+      "loss": 0.8057,
+      "step": 3345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173247552181768e-05,
+      "loss": 0.8034,
+      "step": 3346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167643673979267e-05,
+      "loss": 0.8117,
+      "step": 3347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162040762645576e-05,
+      "loss": 0.7627,
+      "step": 3348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156438820093332e-05,
+      "loss": 0.7171,
+      "step": 3349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150837848234846e-05,
+      "loss": 0.7894,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145237848982095e-05,
+      "loss": 0.859,
+      "step": 3351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139638824246729e-05,
+      "loss": 0.7751,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134040775940059e-05,
+      "loss": 0.7928,
+      "step": 3353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128443705973067e-05,
+      "loss": 0.777,
+      "step": 3354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1228476162564e-05,
+      "loss": 0.6797,
+      "step": 3355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117252508700367e-05,
+      "loss": 0.7413,
+      "step": 3356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11165838521495e-05,
+      "loss": 0.6983,
+      "step": 3357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106065247709785e-05,
+      "loss": 0.2619,
+      "step": 3358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10047309809418e-05,
+      "loss": 0.7268,
+      "step": 3359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094881938277101e-05,
+      "loss": 0.837,
+      "step": 3360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089291770167176e-05,
+      "loss": 0.727,
+      "step": 3361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083702595672697e-05,
+      "loss": 0.7505,
+      "step": 3362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078114416701615e-05,
+      "loss": 0.7301,
+      "step": 3363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07252723516154e-05,
+      "loss": 0.7661,
+      "step": 3364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066941052959748e-05,
+      "loss": 0.7234,
+      "step": 3365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061355872003164e-05,
+      "loss": 0.7876,
+      "step": 3366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055771694198379e-05,
+      "loss": 0.7821,
+      "step": 3367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050188521451637e-05,
+      "loss": 0.7531,
+      "step": 3368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044606355668842e-05,
+      "loss": 0.2741,
+      "step": 3369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039025198755552e-05,
+      "loss": 0.7209,
+      "step": 3370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033445052616983e-05,
+      "loss": 0.8033,
+      "step": 3371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027865919157999e-05,
+      "loss": 0.8095,
+      "step": 3372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022287800283133e-05,
+      "loss": 0.8115,
+      "step": 3373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016710697896558e-05,
+      "loss": 0.253,
+      "step": 3374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011134613902106e-05,
+      "loss": 0.7884,
+      "step": 3375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005559550203255e-05,
+      "loss": 0.8295,
+      "step": 3376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999985508703145e-05,
+      "loss": 0.756,
+      "step": 3377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99441249130456e-05,
+      "loss": 0.747,
+      "step": 3378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988840499909933e-05,
+      "loss": 0.7818,
+      "step": 3379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983269536421355e-05,
+      "loss": 0.7601,
+      "step": 3380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977699602740555e-05,
+      "loss": 0.7712,
+      "step": 3381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972130700768923e-05,
+      "loss": 0.8365,
+      "step": 3382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966562832407484e-05,
+      "loss": 0.805,
+      "step": 3383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960995999556918e-05,
+      "loss": 0.775,
+      "step": 3384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955430204117555e-05,
+      "loss": 0.7536,
+      "step": 3385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949865447989357e-05,
+      "loss": 0.7141,
+      "step": 3386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944301733071948e-05,
+      "loss": 0.7699,
+      "step": 3387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938739061264582e-05,
+      "loss": 0.7554,
+      "step": 3388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.933177434466166e-05,
+      "loss": 0.751,
+      "step": 3389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92761685457525e-05,
+      "loss": 0.801,
+      "step": 3390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922057323490021e-05,
+      "loss": 0.795,
+      "step": 3391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916498843108315e-05,
+      "loss": 0.7878,
+      "step": 3392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910941415327598e-05,
+      "loss": 0.7929,
+      "step": 3393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905385042044994e-05,
+      "loss": 0.789,
+      "step": 3394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89982972515725e-05,
+      "loss": 0.7953,
+      "step": 3395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894275466560758e-05,
+      "loss": 0.7926,
+      "step": 3396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88872226815156e-05,
+      "loss": 0.7455,
+      "step": 3397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88317013182532e-05,
+      "loss": 0.743,
+      "step": 3398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877619059477346e-05,
+      "loss": 0.7369,
+      "step": 3399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872069053002588e-05,
+      "loss": 0.8478,
+      "step": 3400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866520114295618e-05,
+      "loss": 0.7498,
+      "step": 3401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860972245250658e-05,
+      "loss": 0.7874,
+      "step": 3402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855425447761561e-05,
+      "loss": 0.7757,
+      "step": 3403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849879723721806e-05,
+      "loss": 0.7943,
+      "step": 3404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844335075024518e-05,
+      "loss": 0.7375,
+      "step": 3405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838791503562445e-05,
+      "loss": 0.7952,
+      "step": 3406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833249011227974e-05,
+      "loss": 0.7149,
+      "step": 3407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82770759991312e-05,
+      "loss": 0.8031,
+      "step": 3408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822167271509528e-05,
+      "loss": 0.7731,
+      "step": 3409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816628027908478e-05,
+      "loss": 0.7983,
+      "step": 3410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811089871000873e-05,
+      "loss": 0.721,
+      "step": 3411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805552802677256e-05,
+      "loss": 0.7883,
+      "step": 3412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80001682482778e-05,
+      "loss": 0.7442,
+      "step": 3413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794481939342247e-05,
+      "loss": 0.746,
+      "step": 3414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788948148110073e-05,
+      "loss": 0.7994,
+      "step": 3415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.783415453020301e-05,
+      "loss": 0.2879,
+      "step": 3416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777883855961606e-05,
+      "loss": 0.2553,
+      "step": 3417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772353358822282e-05,
+      "loss": 0.821,
+      "step": 3418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766823963490251e-05,
+      "loss": 0.8021,
+      "step": 3419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761295671853059e-05,
+      "loss": 0.7641,
+      "step": 3420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755768485797868e-05,
+      "loss": 0.2812,
+      "step": 3421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750242407211479e-05,
+      "loss": 0.786,
+      "step": 3422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744717437980299e-05,
+      "loss": 0.8454,
+      "step": 3423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739193579990362e-05,
+      "loss": 0.8399,
+      "step": 3424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733670835127323e-05,
+      "loss": 0.7606,
+      "step": 3425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728149205276459e-05,
+      "loss": 0.8299,
+      "step": 3426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722628692322662e-05,
+      "loss": 0.7594,
+      "step": 3427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717109298150445e-05,
+      "loss": 0.7646,
+      "step": 3428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711591024643938e-05,
+      "loss": 0.7936,
+      "step": 3429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706073873686892e-05,
+      "loss": 0.7903,
+      "step": 3430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.700557847162667e-05,
+      "loss": 0.75,
+      "step": 3431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695042946954248e-05,
+      "loss": 0.7737,
+      "step": 3432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689529174944233e-05,
+      "loss": 0.7611,
+      "step": 3433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684016533014831e-05,
+      "loss": 0.7924,
+      "step": 3434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67850502304787e-05,
+      "loss": 0.7581,
+      "step": 3435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672994646924788e-05,
+      "loss": 0.2696,
+      "step": 3436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667485406526639e-05,
+      "loss": 0.7651,
+      "step": 3437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661977303734085e-05,
+      "loss": 0.7363,
+      "step": 3438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656470340427403e-05,
+      "loss": 0.7436,
+      "step": 3439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650964518486485e-05,
+      "loss": 0.7441,
+      "step": 3440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645459839790824e-05,
+      "loss": 0.7812,
+      "step": 3441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63995630621953e-05,
+      "loss": 0.7647,
+      "step": 3442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634453919651315e-05,
+      "loss": 0.7604,
+      "step": 3443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628952681964509e-05,
+      "loss": 0.757,
+      "step": 3444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62345259503704e-05,
+      "loss": 0.2797,
+      "step": 3445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617953660746459e-05,
+      "loss": 0.7387,
+      "step": 3446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612455880969901e-05,
+      "loss": 0.7739,
+      "step": 3447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606959257584124e-05,
+      "loss": 0.7918,
+      "step": 3448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601463792465483e-05,
+      "loss": 0.779,
+      "step": 3449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595969487489941e-05,
+      "loss": 0.2437,
+      "step": 3450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590476344533067e-05,
+      "loss": 0.7602,
+      "step": 3451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584984365470028e-05,
+      "loss": 0.8049,
+      "step": 3452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5794935521756e-05,
+      "loss": 0.2615,
+      "step": 3453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574003906524148e-05,
+      "loss": 0.7825,
+      "step": 3454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56851543038966e-05,
+      "loss": 0.7999,
+      "step": 3455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563028125645704e-05,
+      "loss": 0.7531,
+      "step": 3456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557541994165458e-05,
+      "loss": 0.2545,
+      "step": 3457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552057037821702e-05,
+      "loss": 0.79,
+      "step": 3458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546573258486808e-05,
+      "loss": 0.7342,
+      "step": 3459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541090658032749e-05,
+      "loss": 0.8106,
+      "step": 3460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535609238331096e-05,
+      "loss": 0.8026,
+      "step": 3461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530129001253018e-05,
+      "loss": 0.7663,
+      "step": 3462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524649948669279e-05,
+      "loss": 0.7645,
+      "step": 3463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519172082450236e-05,
+      "loss": 0.7978,
+      "step": 3464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.513695404465847e-05,
+      "loss": 0.7879,
+      "step": 3465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508219916585659e-05,
+      "loss": 0.759,
+      "step": 3466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502745620678817e-05,
+      "loss": 0.8183,
+      "step": 3467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497272518614052e-05,
+      "loss": 0.2477,
+      "step": 3468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491800612259698e-05,
+      "loss": 0.2731,
+      "step": 3469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48632990348367e-05,
+      "loss": 0.7807,
+      "step": 3470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480860394153485e-05,
+      "loss": 0.2643,
+      "step": 3471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47539208613624e-05,
+      "loss": 0.7416,
+      "step": 3472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469924981298634e-05,
+      "loss": 0.7877,
+      "step": 3473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464459081506938e-05,
+      "loss": 0.7997,
+      "step": 3474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458994388627029e-05,
+      "loss": 0.8038,
+      "step": 3475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453530904524362e-05,
+      "loss": 0.8093,
+      "step": 3476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448068631063984e-05,
+      "loss": 0.7816,
+      "step": 3477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442607570110527e-05,
+      "loss": 0.7644,
+      "step": 3478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437147723528204e-05,
+      "loss": 0.8386,
+      "step": 3479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.431689093180825e-05,
+      "loss": 0.7998,
+      "step": 3480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426231680931773e-05,
+      "loss": 0.7705,
+      "step": 3481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420775488644021e-05,
+      "loss": 0.7705,
+      "step": 3482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415320518180131e-05,
+      "loss": 0.2777,
+      "step": 3483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409866771402235e-05,
+      "loss": 0.7949,
+      "step": 3484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404414250172059e-05,
+      "loss": 0.6833,
+      "step": 3485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3989629563509e-05,
+      "loss": 0.7699,
+      "step": 3486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393512891799644e-05,
+      "loss": 0.7668,
+      "step": 3487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388064058378761e-05,
+      "loss": 0.7139,
+      "step": 3488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382616457948288e-05,
+      "loss": 0.7484,
+      "step": 3489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37717009236785e-05,
+      "loss": 0.827,
+      "step": 3490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371724963496648e-05,
+      "loss": 0.7508,
+      "step": 3491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366281073193463e-05,
+      "loss": 0.8122,
+      "step": 3492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360838423316649e-05,
+      "loss": 0.7568,
+      "step": 3493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355397015724139e-05,
+      "loss": 0.7871,
+      "step": 3494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349956852273445e-05,
+      "loss": 0.7807,
+      "step": 3495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34451793482165e-05,
+      "loss": 0.7921,
+      "step": 3496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339080265225412e-05,
+      "loss": 0.7682,
+      "step": 3497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333643845340967e-05,
+      "loss": 0.8438,
+      "step": 3498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328208677024115e-05,
+      "loss": 0.8298,
+      "step": 3499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32277476213024e-05,
+      "loss": 0.8294,
+      "step": 3500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317342102514294e-05,
+      "loss": 0.7325,
+      "step": 3501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311910700030796e-05,
+      "loss": 0.7576,
+      "step": 3502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306480556533843e-05,
+      "loss": 0.8315,
+      "step": 3503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301051673877095e-05,
+      "loss": 0.7268,
+      "step": 3504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295624053913788e-05,
+      "loss": 0.6603,
+      "step": 3505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290197698496726e-05,
+      "loss": 0.7599,
+      "step": 3506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284772609478276e-05,
+      "loss": 0.8079,
+      "step": 3507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279348788710377e-05,
+      "loss": 0.8058,
+      "step": 3508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273926238044536e-05,
+      "loss": 0.8003,
+      "step": 3509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268504959331826e-05,
+      "loss": 0.7875,
+      "step": 3510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26308495442288e-05,
+      "loss": 0.7195,
+      "step": 3511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257666225167905e-05,
+      "loss": 0.7133,
+      "step": 3512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252248773416669e-05,
+      "loss": 0.7481,
+      "step": 3513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246832601018498e-05,
+      "loss": 0.8152,
+      "step": 3514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241417709822293e-05,
+      "loss": 0.7927,
+      "step": 3515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236004101676505e-05,
+      "loss": 0.7574,
+      "step": 3516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23059177842916e-05,
+      "loss": 0.7408,
+      "step": 3517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225180741927831e-05,
+      "loss": 0.2493,
+      "step": 3518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219770994019661e-05,
+      "loss": 0.7858,
+      "step": 3519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214362536551357e-05,
+      "loss": 0.75,
+      "step": 3520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208955371369178e-05,
+      "loss": 0.2592,
+      "step": 3521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20354950031894e-05,
+      "loss": 0.7752,
+      "step": 3522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198144925246023e-05,
+      "loss": 0.7905,
+      "step": 3523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192741647995367e-05,
+      "loss": 0.8073,
+      "step": 3524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187339670411455e-05,
+      "loss": 0.7643,
+      "step": 3525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181938994338345e-05,
+      "loss": 0.8219,
+      "step": 3526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176539621619637e-05,
+      "loss": 0.7603,
+      "step": 3527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171141554098491e-05,
+      "loss": 0.8444,
+      "step": 3528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165744793617622e-05,
+      "loss": 0.2562,
+      "step": 3529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1603493420193e-05,
+      "loss": 0.7964,
+      "step": 3530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154955201145342e-05,
+      "loss": 0.68,
+      "step": 3531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149562372837125e-05,
+      "loss": 0.7846,
+      "step": 3532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144170858935574e-05,
+      "loss": 0.7053,
+      "step": 3533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138780661281163e-05,
+      "loss": 0.2561,
+      "step": 3534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133391781713924e-05,
+      "loss": 0.7982,
+      "step": 3535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128004222073431e-05,
+      "loss": 0.275,
+      "step": 3536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122617984198814e-05,
+      "loss": 0.7752,
+      "step": 3537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11723306992875e-05,
+      "loss": 0.7529,
+      "step": 3538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111849481101457e-05,
+      "loss": 0.7676,
+      "step": 3539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106467219554713e-05,
+      "loss": 0.7855,
+      "step": 3540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101086287125833e-05,
+      "loss": 0.717,
+      "step": 3541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095706685651683e-05,
+      "loss": 0.8007,
+      "step": 3542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090328416968671e-05,
+      "loss": 0.7719,
+      "step": 3543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08495148291276e-05,
+      "loss": 0.7517,
+      "step": 3544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079575885319442e-05,
+      "loss": 0.8383,
+      "step": 3545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0742016260237645e-05,
+      "loss": 0.8188,
+      "step": 3546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068828706860312e-05,
+      "loss": 0.7832,
+      "step": 3547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063457129663217e-05,
+      "loss": 0.7485,
+      "step": 3548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058086896266148e-05,
+      "loss": 0.7988,
+      "step": 3549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052718008502319e-05,
+      "loss": 0.7498,
+      "step": 3550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047350468204485e-05,
+      "loss": 0.7932,
+      "step": 3551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041984277204934e-05,
+      "loss": 0.7546,
+      "step": 3552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036619437335504e-05,
+      "loss": 0.7456,
+      "step": 3553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0312559504275614e-05,
+      "loss": 0.7549,
+      "step": 3554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025893818312019e-05,
+      "loss": 0.8251,
+      "step": 3555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0205330428193254e-05,
+      "loss": 0.7558,
+      "step": 3556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01517362577946e-05,
+      "loss": 0.7806,
+      "step": 3557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0098155690219485e-05,
+      "loss": 0.7343,
+      "step": 3558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00445887437584e-05,
+      "loss": 0.786,
+      "step": 3559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999103543669729e-05,
+      "loss": 0.7346,
+      "step": 3560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9937495787317436e-05,
+      "loss": 0.8072,
+      "step": 3561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988396981389537e-05,
+      "loss": 0.8132,
+      "step": 3562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983045753470308e-05,
+      "loss": 0.7612,
+      "step": 3563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9776958968007743e-05,
+      "loss": 0.2604,
+      "step": 3564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9723474132072e-05,
+      "loss": 0.6438,
+      "step": 3565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967000304515366e-05,
+      "loss": 0.8705,
+      "step": 3566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961654572550592e-05,
+      "loss": 0.8287,
+      "step": 3567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956310219137729e-05,
+      "loss": 0.7981,
+      "step": 3568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950967246101158e-05,
+      "loss": 0.7989,
+      "step": 3569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945625655264781e-05,
+      "loss": 0.7655,
+      "step": 3570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940285448452034e-05,
+      "loss": 0.7523,
+      "step": 3571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93494662748588e-05,
+      "loss": 0.8127,
+      "step": 3572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929609194188808e-05,
+      "loss": 0.7691,
+      "step": 3573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9242731503828356e-05,
+      "loss": 0.8209,
+      "step": 3574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9189384978894993e-05,
+      "loss": 0.706,
+      "step": 3575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.913605238529871e-05,
+      "loss": 0.7916,
+      "step": 3576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90827337412454e-05,
+      "loss": 0.7976,
+      "step": 3577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90294290649362e-05,
+      "loss": 0.7275,
+      "step": 3578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897613837456748e-05,
+      "loss": 0.7672,
+      "step": 3579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892286168833085e-05,
+      "loss": 0.7461,
+      "step": 3580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886959902441318e-05,
+      "loss": 0.8396,
+      "step": 3581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881635040099642e-05,
+      "loss": 0.7341,
+      "step": 3582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876311583625791e-05,
+      "loss": 0.7742,
+      "step": 3583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870989534837001e-05,
+      "loss": 0.7956,
+      "step": 3584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8656688955500406e-05,
+      "loss": 0.783,
+      "step": 3585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8603496675811943e-05,
+      "loss": 0.7032,
+      "step": 3586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855031852746258e-05,
+      "loss": 0.7348,
+      "step": 3587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8497154528605555e-05,
+      "loss": 0.7885,
+      "step": 3588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844400469738918e-05,
+      "loss": 0.2809,
+      "step": 3589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8390869051957007e-05,
+      "loss": 0.7698,
+      "step": 3590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833774761044768e-05,
+      "loss": 0.7304,
+      "step": 3591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828464039099508e-05,
+      "loss": 0.7904,
+      "step": 3592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823154741172811e-05,
+      "loss": 0.761,
+      "step": 3593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817846869077092e-05,
+      "loss": 0.2886,
+      "step": 3594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8125404246242786e-05,
+      "loss": 0.7567,
+      "step": 3595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807235409625806e-05,
+      "loss": 0.7519,
+      "step": 3596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801931825892619e-05,
+      "loss": 0.2355,
+      "step": 3597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796629675235186e-05,
+      "loss": 0.7586,
+      "step": 3598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7913289594634744e-05,
+      "loss": 0.7638,
+      "step": 3599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786029680386966e-05,
+      "loss": 0.8216,
+      "step": 3600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780731839814647e-05,
+      "loss": 0.7983,
+      "step": 3601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775435439555027e-05,
+      "loss": 0.7585,
+      "step": 3602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7701404814161097e-05,
+      "loss": 0.7336,
+      "step": 3603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764846967205407e-05,
+      "loss": 0.7745,
+      "step": 3604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759554898729951e-05,
+      "loss": 0.7693,
+      "step": 3605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7542642777962664e-05,
+      "loss": 0.7457,
+      "step": 3606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7489751062103906e-05,
+      "loss": 0.7758,
+      "step": 3607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7436873857778575e-05,
+      "loss": 0.7983,
+      "step": 3608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738401118303722e-05,
+      "loss": 0.8303,
+      "step": 3609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733116305592532e-05,
+      "loss": 0.7933,
+      "step": 3610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727832949448333e-05,
+      "loss": 0.7805,
+      "step": 3611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722551051674689e-05,
+      "loss": 0.7917,
+      "step": 3612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717270614074656e-05,
+      "loss": 0.7315,
+      "step": 3613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711991638450791e-05,
+      "loss": 0.8247,
+      "step": 3614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7067141266051524e-05,
+      "loss": 0.6673,
+      "step": 3615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7014380803393064e-05,
+      "loss": 0.7634,
+      "step": 3616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.696163501454308e-05,
+      "loss": 0.6865,
+      "step": 3617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690890391750723e-05,
+      "loss": 0.7337,
+      "step": 3618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685618753028601e-05,
+      "loss": 0.813,
+      "step": 3619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680348587087505e-05,
+      "loss": 0.8048,
+      "step": 3620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675079895726487e-05,
+      "loss": 0.7483,
+      "step": 3621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6698126807440896e-05,
+      "loss": 0.7436,
+      "step": 3622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664546943938369e-05,
+      "loss": 0.7126,
+      "step": 3623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6592826871068594e-05,
+      "loss": 0.7601,
+      "step": 3624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654019912046599e-05,
+      "loss": 0.7835,
+      "step": 3625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648758620554112e-05,
+      "loss": 0.75,
+      "step": 3626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643498814425432e-05,
+      "loss": 0.7848,
+      "step": 3627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6382404954560685e-05,
+      "loss": 0.8233,
+      "step": 3628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632983665441029e-05,
+      "loss": 0.7731,
+      "step": 3629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62772832617482e-05,
+      "loss": 0.7788,
+      "step": 3630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6224744794514314e-05,
+      "loss": 0.7494,
+      "step": 3631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617222127064345e-05,
+      "loss": 0.778,
+      "step": 3632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611971270806526e-05,
+      "loss": 0.7675,
+      "step": 3633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6067219124704476e-05,
+      "loss": 0.7701,
+      "step": 3634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6014740538480515e-05,
+      "loss": 0.7384,
+      "step": 3635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5962276967307755e-05,
+      "loss": 0.772,
+      "step": 3636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5909828429095515e-05,
+      "loss": 0.7843,
+      "step": 3637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585739494174789e-05,
+      "loss": 0.7552,
+      "step": 3638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5804976523163835e-05,
+      "loss": 0.7579,
+      "step": 3639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575257319123719e-05,
+      "loss": 0.8043,
+      "step": 3640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5700184963856696e-05,
+      "loss": 0.7176,
+      "step": 3641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5647811858905804e-05,
+      "loss": 0.7294,
+      "step": 3642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5595453894262994e-05,
+      "loss": 0.7487,
+      "step": 3643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554311108780139e-05,
+      "loss": 0.7525,
+      "step": 3644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549078345738907e-05,
+      "loss": 0.8087,
+      "step": 3645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543847102088886e-05,
+      "loss": 0.7948,
+      "step": 3646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53861737961584e-05,
+      "loss": 0.7636,
+      "step": 3647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5333891801050216e-05,
+      "loss": 0.7186,
+      "step": 3648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5281625053411544e-05,
+      "loss": 0.7976,
+      "step": 3649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5229373571084456e-05,
+      "loss": 0.7442,
+      "step": 3650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517713737190575e-05,
+      "loss": 0.7465,
+      "step": 3651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512491647370717e-05,
+      "loss": 0.8306,
+      "step": 3652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507271089431507e-05,
+      "loss": 0.7764,
+      "step": 3653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502052065155061e-05,
+      "loss": 0.7305,
+      "step": 3654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49683457632298e-05,
+      "loss": 0.2551,
+      "step": 3655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491618624716333e-05,
+      "loss": 0.7718,
+      "step": 3656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486404212115664e-05,
+      "loss": 0.7698,
+      "step": 3657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481191340300992e-05,
+      "loss": 0.7522,
+      "step": 3658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475980011051818e-05,
+      "loss": 0.7745,
+      "step": 3659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470770226147106e-05,
+      "loss": 0.8653,
+      "step": 3660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465561987365294e-05,
+      "loss": 0.749,
+      "step": 3661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460355296484303e-05,
+      "loss": 0.7797,
+      "step": 3662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455150155281513e-05,
+      "loss": 0.7343,
+      "step": 3663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449946565533781e-05,
+      "loss": 0.7533,
+      "step": 3664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444744529017427e-05,
+      "loss": 0.8264,
+      "step": 3665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439544047508254e-05,
+      "loss": 0.7652,
+      "step": 3666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4343451227815276e-05,
+      "loss": 0.7963,
+      "step": 3667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429147756611981e-05,
+      "loss": 0.838,
+      "step": 3668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42395195077381e-05,
+      "loss": 0.8013,
+      "step": 3669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4187577070406917e-05,
+      "loss": 0.7515,
+      "step": 3670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4135650271857576e-05,
+      "loss": 0.2392,
+      "step": 3671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408373912981606e-05,
+      "loss": 0.7406,
+      "step": 3672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4031843662003124e-05,
+      "loss": 0.7762,
+      "step": 3673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397996388613404e-05,
+      "loss": 0.7295,
+      "step": 3674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39280998199188e-05,
+      "loss": 0.8017,
+      "step": 3675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3876251481061944e-05,
+      "loss": 0.7695,
+      "step": 3676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382441888726279e-05,
+      "loss": 0.7585,
+      "step": 3677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3772602056215184e-05,
+      "loss": 0.245,
+      "step": 3678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372080100560753e-05,
+      "loss": 0.7764,
+      "step": 3679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366901575312303e-05,
+      "loss": 0.7584,
+      "step": 3680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361724631643932e-05,
+      "loss": 0.8201,
+      "step": 3681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3565492713228723e-05,
+      "loss": 0.7776,
+      "step": 3682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351375496115809e-05,
+      "loss": 0.7622,
+      "step": 3683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346203307788898e-05,
+      "loss": 0.7162,
+      "step": 3684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3410327081077404e-05,
+      "loss": 0.244,
+      "step": 3685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335863698837399e-05,
+      "loss": 0.7572,
+      "step": 3686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3306962817424036e-05,
+      "loss": 0.2651,
+      "step": 3687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3255304585867274e-05,
+      "loss": 0.7745,
+      "step": 3688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3203662311338e-05,
+      "loss": 0.8122,
+      "step": 3689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315203601146519e-05,
+      "loss": 0.8039,
+      "step": 3690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3100425703872195e-05,
+      "loss": 0.7728,
+      "step": 3691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304883140617709e-05,
+      "loss": 0.8416,
+      "step": 3692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299725313599235e-05,
+      "loss": 0.7656,
+      "step": 3693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2945690910924964e-05,
+      "loss": 0.7549,
+      "step": 3694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289414474857659e-05,
+      "loss": 0.7443,
+      "step": 3695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2842614666543256e-05,
+      "loss": 0.7694,
+      "step": 3696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279110068241554e-05,
+      "loss": 0.7127,
+      "step": 3697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273960281377859e-05,
+      "loss": 0.832,
+      "step": 3698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268812107821199e-05,
+      "loss": 0.8015,
+      "step": 3699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2636655493289824e-05,
+      "loss": 0.7621,
+      "step": 3700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2585206076580604e-05,
+      "loss": 0.8248,
+      "step": 3701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25337728456475e-05,
+      "loss": 0.6734,
+      "step": 3702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248235581804799e-05,
+      "loss": 0.8086,
+      "step": 3703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243095501133403e-05,
+      "loss": 0.2618,
+      "step": 3704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2379570443052164e-05,
+      "loss": 0.799,
+      "step": 3705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232820213074327e-05,
+      "loss": 0.7643,
+      "step": 3706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22768500919427e-05,
+      "loss": 0.7569,
+      "step": 3707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222551434418034e-05,
+      "loss": 0.7405,
+      "step": 3708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217419490498039e-05,
+      "loss": 0.7698,
+      "step": 3709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212289179186154e-05,
+      "loss": 0.8099,
+      "step": 3710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207160502233688e-05,
+      "loss": 0.7575,
+      "step": 3711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202033461391401e-05,
+      "loss": 0.75,
+      "step": 3712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1969080584094854e-05,
+      "loss": 0.7935,
+      "step": 3713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191784295037572e-05,
+      "loss": 0.7623,
+      "step": 3714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186662173024741e-05,
+      "loss": 0.7482,
+      "step": 3715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181541694119513e-05,
+      "loss": 0.6752,
+      "step": 3716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176422860069838e-05,
+      "loss": 0.7497,
+      "step": 3717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1713056726231105e-05,
+      "loss": 0.7921,
+      "step": 3718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1661901335261567e-05,
+      "loss": 0.7019,
+      "step": 3719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161076244525254e-05,
+      "loss": 0.7833,
+      "step": 3720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1559640073661044e-05,
+      "loss": 0.7818,
+      "step": 3721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150853423793844e-05,
+      "loss": 0.782,
+      "step": 3722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14574449555306e-05,
+      "loss": 0.85,
+      "step": 3723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140637224387756e-05,
+      "loss": 0.7815,
+      "step": 3724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135531612041384e-05,
+      "loss": 0.7289,
+      "step": 3725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130427660256818e-05,
+      "loss": 0.7007,
+      "step": 3726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1253253707763774e-05,
+      "loss": 0.756,
+      "step": 3727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120224745341804e-05,
+      "loss": 0.7032,
+      "step": 3728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1151257856942747e-05,
+      "loss": 0.7999,
+      "step": 3729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110028493574403e-05,
+      "loss": 0.8188,
+      "step": 3730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1049328707222254e-05,
+      "loss": 0.7818,
+      "step": 3731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099838918877211e-05,
+      "loss": 0.7878,
+      "step": 3732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094746639778263e-05,
+      "loss": 0.7614,
+      "step": 3733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089656035163708e-05,
+      "loss": 0.781,
+      "step": 3734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084567106771303e-05,
+      "loss": 0.7914,
+      "step": 3735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079479856338229e-05,
+      "loss": 0.7533,
+      "step": 3736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074394285601105e-05,
+      "loss": 0.7773,
+      "step": 3737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0693103962959664e-05,
+      "loss": 0.7296,
+      "step": 3738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064228190158273e-05,
+      "loss": 0.7001,
+      "step": 3739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059147668922919e-05,
+      "loss": 0.7848,
+      "step": 3740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0540688343242236e-05,
+      "loss": 0.7456,
+      "step": 3741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048991688095922e-05,
+      "loss": 0.7173,
+      "step": 3742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043916231971175e-05,
+      "loss": 0.7755,
+      "step": 3743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0388424676825674e-05,
+      "loss": 0.7389,
+      "step": 3744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033770396962112e-05,
+      "loss": 0.6976,
+      "step": 3745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0287000215412386e-05,
+      "loss": 0.2818,
+      "step": 3746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023631343150792e-05,
+      "loss": 0.8004,
+      "step": 3747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018564363521052e-05,
+      "loss": 0.7519,
+      "step": 3748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0134990843817074e-05,
+      "loss": 0.7049,
+      "step": 3749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0084355074618684e-05,
+      "loss": 0.7896,
+      "step": 3750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0033736344900696e-05,
+      "loss": 0.7415,
+      "step": 3751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9983134671942576e-05,
+      "loss": 0.7445,
+      "step": 3752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9932550073017994e-05,
+      "loss": 0.7625,
+      "step": 3753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988198256539475e-05,
+      "loss": 0.798,
+      "step": 3754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9831432166334915e-05,
+      "loss": 0.788,
+      "step": 3755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978089889309463e-05,
+      "loss": 0.2461,
+      "step": 3756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973038276292415e-05,
+      "loss": 0.7215,
+      "step": 3757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967988379306805e-05,
+      "loss": 0.8099,
+      "step": 3758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962940200076488e-05,
+      "loss": 0.7884,
+      "step": 3759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957893740324737e-05,
+      "loss": 0.8041,
+      "step": 3760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952849001774239e-05,
+      "loss": 0.8423,
+      "step": 3761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947805986147098e-05,
+      "loss": 0.761,
+      "step": 3762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9427646951648245e-05,
+      "loss": 0.7499,
+      "step": 3763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937725130548335e-05,
+      "loss": 0.7295,
+      "step": 3764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932687294017969e-05,
+      "loss": 0.7944,
+      "step": 3765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927651187293474e-05,
+      "loss": 0.7934,
+      "step": 3766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9226168120939986e-05,
+      "loss": 0.8206,
+      "step": 3767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917584170138102e-05,
+      "loss": 0.7907,
+      "step": 3768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912553263143761e-05,
+      "loss": 0.7729,
+      "step": 3769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907524092828351e-05,
+      "loss": 0.7816,
+      "step": 3770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9024966609086565e-05,
+      "loss": 0.7497,
+      "step": 3771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897470969100867e-05,
+      "loss": 0.8133,
+      "step": 3772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892447019120587e-05,
+      "loss": 0.773,
+      "step": 3773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887424812682816e-05,
+      "loss": 0.7862,
+      "step": 3774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882404351501959e-05,
+      "loss": 0.761,
+      "step": 3775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877385637291835e-05,
+      "loss": 0.7619,
+      "step": 3776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872368671765657e-05,
+      "loss": 0.773,
+      "step": 3777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867353456636045e-05,
+      "loss": 0.7819,
+      "step": 3778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862339993615014e-05,
+      "loss": 0.7493,
+      "step": 3779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857328284413998e-05,
+      "loss": 0.774,
+      "step": 3780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852318330743815e-05,
+      "loss": 0.8074,
+      "step": 3781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8473101343146895e-05,
+      "loss": 0.7722,
+      "step": 3782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8423036968362535e-05,
+      "loss": 0.7779,
+      "step": 3783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837299020017527e-05,
+      "loss": 0.246,
+      "step": 3784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832296105566937e-05,
+      "loss": 0.7388,
+      "step": 3785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8272949551923e-05,
+      "loss": 0.751,
+      "step": 3786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822295570600845e-05,
+      "loss": 0.8182,
+      "step": 3787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817297953499186e-05,
+      "loss": 0.7456,
+      "step": 3788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8123021055933317e-05,
+      "loss": 0.7422,
+      "step": 3789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807308028588697e-05,
+      "loss": 0.7788,
+      "step": 3790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802315724190092e-05,
+      "loss": 0.7578,
+      "step": 3791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797325194101714e-05,
+      "loss": 0.8058,
+      "step": 3792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792336440027152e-05,
+      "loss": 0.6909,
+      "step": 3793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787349463669404e-05,
+      "loss": 0.7391,
+      "step": 3794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7823642667308476e-05,
+      "loss": 0.691,
+      "step": 3795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777380850913258e-05,
+      "loss": 0.7832,
+      "step": 3796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772399217917796e-05,
+      "loss": 0.7556,
+      "step": 3797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7674193694450274e-05,
+      "loss": 0.7734,
+      "step": 3798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7624413071948984e-05,
+      "loss": 0.8142,
+      "step": 3799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7574650328667425e-05,
+      "loss": 0.776,
+      "step": 3800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7524905481592966e-05,
+      "loss": 0.7642,
+      "step": 3801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7475178547706746e-05,
+      "loss": 0.7932,
+      "step": 3802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7425469543983814e-05,
+      "loss": 0.7298,
+      "step": 3803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7375778487393086e-05,
+      "loss": 0.7201,
+      "step": 3804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732610539489745e-05,
+      "loss": 0.7253,
+      "step": 3805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7276450283453535e-05,
+      "loss": 0.8051,
+      "step": 3806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722681317001186e-05,
+      "loss": 0.6779,
+      "step": 3807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7177194071516906e-05,
+      "loss": 0.8057,
+      "step": 3808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7127593004906875e-05,
+      "loss": 0.25,
+      "step": 3809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707800998711386e-05,
+      "loss": 0.6895,
+      "step": 3810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7028445035063764e-05,
+      "loss": 0.7778,
+      "step": 3811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697889816567641e-05,
+      "loss": 0.755,
+      "step": 3812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6929369395865375e-05,
+      "loss": 0.7761,
+      "step": 3813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687985874253802e-05,
+      "loss": 0.7973,
+      "step": 3814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683036622259562e-05,
+      "loss": 0.732,
+      "step": 3815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6780891852933264e-05,
+      "loss": 0.7631,
+      "step": 3816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6731435650439745e-05,
+      "loss": 0.7479,
+      "step": 3817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6681997631997665e-05,
+      "loss": 0.7773,
+      "step": 3818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6632577814483546e-05,
+      "loss": 0.7801,
+      "step": 3819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6583176214767554e-05,
+      "loss": 0.784,
+      "step": 3820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653379284971372e-05,
+      "loss": 0.7525,
+      "step": 3821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648442773617976e-05,
+      "loss": 0.7607,
+      "step": 3822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64350808910173e-05,
+      "loss": 0.7868,
+      "step": 3823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638575233107163e-05,
+      "loss": 0.7948,
+      "step": 3824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6336442073181763e-05,
+      "loss": 0.7685,
+      "step": 3825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628715013418061e-05,
+      "loss": 0.7685,
+      "step": 3826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6237876530894706e-05,
+      "loss": 0.7568,
+      "step": 3827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618862128014435e-05,
+      "loss": 0.7711,
+      "step": 3828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613938439874355e-05,
+      "loss": 0.7592,
+      "step": 3829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609016590350015e-05,
+      "loss": 0.7991,
+      "step": 3830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6040965811215616e-05,
+      "loss": 0.8393,
+      "step": 3831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599178413868514e-05,
+      "loss": 0.7776,
+      "step": 3832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59426209026977e-05,
+      "loss": 0.7572,
+      "step": 3833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5893476120035896e-05,
+      "loss": 0.7778,
+      "step": 3834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584434980747609e-05,
+      "loss": 0.7948,
+      "step": 3835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579524198178823e-05,
+      "loss": 0.7199,
+      "step": 3836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574615265973615e-05,
+      "loss": 0.7651,
+      "step": 3837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5697081858077154e-05,
+      "loss": 0.7468,
+      "step": 3838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5648029593562404e-05,
+      "loss": 0.7265,
+      "step": 3839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559899588293658e-05,
+      "loss": 0.7474,
+      "step": 3840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554998074293817e-05,
+      "loss": 0.7919,
+      "step": 3841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5500984190299226e-05,
+      "loss": 0.7735,
+      "step": 3842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545200624174541e-05,
+      "loss": 0.8116,
+      "step": 3843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540304691399624e-05,
+      "loss": 0.8185,
+      "step": 3844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535410622376465e-05,
+      "loss": 0.7288,
+      "step": 3845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530518418775733e-05,
+      "loss": 0.7438,
+      "step": 3846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5256280822674544e-05,
+      "loss": 0.7881,
+      "step": 3847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5207396145210266e-05,
+      "loss": 0.7617,
+      "step": 3848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515853017205203e-05,
+      "loss": 0.6843,
+      "step": 3849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5109682919880935e-05,
+      "loss": 0.7751,
+      "step": 3850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5060854405371844e-05,
+      "loss": 0.2601,
+      "step": 3851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5012044645193076e-05,
+      "loss": 0.7801,
+      "step": 3852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4963253656006596e-05,
+      "loss": 0.7079,
+      "step": 3853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491448145446795e-05,
+      "loss": 0.7375,
+      "step": 3854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486572805722634e-05,
+      "loss": 0.7663,
+      "step": 3855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4816993480924466e-05,
+      "loss": 0.8055,
+      "step": 3856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476827774219859e-05,
+      "loss": 0.7622,
+      "step": 3857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471958085767868e-05,
+      "loss": 0.746,
+      "step": 3858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4670902843988106e-05,
+      "loss": 0.7463,
+      "step": 3859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462224371774388e-05,
+      "loss": 0.2539,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457360349555653e-05,
+      "loss": 0.7815,
+      "step": 3861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4524982194030185e-05,
+      "loss": 0.744,
+      "step": 3862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4476379829762436e-05,
+      "loss": 0.7645,
+      "step": 3863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442779641934452e-05,
+      "loss": 0.7416,
+      "step": 3864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437923197936108e-05,
+      "loss": 0.7829,
+      "step": 3865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4330686526390394e-05,
+      "loss": 0.778,
+      "step": 3866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428216007700418e-05,
+      "loss": 0.7464,
+      "step": 3867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4233652647767645e-05,
+      "loss": 0.7657,
+      "step": 3868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418516425523964e-05,
+      "loss": 0.8158,
+      "step": 3869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41366949159724e-05,
+      "loss": 0.7795,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408824464651167e-05,
+      "loss": 0.7571,
+      "step": 3871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4039813463396676e-05,
+      "loss": 0.7945,
+      "step": 3872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399140138316022e-05,
+      "loss": 0.7363,
+      "step": 3873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394300842232849e-05,
+      "loss": 0.7,
+      "step": 3874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3894634597421124e-05,
+      "loss": 0.7643,
+      "step": 3875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384627992495136e-05,
+      "loss": 0.7205,
+      "step": 3876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379794442142578e-05,
+      "loss": 0.7574,
+      "step": 3877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374962810334447e-05,
+      "loss": 0.7907,
+      "step": 3878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3701330987200905e-05,
+      "loss": 0.2452,
+      "step": 3879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365305308948212e-05,
+      "loss": 0.7752,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36047944266685e-05,
+      "loss": 0.7042,
+      "step": 3881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355655501523386e-05,
+      "loss": 0.7331,
+      "step": 3882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350833487164553e-05,
+      "loss": 0.782,
+      "step": 3883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3460134012364174e-05,
+      "loss": 0.7498,
+      "step": 3884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3411952453843864e-05,
+      "loss": 0.7594,
+      "step": 3885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336379021253221e-05,
+      "loss": 0.7668,
+      "step": 3886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3315647304870046e-05,
+      "loss": 0.7667,
+      "step": 3887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3267523747291784e-05,
+      "loss": 0.2359,
+      "step": 3888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321941955622512e-05,
+      "loss": 0.7901,
+      "step": 3889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317133474809112e-05,
+      "loss": 0.7481,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312326933930435e-05,
+      "loss": 0.7367,
+      "step": 3891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3075223346272666e-05,
+      "loss": 0.7574,
+      "step": 3892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302719678539725e-05,
+      "loss": 0.7628,
+      "step": 3893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2979189673072804e-05,
+      "loss": 0.8024,
+      "step": 3894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293120202568727e-05,
+      "loss": 0.7539,
+      "step": 3895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288323385962195e-05,
+      "loss": 0.8164,
+      "step": 3896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283528519125152e-05,
+      "loss": 0.7555,
+      "step": 3897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278735603694405e-05,
+      "loss": 0.7548,
+      "step": 3898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273944641306088e-05,
+      "loss": 0.8014,
+      "step": 3899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269155633595666e-05,
+      "loss": 0.7463,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264368582197947e-05,
+      "loss": 0.7873,
+      "step": 3901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2595834887470655e-05,
+      "loss": 0.739,
+      "step": 3902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2548003548764804e-05,
+      "loss": 0.7511,
+      "step": 3903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250019182218996e-05,
+      "loss": 0.8039,
+      "step": 3904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2452399724067396e-05,
+      "loss": 0.8282,
+      "step": 3905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240462727071164e-05,
+      "loss": 0.7132,
+      "step": 3906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2356874478430565e-05,
+      "loss": 0.7637,
+      "step": 3907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2309141363525375e-05,
+      "loss": 0.7782,
+      "step": 3908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.226142794229048e-05,
+      "loss": 0.2534,
+      "step": 3909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221373423101356e-05,
+      "loss": 0.7897,
+      "step": 3910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216606024597569e-05,
+      "loss": 0.7623,
+      "step": 3911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2118406003451025e-05,
+      "loss": 0.7456,
+      "step": 3912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2070771519707186e-05,
+      "loss": 0.7536,
+      "step": 3913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202315681100488e-05,
+      "loss": 0.769,
+      "step": 3914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1975561893598114e-05,
+      "loss": 0.2319,
+      "step": 3915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1927986783734206e-05,
+      "loss": 0.7685,
+      "step": 3916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1880431497653626e-05,
+      "loss": 0.8127,
+      "step": 3917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183289605159007e-05,
+      "loss": 0.7566,
+      "step": 3918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178538046177058e-05,
+      "loss": 0.7841,
+      "step": 3919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1737884744415314e-05,
+      "loss": 0.7833,
+      "step": 3920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169040891573766e-05,
+      "loss": 0.774,
+      "step": 3921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164295299194419e-05,
+      "loss": 0.7011,
+      "step": 3922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1595516989234785e-05,
+      "loss": 0.7709,
+      "step": 3923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1548100923802446e-05,
+      "loss": 0.8201,
+      "step": 3924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1500704811833346e-05,
+      "loss": 0.7611,
+      "step": 3925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145332866950693e-05,
+      "loss": 0.783,
+      "step": 3926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1405972512995774e-05,
+      "loss": 0.7332,
+      "step": 3927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135863635846558e-05,
+      "loss": 0.8121,
+      "step": 3928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131132022207537e-05,
+      "loss": 0.749,
+      "step": 3929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12640241199772e-05,
+      "loss": 0.7905,
+      "step": 3930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12167480683163e-05,
+      "loss": 0.7206,
+      "step": 3931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116949208323109e-05,
+      "loss": 0.7295,
+      "step": 3932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1122256180853193e-05,
+      "loss": 0.7894,
+      "step": 3933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107504037730726e-05,
+      "loss": 0.7294,
+      "step": 3934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102784468871114e-05,
+      "loss": 0.7938,
+      "step": 3935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0980669131175806e-05,
+      "loss": 0.791,
+      "step": 3936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093351372080545e-05,
+      "loss": 0.709,
+      "step": 3937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088637847369723e-05,
+      "loss": 0.7442,
+      "step": 3938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083926340594152e-05,
+      "loss": 0.7772,
+      "step": 3939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079216853362172e-05,
+      "loss": 0.7683,
+      "step": 3940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074509387281449e-05,
+      "loss": 0.7386,
+      "step": 3941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0698039439589455e-05,
+      "loss": 0.7489,
+      "step": 3942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0651005250009324e-05,
+      "loss": 0.7663,
+      "step": 3943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060399132013004e-05,
+      "loss": 0.7344,
+      "step": 3944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055699766600048e-05,
+      "loss": 0.745,
+      "step": 3945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0510024303662656e-05,
+      "loss": 0.8178,
+      "step": 3946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0463071249151695e-05,
+      "loss": 0.7425,
+      "step": 3947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0416138518495726e-05,
+      "loss": 0.7513,
+      "step": 3948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036922612771598e-05,
+      "loss": 0.7197,
+      "step": 3949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0322334092826664e-05,
+      "loss": 0.707,
+      "step": 3950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0275462429835184e-05,
+      "loss": 0.7987,
+      "step": 3951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0228611154741894e-05,
+      "loss": 0.7427,
+      "step": 3952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0181780283540146e-05,
+      "loss": 0.7856,
+      "step": 3953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013496983221646e-05,
+      "loss": 0.7339,
+      "step": 3954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0088179816750296e-05,
+      "loss": 0.7538,
+      "step": 3955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004141025311412e-05,
+      "loss": 0.7361,
+      "step": 3956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999466115727343e-05,
+      "loss": 0.7359,
+      "step": 3957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9947932545186826e-05,
+      "loss": 0.6768,
+      "step": 3958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990122443280582e-05,
+      "loss": 0.7034,
+      "step": 3959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985453683607488e-05,
+      "loss": 0.7504,
+      "step": 3960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980786977093163e-05,
+      "loss": 0.6925,
+      "step": 3961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976122325330659e-05,
+      "loss": 0.7194,
+      "step": 3962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9714597299123254e-05,
+      "loss": 0.7972,
+      "step": 3963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9667991924298075e-05,
+      "loss": 0.2447,
+      "step": 3964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9621407144740596e-05,
+      "loss": 0.7823,
+      "step": 3965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957484297635322e-05,
+      "loss": 0.7301,
+      "step": 3966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9528299435031354e-05,
+      "loss": 0.679,
+      "step": 3967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948177653666332e-05,
+      "loss": 0.2735,
+      "step": 3968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9435274297130485e-05,
+      "loss": 0.7758,
+      "step": 3969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93887927323071e-05,
+      "loss": 0.7401,
+      "step": 3970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934233185806031e-05,
+      "loss": 0.7643,
+      "step": 3971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9295891690250354e-05,
+      "loss": 0.8129,
+      "step": 3972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924947224473024e-05,
+      "loss": 0.2239,
+      "step": 3973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920307353734599e-05,
+      "loss": 0.2684,
+      "step": 3974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9156695583936474e-05,
+      "loss": 0.7089,
+      "step": 3975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911033840033361e-05,
+      "loss": 0.7365,
+      "step": 3976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9064002002362086e-05,
+      "loss": 0.7834,
+      "step": 3977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901768640583954e-05,
+      "loss": 0.7953,
+      "step": 3978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897139162657657e-05,
+      "loss": 0.7335,
+      "step": 3979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89251176803766e-05,
+      "loss": 0.7614,
+      "step": 3980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887886458303593e-05,
+      "loss": 0.7864,
+      "step": 3981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8832632350343776e-05,
+      "loss": 0.7735,
+      "step": 3982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8786420998082286e-05,
+      "loss": 0.7827,
+      "step": 3983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874023054202638e-05,
+      "loss": 0.7904,
+      "step": 3984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869406099794385e-05,
+      "loss": 0.7449,
+      "step": 3985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8647912381595416e-05,
+      "loss": 0.2586,
+      "step": 3986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860178470873469e-05,
+      "loss": 0.7513,
+      "step": 3987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8555677995108006e-05,
+      "loss": 0.6591,
+      "step": 3988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8509592256454586e-05,
+      "loss": 0.8026,
+      "step": 3989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846352750850657e-05,
+      "loss": 0.7871,
+      "step": 3990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841748376698885e-05,
+      "loss": 0.7484,
+      "step": 3991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837146104761916e-05,
+      "loss": 0.7625,
+      "step": 3992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832545936610805e-05,
+      "loss": 0.7467,
+      "step": 3993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827947873815898e-05,
+      "loss": 0.7456,
+      "step": 3994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8233519179468105e-05,
+      "loss": 0.698,
+      "step": 3995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818758070572439e-05,
+      "loss": 0.7389,
+      "step": 3996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814166333260973e-05,
+      "loss": 0.8039,
+      "step": 3997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809576707579871e-05,
+      "loss": 0.8152,
+      "step": 3998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8049891950958704e-05,
+      "loss": 0.7748,
+      "step": 3999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800403797374988e-05,
+      "loss": 0.2641,
+      "step": 4000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795820515982528e-05,
+      "loss": 0.7431,
+      "step": 4001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.791239352483058e-05,
+      "loss": 0.6776,
+      "step": 4002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7866603084404295e-05,
+      "loss": 0.673,
+      "step": 4003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7820833854177764e-05,
+      "loss": 0.7702,
+      "step": 4004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.777508584977497e-05,
+      "loss": 0.7275,
+      "step": 4005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.772935908681272e-05,
+      "loss": 0.6983,
+      "step": 4006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768365358090051e-05,
+      "loss": 0.7385,
+      "step": 4007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763796934764071e-05,
+      "loss": 0.7732,
+      "step": 4008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759230640262831e-05,
+      "loss": 0.7268,
+      "step": 4009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7546664761451e-05,
+      "loss": 0.7268,
+      "step": 4010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.750104443968933e-05,
+      "loss": 0.7807,
+      "step": 4011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7455445452916517e-05,
+      "loss": 0.7834,
+      "step": 4012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740986781669846e-05,
+      "loss": 0.2671,
+      "step": 4013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7364311546593757e-05,
+      "loss": 0.7445,
+      "step": 4014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731877665815381e-05,
+      "loss": 0.757,
+      "step": 4015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727326316692265e-05,
+      "loss": 0.7791,
+      "step": 4016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722777108843697e-05,
+      "loss": 0.2741,
+      "step": 4017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71823004382262e-05,
+      "loss": 0.7757,
+      "step": 4018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713685123181252e-05,
+      "loss": 0.798,
+      "step": 4019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7091423484710665e-05,
+      "loss": 0.7586,
+      "step": 4020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704601721242809e-05,
+      "loss": 0.7276,
+      "step": 4021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7000632430464977e-05,
+      "loss": 0.7847,
+      "step": 4022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695526915431411e-05,
+      "loss": 0.8013,
+      "step": 4023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690992739946094e-05,
+      "loss": 0.7917,
+      "step": 4024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6864607181383546e-05,
+      "loss": 0.7671,
+      "step": 4025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6819308515552767e-05,
+      "loss": 0.719,
+      "step": 4026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6774031417431953e-05,
+      "loss": 0.7654,
+      "step": 4027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.672877590247712e-05,
+      "loss": 0.8314,
+      "step": 4028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6683541986137015e-05,
+      "loss": 0.7669,
+      "step": 4029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663832968385289e-05,
+      "loss": 0.7701,
+      "step": 4030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659313901105867e-05,
+      "loss": 0.7933,
+      "step": 4031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6547969983180874e-05,
+      "loss": 0.7055,
+      "step": 4032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65028226156387e-05,
+      "loss": 0.7685,
+      "step": 4033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6457696923843864e-05,
+      "loss": 0.7852,
+      "step": 4034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6412592923200714e-05,
+      "loss": 0.7811,
+      "step": 4035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636751062910622e-05,
+      "loss": 0.6848,
+      "step": 4036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6322450056949955e-05,
+      "loss": 0.733,
+      "step": 4037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6277411222114e-05,
+      "loss": 0.2514,
+      "step": 4038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623239413997306e-05,
+      "loss": 0.7234,
+      "step": 4039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618739882589445e-05,
+      "loss": 0.2793,
+      "step": 4040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.614242529523799e-05,
+      "loss": 0.7282,
+      "step": 4041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609747356335611e-05,
+      "loss": 0.6961,
+      "step": 4042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.605254364559374e-05,
+      "loss": 0.7594,
+      "step": 4043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6007635557288475e-05,
+      "loss": 0.6699,
+      "step": 4044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596274931377034e-05,
+      "loss": 0.8328,
+      "step": 4045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5917884930361934e-05,
+      "loss": 0.7457,
+      "step": 4046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587304242237847e-05,
+      "loss": 0.7738,
+      "step": 4047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5828221805127606e-05,
+      "loss": 0.7387,
+      "step": 4048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578342309390954e-05,
+      "loss": 0.7313,
+      "step": 4049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5738646304017e-05,
+      "loss": 0.791,
+      "step": 4050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5693891450735284e-05,
+      "loss": 0.3139,
+      "step": 4051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564915854934213e-05,
+      "loss": 0.8252,
+      "step": 4052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5604447615107774e-05,
+      "loss": 0.7881,
+      "step": 4053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5559758663295065e-05,
+      "loss": 0.7606,
+      "step": 4054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551509170915921e-05,
+      "loss": 0.7815,
+      "step": 4055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547044676794799e-05,
+      "loss": 0.8276,
+      "step": 4056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542582385490162e-05,
+      "loss": 0.7647,
+      "step": 4057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538122298525288e-05,
+      "loss": 0.7759,
+      "step": 4058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5336644174226895e-05,
+      "loss": 0.7095,
+      "step": 4059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529208743704142e-05,
+      "loss": 0.7997,
+      "step": 4060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524755278890651e-05,
+      "loss": 0.7044,
+      "step": 4061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520304024502482e-05,
+      "loss": 0.7717,
+      "step": 4062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515854982059138e-05,
+      "loss": 0.7269,
+      "step": 4063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511408153079364e-05,
+      "loss": 0.8158,
+      "step": 4064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.506963539081162e-05,
+      "loss": 0.7829,
+      "step": 4065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502521141581766e-05,
+      "loss": 0.2396,
+      "step": 4066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4980809620976564e-05,
+      "loss": 0.7427,
+      "step": 4067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493643002144557e-05,
+      "loss": 0.7895,
+      "step": 4068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4892072632374376e-05,
+      "loss": 0.699,
+      "step": 4069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484773746890504e-05,
+      "loss": 0.7068,
+      "step": 4070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4803424546172034e-05,
+      "loss": 0.7396,
+      "step": 4071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475913387930234e-05,
+      "loss": 0.7778,
+      "step": 4072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471486548341521e-05,
+      "loss": 0.8229,
+      "step": 4073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467061937362236e-05,
+      "loss": 0.7543,
+      "step": 4074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4626395565027846e-05,
+      "loss": 0.7534,
+      "step": 4075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458219407272821e-05,
+      "loss": 0.7111,
+      "step": 4076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4538014911812325e-05,
+      "loss": 0.7736,
+      "step": 4077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4493858097361345e-05,
+      "loss": 0.7804,
+      "step": 4078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4449723644448985e-05,
+      "loss": 0.7992,
+      "step": 4079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4405611568141195e-05,
+      "loss": 0.7273,
+      "step": 4080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.436152188349626e-05,
+      "loss": 0.7667,
+      "step": 4081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.431745460556496e-05,
+      "loss": 0.7079,
+      "step": 4082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427340974939032e-05,
+      "loss": 0.8093,
+      "step": 4083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422938733000768e-05,
+      "loss": 0.8285,
+      "step": 4084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4185387362444853e-05,
+      "loss": 0.7639,
+      "step": 4085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414140986172185e-05,
+      "loss": 0.7674,
+      "step": 4086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409745484285114e-05,
+      "loss": 0.7586,
+      "step": 4087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4053522320837414e-05,
+      "loss": 0.7417,
+      "step": 4088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4009612310677685e-05,
+      "loss": 0.7274,
+      "step": 4089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396572482736139e-05,
+      "loss": 0.7441,
+      "step": 4090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3921859885870166e-05,
+      "loss": 0.8009,
+      "step": 4091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387801750117798e-05,
+      "loss": 0.7696,
+      "step": 4092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38341976882511e-05,
+      "loss": 0.7488,
+      "step": 4093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.379040046204815e-05,
+      "loss": 0.7532,
+      "step": 4094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374662583751996e-05,
+      "loss": 0.8244,
+      "step": 4095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370287382960966e-05,
+      "loss": 0.7807,
+      "step": 4096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.365914445325272e-05,
+      "loss": 0.7878,
+      "step": 4097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361543772337683e-05,
+      "loss": 0.7413,
+      "step": 4098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3571753654901915e-05,
+      "loss": 0.7776,
+      "step": 4099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352809226274028e-05,
+      "loss": 0.8196,
+      "step": 4100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348445356179639e-05,
+      "loss": 0.7252,
+      "step": 4101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3440837566966987e-05,
+      "loss": 0.7598,
+      "step": 4102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3397244293141027e-05,
+      "loss": 0.6944,
+      "step": 4103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335367375519984e-05,
+      "loss": 0.7464,
+      "step": 4104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.331012596801686e-05,
+      "loss": 0.7964,
+      "step": 4105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326660094645776e-05,
+      "loss": 0.7047,
+      "step": 4106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322309870538055e-05,
+      "loss": 0.7228,
+      "step": 4107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317961925963534e-05,
+      "loss": 0.7917,
+      "step": 4108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3136162624064574e-05,
+      "loss": 0.7025,
+      "step": 4109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3092728813502815e-05,
+      "loss": 0.8083,
+      "step": 4110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304931784277684e-05,
+      "loss": 0.7602,
+      "step": 4111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300592972670571e-05,
+      "loss": 0.7821,
+      "step": 4112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296256448010062e-05,
+      "loss": 0.7881,
+      "step": 4113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2919222117764924e-05,
+      "loss": 0.7293,
+      "step": 4114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2875902654494275e-05,
+      "loss": 0.7687,
+      "step": 4115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283260610507643e-05,
+      "loss": 0.7045,
+      "step": 4116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2789332484291266e-05,
+      "loss": 0.8081,
+      "step": 4117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274608180691101e-05,
+      "loss": 0.764,
+      "step": 4118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270285408769991e-05,
+      "loss": 0.7406,
+      "step": 4119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.265964934141441e-05,
+      "loss": 0.7978,
+      "step": 4120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.261646758280309e-05,
+      "loss": 0.7449,
+      "step": 4121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2573308826606786e-05,
+      "loss": 0.7958,
+      "step": 4122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253017308755837e-05,
+      "loss": 0.7722,
+      "step": 4123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248706038038286e-05,
+      "loss": 0.7496,
+      "step": 4124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2443970719797534e-05,
+      "loss": 0.7904,
+      "step": 4125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2400904120511654e-05,
+      "loss": 0.7712,
+      "step": 4126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2357860597226696e-05,
+      "loss": 0.7811,
+      "step": 4127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2314840164636195e-05,
+      "loss": 0.7842,
+      "step": 4128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227184283742591e-05,
+      "loss": 0.7973,
+      "step": 4129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.222886863027361e-05,
+      "loss": 0.7884,
+      "step": 4130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2185917557849176e-05,
+      "loss": 0.844,
+      "step": 4131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214298963481468e-05,
+      "loss": 0.761,
+      "step": 4132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210008487582419e-05,
+      "loss": 0.7161,
+      "step": 4133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2057203295523965e-05,
+      "loss": 0.7903,
+      "step": 4134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201434490855226e-05,
+      "loss": 0.689,
+      "step": 4135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197150972953943e-05,
+      "loss": 0.7777,
+      "step": 4136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1928697773107994e-05,
+      "loss": 0.7577,
+      "step": 4137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.188590905387243e-05,
+      "loss": 0.7781,
+      "step": 4138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1843143586439315e-05,
+      "loss": 0.7773,
+      "step": 4139
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180040138540736e-05,
+      "loss": 0.741,
+      "step": 4140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1757682465367253e-05,
+      "loss": 0.7199,
+      "step": 4141
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171498684090173e-05,
+      "loss": 0.7501,
+      "step": 4142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1672314526585655e-05,
+      "loss": 0.7686,
+      "step": 4143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162966553698587e-05,
+      "loss": 0.782,
+      "step": 4144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.158703988666126e-05,
+      "loss": 0.7287,
+      "step": 4145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1544437590162735e-05,
+      "loss": 0.7754,
+      "step": 4146
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150185866203329e-05,
+      "loss": 0.7577,
+      "step": 4147
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145930311680789e-05,
+      "loss": 0.8252,
+      "step": 4148
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1416770969013496e-05,
+      "loss": 0.6873,
+      "step": 4149
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137426223316918e-05,
+      "loss": 0.7067,
+      "step": 4150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.133177692378593e-05,
+      "loss": 0.8045,
+      "step": 4151
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128931505536674e-05,
+      "loss": 0.7726,
+      "step": 4152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124687664240664e-05,
+      "loss": 0.7215,
+      "step": 4153
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1204461699392674e-05,
+      "loss": 0.7183,
+      "step": 4154
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116207024080381e-05,
+      "loss": 0.752,
+      "step": 4155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1119702281111016e-05,
+      "loss": 0.7008,
+      "step": 4156
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107735783477728e-05,
+      "loss": 0.7305,
+      "step": 4157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103503691625756e-05,
+      "loss": 0.2601,
+      "step": 4158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0992739539998715e-05,
+      "loss": 0.7754,
+      "step": 4159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0950465720439617e-05,
+      "loss": 0.7869,
+      "step": 4160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.090821547201111e-05,
+      "loss": 0.8372,
+      "step": 4161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086598880913598e-05,
+      "loss": 0.8488,
+      "step": 4162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0823785746228925e-05,
+      "loss": 0.7399,
+      "step": 4163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078160629769661e-05,
+      "loss": 0.7348,
+      "step": 4164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073945047793768e-05,
+      "loss": 0.7701,
+      "step": 4165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069731830134266e-05,
+      "loss": 0.7303,
+      "step": 4166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0655209782294004e-05,
+      "loss": 0.7273,
+      "step": 4167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.061312493516615e-05,
+      "loss": 0.7245,
+      "step": 4168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0571063774325396e-05,
+      "loss": 0.7757,
+      "step": 4169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.052902631412995e-05,
+      "loss": 0.6928,
+      "step": 4170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.048701256892995e-05,
+      "loss": 0.7355,
+      "step": 4171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0445022553067504e-05,
+      "loss": 0.2563,
+      "step": 4172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0403056280876495e-05,
+      "loss": 0.2484,
+      "step": 4173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0361113766682747e-05,
+      "loss": 0.7171,
+      "step": 4174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0319195024804058e-05,
+      "loss": 0.6926,
+      "step": 4175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0277300069550006e-05,
+      "loss": 0.7095,
+      "step": 4176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0235428915222076e-05,
+      "loss": 0.7907,
+      "step": 4177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0193581576113606e-05,
+      "loss": 0.7757,
+      "step": 4178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0151758066509905e-05,
+      "loss": 0.7441,
+      "step": 4179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0109958400688055e-05,
+      "loss": 0.732,
+      "step": 4180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006818259291696e-05,
+      "loss": 0.7571,
+      "step": 4181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0026430657457504e-05,
+      "loss": 0.749,
+      "step": 4182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9984702608562364e-05,
+      "loss": 0.7448,
+      "step": 4183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9942998460476047e-05,
+      "loss": 0.7467,
+      "step": 4184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990131822743486e-05,
+      "loss": 0.693,
+      "step": 4185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9859661923667082e-05,
+      "loss": 0.7759,
+      "step": 4186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981802956339269e-05,
+      "loss": 0.7976,
+      "step": 4187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9776421160823553e-05,
+      "loss": 0.7697,
+      "step": 4188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9734836730163296e-05,
+      "loss": 0.7824,
+      "step": 4189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9693276285607475e-05,
+      "loss": 0.7726,
+      "step": 4190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9651739841343372e-05,
+      "loss": 0.7509,
+      "step": 4191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9610227411550063e-05,
+      "loss": 0.6398,
+      "step": 4192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9568739010398526e-05,
+      "loss": 0.7825,
+      "step": 4193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9527274652051428e-05,
+      "loss": 0.6994,
+      "step": 4194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948583435066329e-05,
+      "loss": 0.8402,
+      "step": 4195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9444418120380348e-05,
+      "loss": 0.2514,
+      "step": 4196
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9403025975340747e-05,
+      "loss": 0.7697,
+      "step": 4197
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9361657929674313e-05,
+      "loss": 0.7818,
+      "step": 4198
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9320313997502635e-05,
+      "loss": 0.2508,
+      "step": 4199
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927899419293917e-05,
+      "loss": 0.789,
+      "step": 4200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9237698530089042e-05,
+      "loss": 0.722,
+      "step": 4201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9196427023049177e-05,
+      "loss": 0.2474,
+      "step": 4202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9155179685908196e-05,
+      "loss": 0.71,
+      "step": 4203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911395653274659e-05,
+      "loss": 0.7721,
+      "step": 4204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907275757763649e-05,
+      "loss": 0.7624,
+      "step": 4205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903158283464176e-05,
+      "loss": 0.28,
+      "step": 4206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8990432317818085e-05,
+      "loss": 0.2862,
+      "step": 4207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8949306041212853e-05,
+      "loss": 0.7412,
+      "step": 4208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8908204018865124e-05,
+      "loss": 0.7747,
+      "step": 4209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.886712626480569e-05,
+      "loss": 0.7274,
+      "step": 4210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8826072793057137e-05,
+      "loss": 0.7374,
+      "step": 4211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8785043617633668e-05,
+      "loss": 0.7392,
+      "step": 4212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8744038752541237e-05,
+      "loss": 0.7334,
+      "step": 4213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8703058211777466e-05,
+      "loss": 0.7611,
+      "step": 4214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866210200933175e-05,
+      "loss": 0.7732,
+      "step": 4215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8621170159185085e-05,
+      "loss": 0.7611,
+      "step": 4216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8580262675310188e-05,
+      "loss": 0.6698,
+      "step": 4217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853937957167151e-05,
+      "loss": 0.7073,
+      "step": 4218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8498520862225096e-05,
+      "loss": 0.764,
+      "step": 4219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8457686560918718e-05,
+      "loss": 0.7761,
+      "step": 4220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841687668169175e-05,
+      "loss": 0.7257,
+      "step": 4221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8376091238475356e-05,
+      "loss": 0.7866,
+      "step": 4222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8335330245192238e-05,
+      "loss": 0.7058,
+      "step": 4223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829459371575677e-05,
+      "loss": 0.7758,
+      "step": 4224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8253881664075042e-05,
+      "loss": 0.7545,
+      "step": 4225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821319410404475e-05,
+      "loss": 0.6851,
+      "step": 4226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8172531049555195e-05,
+      "loss": 0.7931,
+      "step": 4227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8131892514487323e-05,
+      "loss": 0.7685,
+      "step": 4228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8091278512713783e-05,
+      "loss": 0.7127,
+      "step": 4229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8050689058098768e-05,
+      "loss": 0.6398,
+      "step": 4230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8010124164498098e-05,
+      "loss": 0.7709,
+      "step": 4231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796958384575924e-05,
+      "loss": 0.7775,
+      "step": 4232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79290681157213e-05,
+      "loss": 0.823,
+      "step": 4233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7888576988214932e-05,
+      "loss": 0.3031,
+      "step": 4234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7848110477062372e-05,
+      "loss": 0.7683,
+      "step": 4235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7807668596077553e-05,
+      "loss": 0.7358,
+      "step": 4236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7767251359065903e-05,
+      "loss": 0.7737,
+      "step": 4237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7726858779824483e-05,
+      "loss": 0.7019,
+      "step": 4238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768649087214188e-05,
+      "loss": 0.73,
+      "step": 4239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7646147649798393e-05,
+      "loss": 0.7255,
+      "step": 4240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7605829126565752e-05,
+      "loss": 0.703,
+      "step": 4241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7565535316207304e-05,
+      "loss": 0.2444,
+      "step": 4242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7525266232478007e-05,
+      "loss": 0.7439,
+      "step": 4243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7485021889124318e-05,
+      "loss": 0.7383,
+      "step": 4244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7444802299884264e-05,
+      "loss": 0.7464,
+      "step": 4245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74046074784874e-05,
+      "loss": 0.7691,
+      "step": 4246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73644374386549e-05,
+      "loss": 0.8348,
+      "step": 4247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7324292194099423e-05,
+      "loss": 0.7679,
+      "step": 4248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7284171758525122e-05,
+      "loss": 0.7779,
+      "step": 4249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7244076145627806e-05,
+      "loss": 0.7493,
+      "step": 4250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7204005369094698e-05,
+      "loss": 0.829,
+      "step": 4251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716395944260457e-05,
+      "loss": 0.7509,
+      "step": 4252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7123938379827707e-05,
+      "loss": 0.7674,
+      "step": 4253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7083942194425972e-05,
+      "loss": 0.7288,
+      "step": 4254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7043970900052617e-05,
+      "loss": 0.8143,
+      "step": 4255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7004024510352543e-05,
+      "loss": 0.2357,
+      "step": 4256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6964103038961984e-05,
+      "loss": 0.7834,
+      "step": 4257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6924206499508843e-05,
+      "loss": 0.8019,
+      "step": 4258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6884334905612373e-05,
+      "loss": 0.7804,
+      "step": 4259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684448827088334e-05,
+      "loss": 0.7888,
+      "step": 4260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6804666608924066e-05,
+      "loss": 0.7731,
+      "step": 4261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6764869933328275e-05,
+      "loss": 0.7542,
+      "step": 4262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672509825768117e-05,
+      "loss": 0.7751,
+      "step": 4263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6685351595559395e-05,
+      "loss": 0.7526,
+      "step": 4264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.664562996053116e-05,
+      "loss": 0.8119,
+      "step": 4265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6605933366156032e-05,
+      "loss": 0.7656,
+      "step": 4266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.656626182598503e-05,
+      "loss": 0.7485,
+      "step": 4267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6526615353560702e-05,
+      "loss": 0.7407,
+      "step": 4268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648699396241696e-05,
+      "loss": 0.7359,
+      "step": 4269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6447397666079176e-05,
+      "loss": 0.776,
+      "step": 4270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6407826478064144e-05,
+      "loss": 0.2889,
+      "step": 4271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6368280411880164e-05,
+      "loss": 0.6653,
+      "step": 4272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6328759481026855e-05,
+      "loss": 0.7489,
+      "step": 4273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6289263698995283e-05,
+      "loss": 0.7365,
+      "step": 4274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624979307926799e-05,
+      "loss": 0.7683,
+      "step": 4275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6210347635318875e-05,
+      "loss": 0.711,
+      "step": 4276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617092738061322e-05,
+      "loss": 0.7547,
+      "step": 4277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6131532328607787e-05,
+      "loss": 0.7714,
+      "step": 4278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.609216249275066e-05,
+      "loss": 0.7375,
+      "step": 4279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605281788648133e-05,
+      "loss": 0.8049,
+      "step": 4280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601349852323074e-05,
+      "loss": 0.8018,
+      "step": 4281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59742044164211e-05,
+      "loss": 0.7733,
+      "step": 4282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5934935579466126e-05,
+      "loss": 0.7308,
+      "step": 4283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5895692025770823e-05,
+      "loss": 0.8041,
+      "step": 4284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585647376873155e-05,
+      "loss": 0.7158,
+      "step": 4285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581728082173612e-05,
+      "loss": 0.7916,
+      "step": 4286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5778113198163645e-05,
+      "loss": 0.7656,
+      "step": 4287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5738970911384585e-05,
+      "loss": 0.7942,
+      "step": 4288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5699853974760744e-05,
+      "loss": 0.7628,
+      "step": 4289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5660762401645354e-05,
+      "loss": 0.2847,
+      "step": 4290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5621696205382896e-05,
+      "loss": 0.7015,
+      "step": 4291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5582655399309195e-05,
+      "loss": 0.2622,
+      "step": 4292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554363999675151e-05,
+      "loss": 0.7707,
+      "step": 4293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5504650011028308e-05,
+      "loss": 0.7376,
+      "step": 4294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5465685455449394e-05,
+      "loss": 0.7034,
+      "step": 4295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5426746343315987e-05,
+      "loss": 0.7262,
+      "step": 4296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5387832687920544e-05,
+      "loss": 0.7749,
+      "step": 4297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5348944502546824e-05,
+      "loss": 0.7041,
+      "step": 4298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5310081800469898e-05,
+      "loss": 0.7321,
+      "step": 4299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52712445949562e-05,
+      "loss": 0.7941,
+      "step": 4300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523243289926339e-05,
+      "loss": 0.7172,
+      "step": 4301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519364672664041e-05,
+      "loss": 0.7516,
+      "step": 4302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515488609032759e-05,
+      "loss": 0.771,
+      "step": 4303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511615100355643e-05,
+      "loss": 0.7695,
+      "step": 4304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5077441479549736e-05,
+      "loss": 0.7704,
+      "step": 4305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5038757531521662e-05,
+      "loss": 0.756,
+      "step": 4306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5000099172677514e-05,
+      "loss": 0.7684,
+      "step": 4307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4961466416213984e-05,
+      "loss": 0.808,
+      "step": 4308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492285927531893e-05,
+      "loss": 0.7221,
+      "step": 4309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4884277763171483e-05,
+      "loss": 0.7335,
+      "step": 4310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4845721892942074e-05,
+      "loss": 0.7993,
+      "step": 4311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480719167779234e-05,
+      "loss": 0.7705,
+      "step": 4312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4768687130875125e-05,
+      "loss": 0.7833,
+      "step": 4313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473020826533462e-05,
+      "loss": 0.7289,
+      "step": 4314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4691755094306158e-05,
+      "loss": 0.8076,
+      "step": 4315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4653327630916323e-05,
+      "loss": 0.7294,
+      "step": 4316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461492588828289e-05,
+      "loss": 0.7396,
+      "step": 4317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4576549879514955e-05,
+      "loss": 0.7276,
+      "step": 4318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453819961771273e-05,
+      "loss": 0.7444,
+      "step": 4319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4499875115967664e-05,
+      "loss": 0.7698,
+      "step": 4320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446157638736244e-05,
+      "loss": 0.7517,
+      "step": 4321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4423303444970926e-05,
+      "loss": 0.7532,
+      "step": 4322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438505630185818e-05,
+      "loss": 0.7735,
+      "step": 4323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4346834971080423e-05,
+      "loss": 0.7493,
+      "step": 4324
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4308639465685158e-05,
+      "loss": 0.7389,
+      "step": 4325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427046979871098e-05,
+      "loss": 0.7658,
+      "step": 4326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4232325983187676e-05,
+      "loss": 0.7149,
+      "step": 4327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419420803213629e-05,
+      "loss": 0.6959,
+      "step": 4328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4156115958568916e-05,
+      "loss": 0.7654,
+      "step": 4329
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4118049775488925e-05,
+      "loss": 0.7912,
+      "step": 4330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4080009495890788e-05,
+      "loss": 0.8195,
+      "step": 4331
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40419951327601e-05,
+      "loss": 0.8038,
+      "step": 4332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4004006699073712e-05,
+      "loss": 0.7702,
+      "step": 4333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396604420779954e-05,
+      "loss": 0.7963,
+      "step": 4334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392810767189666e-05,
+      "loss": 0.7594,
+      "step": 4335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3890197104315325e-05,
+      "loss": 0.7337,
+      "step": 4336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3852312517996878e-05,
+      "loss": 0.7229,
+      "step": 4337
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3814453925873793e-05,
+      "loss": 0.6762,
+      "step": 4338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377662134086973e-05,
+      "loss": 0.7892,
+      "step": 4339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3738814775899397e-05,
+      "loss": 0.744,
+      "step": 4340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3701034243868682e-05,
+      "loss": 0.7254,
+      "step": 4341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3663279757674495e-05,
+      "loss": 0.746,
+      "step": 4342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3625551330204985e-05,
+      "loss": 0.7696,
+      "step": 4343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3587848974339322e-05,
+      "loss": 0.7627,
+      "step": 4344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3550172702947737e-05,
+      "loss": 0.7007,
+      "step": 4345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3512522528891678e-05,
+      "loss": 0.8032,
+      "step": 4346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3474898465023598e-05,
+      "loss": 0.7179,
+      "step": 4347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3437300524187044e-05,
+      "loss": 0.778,
+      "step": 4348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339972871921663e-05,
+      "loss": 0.7766,
+      "step": 4349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3362183062938124e-05,
+      "loss": 0.7793,
+      "step": 4350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3324663568168304e-05,
+      "loss": 0.7582,
+      "step": 4351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3287170247715007e-05,
+      "loss": 0.7528,
+      "step": 4352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3249703114377197e-05,
+      "loss": 0.7417,
+      "step": 4353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3212262180944822e-05,
+      "loss": 0.7066,
+      "step": 4354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317484746019898e-05,
+      "loss": 0.771,
+      "step": 4355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3137458964911695e-05,
+      "loss": 0.7235,
+      "step": 4356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3100096707846188e-05,
+      "loss": 0.7763,
+      "step": 4357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30627607017566e-05,
+      "loss": 0.7756,
+      "step": 4358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3025450959388183e-05,
+      "loss": 0.7193,
+      "step": 4359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.298816749347713e-05,
+      "loss": 0.7448,
+      "step": 4360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2950910316750818e-05,
+      "loss": 0.7775,
+      "step": 4361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2913679441927527e-05,
+      "loss": 0.7945,
+      "step": 4362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2876474881716557e-05,
+      "loss": 0.7789,
+      "step": 4363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2839296648818333e-05,
+      "loss": 0.7262,
+      "step": 4364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280214475592417e-05,
+      "loss": 0.736,
+      "step": 4365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2765019215716464e-05,
+      "loss": 0.7758,
+      "step": 4366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2727920040868568e-05,
+      "loss": 0.7637,
+      "step": 4367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2690847244044912e-05,
+      "loss": 0.2949,
+      "step": 4368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265380083790084e-05,
+      "loss": 0.7576,
+      "step": 4369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2616780835082683e-05,
+      "loss": 0.688,
+      "step": 4370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.257978724822787e-05,
+      "loss": 0.7627,
+      "step": 4371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.254282008996469e-05,
+      "loss": 0.26,
+      "step": 4372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2505879372912465e-05,
+      "loss": 0.2527,
+      "step": 4373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246896510968145e-05,
+      "loss": 0.7443,
+      "step": 4374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2432077312872966e-05,
+      "loss": 0.7483,
+      "step": 4375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23952159950792e-05,
+      "loss": 0.8127,
+      "step": 4376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2358381168883303e-05,
+      "loss": 0.7684,
+      "step": 4377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232157284685945e-05,
+      "loss": 0.2521,
+      "step": 4378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2284791041572762e-05,
+      "loss": 0.8216,
+      "step": 4379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2248035765579244e-05,
+      "loss": 0.7603,
+      "step": 4380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2211307031425865e-05,
+      "loss": 0.8336,
+      "step": 4381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2174604851650592e-05,
+      "loss": 0.7916,
+      "step": 4382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213792923878225e-05,
+      "loss": 0.7355,
+      "step": 4383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2101280205340658e-05,
+      "loss": 0.7741,
+      "step": 4384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206465776383647e-05,
+      "loss": 0.7419,
+      "step": 4385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2028061926771402e-05,
+      "loss": 0.7423,
+      "step": 4386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1991492706637984e-05,
+      "loss": 0.7298,
+      "step": 4387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195495011591965e-05,
+      "loss": 0.7719,
+      "step": 4388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.191843416709084e-05,
+      "loss": 0.751,
+      "step": 4389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188194487261681e-05,
+      "loss": 0.8298,
+      "step": 4390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1845482244953753e-05,
+      "loss": 0.7705,
+      "step": 4391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1809046296548718e-05,
+      "loss": 0.7387,
+      "step": 4392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1772637039839737e-05,
+      "loss": 0.7916,
+      "step": 4393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1736254487255657e-05,
+      "loss": 0.2285,
+      "step": 4394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16998986512162e-05,
+      "loss": 0.7373,
+      "step": 4395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1663569544132033e-05,
+      "loss": 0.7338,
+      "step": 4396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1627267178404643e-05,
+      "loss": 0.7774,
+      "step": 4397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159099156642641e-05,
+      "loss": 0.7496,
+      "step": 4398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1554742720580544e-05,
+      "loss": 0.696,
+      "step": 4399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1518520653241213e-05,
+      "loss": 0.7462,
+      "step": 4400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148232537677336e-05,
+      "loss": 0.7947,
+      "step": 4401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1446156903532756e-05,
+      "loss": 0.7454,
+      "step": 4402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141001524586611e-05,
+      "loss": 0.7589,
+      "step": 4403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1373900416110972e-05,
+      "loss": 0.7771,
+      "step": 4404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133781242659566e-05,
+      "loss": 0.7758,
+      "step": 4405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1301751289639348e-05,
+      "loss": 0.2658,
+      "step": 4406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1265717017552124e-05,
+      "loss": 0.713,
+      "step": 4407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122970962263482e-05,
+      "loss": 0.7609,
+      "step": 4408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1193729117179106e-05,
+      "loss": 0.741,
+      "step": 4409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115777551346747e-05,
+      "loss": 0.7698,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1121848823773283e-05,
+      "loss": 0.7519,
+      "step": 4411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1085949060360654e-05,
+      "loss": 0.7312,
+      "step": 4412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1050076235484495e-05,
+      "loss": 0.7081,
+      "step": 4413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1014230361390607e-05,
+      "loss": 0.7358,
+      "step": 4414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0978411450315504e-05,
+      "loss": 0.79,
+      "step": 4415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.094261951448654e-05,
+      "loss": 0.7937,
+      "step": 4416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0906854566121815e-05,
+      "loss": 0.7227,
+      "step": 4417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0871116617430286e-05,
+      "loss": 0.7544,
+      "step": 4418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0835405680611665e-05,
+      "loss": 0.7662,
+      "step": 4419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0799721767856374e-05,
+      "loss": 0.7624,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076406489134575e-05,
+      "loss": 0.6824,
+      "step": 4421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.072843506325177e-05,
+      "loss": 0.7234,
+      "step": 4422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069283229573724e-05,
+      "loss": 0.7876,
+      "step": 4423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06572566009557e-05,
+      "loss": 0.7789,
+      "step": 4424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0621707991051508e-05,
+      "loss": 0.7379,
+      "step": 4425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058618647815972e-05,
+      "loss": 0.7643,
+      "step": 4426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055069207440611e-05,
+      "loss": 0.7326,
+      "step": 4427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0515224791907283e-05,
+      "loss": 0.7796,
+      "step": 4428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0479784642770572e-05,
+      "loss": 0.6841,
+      "step": 4429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0444371639093997e-05,
+      "loss": 0.7645,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0408985792966305e-05,
+      "loss": 0.7669,
+      "step": 4431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0373627116467066e-05,
+      "loss": 0.7012,
+      "step": 4432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033829562166647e-05,
+      "loss": 0.7264,
+      "step": 4433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0302991320625485e-05,
+      "loss": 0.282,
+      "step": 4434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026771422539575e-05,
+      "loss": 0.7892,
+      "step": 4435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023246434801971e-05,
+      "loss": 0.7395,
+      "step": 4436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0197241700530423e-05,
+      "loss": 0.7311,
+      "step": 4437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0162046294951665e-05,
+      "loss": 0.7276,
+      "step": 4438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0126878143297976e-05,
+      "loss": 0.7562,
+      "step": 4439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0091737257574538e-05,
+      "loss": 0.809,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005662364977724e-05,
+      "loss": 0.7398,
+      "step": 4441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0021537331892614e-05,
+      "loss": 0.7506,
+      "step": 4442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9986478315897982e-05,
+      "loss": 0.7394,
+      "step": 4443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9951446613761272e-05,
+      "loss": 0.7364,
+      "step": 4444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9916442237441048e-05,
+      "loss": 0.7922,
+      "step": 4445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988146519888665e-05,
+      "loss": 0.8432,
+      "step": 4446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9846515510038035e-05,
+      "loss": 0.8172,
+      "step": 4447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9811593182825793e-05,
+      "loss": 0.726,
+      "step": 4448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9776698229171185e-05,
+      "loss": 0.797,
+      "step": 4449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741830660986206e-05,
+      "loss": 0.7699,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9706990490173405e-05,
+      "loss": 0.7629,
+      "step": 4451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9672177728625996e-05,
+      "loss": 0.6885,
+      "step": 4452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9637392388227883e-05,
+      "loss": 0.7166,
+      "step": 4453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9602634480853598e-05,
+      "loss": 0.7765,
+      "step": 4454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9567904018368276e-05,
+      "loss": 0.7885,
+      "step": 4455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9533201012627666e-05,
+      "loss": 0.7854,
+      "step": 4456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9498525475478247e-05,
+      "loss": 0.799,
+      "step": 4457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9463877418757005e-05,
+      "loss": 0.7997,
+      "step": 4458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.942925685429159e-05,
+      "loss": 0.7436,
+      "step": 4459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939466379390027e-05,
+      "loss": 0.7603,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9360098249391944e-05,
+      "loss": 0.7455,
+      "step": 4461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9325560232566097e-05,
+      "loss": 0.2546,
+      "step": 4462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9291049755212776e-05,
+      "loss": 0.8084,
+      "step": 4463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925656682911272e-05,
+      "loss": 0.7508,
+      "step": 4464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9222111466037196e-05,
+      "loss": 0.767,
+      "step": 4465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918768367774807e-05,
+      "loss": 0.7541,
+      "step": 4466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9153283475997773e-05,
+      "loss": 0.77,
+      "step": 4467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9118910872529406e-05,
+      "loss": 0.8047,
+      "step": 4468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9084565879076554e-05,
+      "loss": 0.7763,
+      "step": 4469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9050248507363412e-05,
+      "loss": 0.7682,
+      "step": 4470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9015958769104768e-05,
+      "loss": 0.2633,
+      "step": 4471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8981696676005954e-05,
+      "loss": 0.8123,
+      "step": 4472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8947462239762837e-05,
+      "loss": 0.7682,
+      "step": 4473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8913255472061918e-05,
+      "loss": 0.7292,
+      "step": 4474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8879076384580196e-05,
+      "loss": 0.7861,
+      "step": 4475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8844924988985192e-05,
+      "loss": 0.7764,
+      "step": 4476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8810801296935066e-05,
+      "loss": 0.7618,
+      "step": 4477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8776705320078436e-05,
+      "loss": 0.8218,
+      "step": 4478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8742637070054524e-05,
+      "loss": 0.7347,
+      "step": 4479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8708596558493042e-05,
+      "loss": 0.7801,
+      "step": 4480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.867458379701422e-05,
+      "loss": 0.7515,
+      "step": 4481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.864059879722889e-05,
+      "loss": 0.7407,
+      "step": 4482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8606641570738346e-05,
+      "loss": 0.8138,
+      "step": 4483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8572712129134397e-05,
+      "loss": 0.7588,
+      "step": 4484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.853881048399937e-05,
+      "loss": 0.6918,
+      "step": 4485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.850493664690617e-05,
+      "loss": 0.7285,
+      "step": 4486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847109062941812e-05,
+      "loss": 0.7354,
+      "step": 4487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843727244308907e-05,
+      "loss": 0.7886,
+      "step": 4488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.840348209946343e-05,
+      "loss": 0.7563,
+      "step": 4489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8369719610076043e-05,
+      "loss": 0.7047,
+      "step": 4490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8335984986452204e-05,
+      "loss": 0.7594,
+      "step": 4491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830227824010784e-05,
+      "loss": 0.7629,
+      "step": 4492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826859938254921e-05,
+      "loss": 0.758,
+      "step": 4493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8234948425273147e-05,
+      "loss": 0.7576,
+      "step": 4494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8201325379766876e-05,
+      "loss": 0.7767,
+      "step": 4495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.816773025750822e-05,
+      "loss": 0.8166,
+      "step": 4496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8134163069965347e-05,
+      "loss": 0.7659,
+      "step": 4497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810062382859693e-05,
+      "loss": 0.238,
+      "step": 4498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806711254485215e-05,
+      "loss": 0.7139,
+      "step": 4499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.803362923017059e-05,
+      "loss": 0.7315,
+      "step": 4500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8000173895982263e-05,
+      "loss": 0.7544,
+      "step": 4501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7966746553707716e-05,
+      "loss": 0.7596,
+      "step": 4502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7933347214757857e-05,
+      "loss": 0.6964,
+      "step": 4503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7899975890534093e-05,
+      "loss": 0.7575,
+      "step": 4504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7866632592428244e-05,
+      "loss": 0.7054,
+      "step": 4505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783331733182253e-05,
+      "loss": 0.8541,
+      "step": 4506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7800030120089673e-05,
+      "loss": 0.7184,
+      "step": 4507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7766770968592762e-05,
+      "loss": 0.8129,
+      "step": 4508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7733539888685303e-05,
+      "loss": 0.6833,
+      "step": 4509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7700336891711278e-05,
+      "loss": 0.7378,
+      "step": 4510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7667161989005043e-05,
+      "loss": 0.7946,
+      "step": 4511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763401519189134e-05,
+      "loss": 0.7746,
+      "step": 4512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7600896511685338e-05,
+      "loss": 0.7828,
+      "step": 4513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.756780595969265e-05,
+      "loss": 0.7798,
+      "step": 4514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.753474354720923e-05,
+      "loss": 0.7871,
+      "step": 4515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7501709285521427e-05,
+      "loss": 0.7632,
+      "step": 4516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7468703185906045e-05,
+      "loss": 0.7663,
+      "step": 4517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74357252596302e-05,
+      "loss": 0.7269,
+      "step": 4518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7402775517951408e-05,
+      "loss": 0.7892,
+      "step": 4519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7369853972117568e-05,
+      "loss": 0.8161,
+      "step": 4520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7336960633367007e-05,
+      "loss": 0.7686,
+      "step": 4521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7304095512928352e-05,
+      "loss": 0.2471,
+      "step": 4522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7271258622020593e-05,
+      "loss": 0.6636,
+      "step": 4523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7238449971853167e-05,
+      "loss": 0.2504,
+      "step": 4524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7205669573625795e-05,
+      "loss": 0.758,
+      "step": 4525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172917438528523e-05,
+      "loss": 0.7414,
+      "step": 4526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714019357774189e-05,
+      "loss": 0.8056,
+      "step": 4527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.710749800243662e-05,
+      "loss": 0.7444,
+      "step": 4528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074830723773916e-05,
+      "loss": 0.7266,
+      "step": 4529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042191752905225e-05,
+      "loss": 0.7726,
+      "step": 4530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.700958110097234e-05,
+      "loss": 0.7687,
+      "step": 4531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697699877910748e-05,
+      "loss": 0.697,
+      "step": 4532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694444479843309e-05,
+      "loss": 0.732,
+      "step": 4533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.691191917006195e-05,
+      "loss": 0.7508,
+      "step": 4534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687942190509725e-05,
+      "loss": 0.7833,
+      "step": 4535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6846953014632394e-05,
+      "loss": 0.7837,
+      "step": 4536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6814512509751146e-05,
+      "loss": 0.7682,
+      "step": 4537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6782100401527566e-05,
+      "loss": 0.7957,
+      "step": 4538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6749716701026063e-05,
+      "loss": 0.7445,
+      "step": 4539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6717361419301302e-05,
+      "loss": 0.7847,
+      "step": 4540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668503456739824e-05,
+      "loss": 0.7601,
+      "step": 4541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6652736156352178e-05,
+      "loss": 0.7513,
+      "step": 4542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6620466197188665e-05,
+      "loss": 0.7726,
+      "step": 4543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6588224700923573e-05,
+      "loss": 0.7718,
+      "step": 4544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655601167856299e-05,
+      "loss": 0.7418,
+      "step": 4545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6523827141103376e-05,
+      "loss": 0.7545,
+      "step": 4546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6491671099531414e-05,
+      "loss": 0.7733,
+      "step": 4547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6459543564824044e-05,
+      "loss": 0.7114,
+      "step": 4548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6427444547948532e-05,
+      "loss": 0.7035,
+      "step": 4549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6395374059862344e-05,
+      "loss": 0.7313,
+      "step": 4550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6363332111513274e-05,
+      "loss": 0.7802,
+      "step": 4551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6331318713839293e-05,
+      "loss": 0.7758,
+      "step": 4552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.629933387776873e-05,
+      "loss": 0.7505,
+      "step": 4553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626737761422008e-05,
+      "loss": 0.7623,
+      "step": 4554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6235449934102098e-05,
+      "loss": 0.6759,
+      "step": 4555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6203550848313775e-05,
+      "loss": 0.7382,
+      "step": 4556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.61716803677444e-05,
+      "loss": 0.7643,
+      "step": 4557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6139838503273453e-05,
+      "loss": 0.6865,
+      "step": 4558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6108025265770598e-05,
+      "loss": 0.7214,
+      "step": 4559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6076240666095844e-05,
+      "loss": 0.7725,
+      "step": 4560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6044484715099327e-05,
+      "loss": 0.7188,
+      "step": 4561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.601275742362145e-05,
+      "loss": 0.7842,
+      "step": 4562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5981058802492767e-05,
+      "loss": 0.7372,
+      "step": 4563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5949388862534152e-05,
+      "loss": 0.7766,
+      "step": 4564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.591774761455661e-05,
+      "loss": 0.7308,
+      "step": 4565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5886135069361342e-05,
+      "loss": 0.7174,
+      "step": 4566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5854551237739836e-05,
+      "loss": 0.7968,
+      "step": 4567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58229961304737e-05,
+      "loss": 0.8104,
+      "step": 4568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5791469758334755e-05,
+      "loss": 0.7746,
+      "step": 4569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5759972132084998e-05,
+      "loss": 0.7806,
+      "step": 4570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.572850326247668e-05,
+      "loss": 0.7486,
+      "step": 4571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5697063160252157e-05,
+      "loss": 0.7612,
+      "step": 4572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566565183614398e-05,
+      "loss": 0.2507,
+      "step": 4573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5634269300874938e-05,
+      "loss": 0.7595,
+      "step": 4574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5602915565157893e-05,
+      "loss": 0.7625,
+      "step": 4575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5571590639695986e-05,
+      "loss": 0.7163,
+      "step": 4576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.554029453518241e-05,
+      "loss": 0.7771,
+      "step": 4577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5509027262300636e-05,
+      "loss": 0.8001,
+      "step": 4578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5477788831724206e-05,
+      "loss": 0.2498,
+      "step": 4579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5446579254116835e-05,
+      "loss": 0.2408,
+      "step": 4580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5415398540132374e-05,
+      "loss": 0.8186,
+      "step": 4581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5384246700414895e-05,
+      "loss": 0.7492,
+      "step": 4582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5353123745598553e-05,
+      "loss": 0.799,
+      "step": 4583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5322029686307594e-05,
+      "loss": 0.7942,
+      "step": 4584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5290964533156538e-05,
+      "loss": 0.2453,
+      "step": 4585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5259928296749915e-05,
+      "loss": 0.6983,
+      "step": 4586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5228920987682438e-05,
+      "loss": 0.7356,
+      "step": 4587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5197942616538896e-05,
+      "loss": 0.7253,
+      "step": 4588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5166993193894297e-05,
+      "loss": 0.7725,
+      "step": 4589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5136072730313678e-05,
+      "loss": 0.8151,
+      "step": 4590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5105181236352195e-05,
+      "loss": 0.7503,
+      "step": 4591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5074318722555193e-05,
+      "loss": 0.767,
+      "step": 4592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5043485199458029e-05,
+      "loss": 0.7055,
+      "step": 4593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5012680677586222e-05,
+      "loss": 0.7996,
+      "step": 4594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4981905167455346e-05,
+      "loss": 0.7334,
+      "step": 4595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4951158679571142e-05,
+      "loss": 0.7642,
+      "step": 4596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4920441224429383e-05,
+      "loss": 0.7854,
+      "step": 4597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4889752812515922e-05,
+      "loss": 0.7906,
+      "step": 4598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4859093454306749e-05,
+      "loss": 0.2459,
+      "step": 4599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4828463160267947e-05,
+      "loss": 0.7384,
+      "step": 4600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4797861940855595e-05,
+      "loss": 0.7371,
+      "step": 4601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4767289806515894e-05,
+      "loss": 0.7244,
+      "step": 4602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4736746767685138e-05,
+      "loss": 0.7455,
+      "step": 4603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470623283478968e-05,
+      "loss": 0.7947,
+      "step": 4604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4675748018245883e-05,
+      "loss": 0.7077,
+      "step": 4605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4645292328460226e-05,
+      "loss": 0.7376,
+      "step": 4606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4614865775829256e-05,
+      "loss": 0.6858,
+      "step": 4607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4584468370739534e-05,
+      "loss": 0.762,
+      "step": 4608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455410012356767e-05,
+      "loss": 0.7449,
+      "step": 4609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4523761044680363e-05,
+      "loss": 0.7849,
+      "step": 4610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4493451144434322e-05,
+      "loss": 0.72,
+      "step": 4611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4463170433176299e-05,
+      "loss": 0.78,
+      "step": 4612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4432918921243055e-05,
+      "loss": 0.8044,
+      "step": 4613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4402696618961465e-05,
+      "loss": 0.7693,
+      "step": 4614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4372503536648362e-05,
+      "loss": 0.7316,
+      "step": 4615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4342339684610585e-05,
+      "loss": 0.7713,
+      "step": 4616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4312205073145102e-05,
+      "loss": 0.7888,
+      "step": 4617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4282099712538788e-05,
+      "loss": 0.7677,
+      "step": 4618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4252023613068588e-05,
+      "loss": 0.7022,
+      "step": 4619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4221976785001412e-05,
+      "loss": 0.7447,
+      "step": 4620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4191959238594265e-05,
+      "loss": 0.7557,
+      "step": 4621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4161970984094075e-05,
+      "loss": 0.7588,
+      "step": 4622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4132012031737785e-05,
+      "loss": 0.7698,
+      "step": 4623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4102082391752358e-05,
+      "loss": 0.7486,
+      "step": 4624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4072182074354768e-05,
+      "loss": 0.7632,
+      "step": 4625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4042311089751936e-05,
+      "loss": 0.7714,
+      "step": 4626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4012469448140764e-05,
+      "loss": 0.7498,
+      "step": 4627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.398265715970819e-05,
+      "loss": 0.7821,
+      "step": 4628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3952874234631096e-05,
+      "loss": 0.7908,
+      "step": 4629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3923120683076329e-05,
+      "loss": 0.7504,
+      "step": 4630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.389339651520073e-05,
+      "loss": 0.752,
+      "step": 4631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3863701741151125e-05,
+      "loss": 0.7665,
+      "step": 4632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3834036371064263e-05,
+      "loss": 0.7478,
+      "step": 4633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3804400415066876e-05,
+      "loss": 0.7485,
+      "step": 4634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3774793883275695e-05,
+      "loss": 0.6786,
+      "step": 4635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3745216785797343e-05,
+      "loss": 0.7305,
+      "step": 4636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3715669132728415e-05,
+      "loss": 0.7588,
+      "step": 4637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368615093415545e-05,
+      "loss": 0.7314,
+      "step": 4638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.365666220015498e-05,
+      "loss": 0.7668,
+      "step": 4639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3627202940793426e-05,
+      "loss": 0.7661,
+      "step": 4640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3597773166127147e-05,
+      "loss": 0.7776,
+      "step": 4641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3568372886202485e-05,
+      "loss": 0.8031,
+      "step": 4642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3539002111055664e-05,
+      "loss": 0.728,
+      "step": 4643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3509660850712858e-05,
+      "loss": 0.7501,
+      "step": 4644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3480349115190138e-05,
+      "loss": 0.7601,
+      "step": 4645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.345106691449356e-05,
+      "loss": 0.814,
+      "step": 4646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3421814258619036e-05,
+      "loss": 0.8132,
+      "step": 4647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3392591157552404e-05,
+      "loss": 0.7869,
+      "step": 4648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3363397621269424e-05,
+      "loss": 0.7576,
+      "step": 4649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3334233659735806e-05,
+      "loss": 0.6995,
+      "step": 4650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3305099282907085e-05,
+      "loss": 0.7449,
+      "step": 4651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3275994500728727e-05,
+      "loss": 0.758,
+      "step": 4652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3246919323136119e-05,
+      "loss": 0.7531,
+      "step": 4653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3217873760054533e-05,
+      "loss": 0.7476,
+      "step": 4654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3188857821399104e-05,
+      "loss": 0.7143,
+      "step": 4655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.315987151707485e-05,
+      "loss": 0.7437,
+      "step": 4656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.313091485697674e-05,
+      "loss": 0.2525,
+      "step": 4657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101987850989572e-05,
+      "loss": 0.7652,
+      "step": 4658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3073090508987995e-05,
+      "loss": 0.7532,
+      "step": 4659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3044222840836606e-05,
+      "loss": 0.762,
+      "step": 4660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3015384856389822e-05,
+      "loss": 0.7518,
+      "step": 4661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2986576565491926e-05,
+      "loss": 0.7281,
+      "step": 4662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295779797797706e-05,
+      "loss": 0.6929,
+      "step": 4663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2929049103669277e-05,
+      "loss": 0.785,
+      "step": 4664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.290032995238244e-05,
+      "loss": 0.7135,
+      "step": 4665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2871640533920249e-05,
+      "loss": 0.8094,
+      "step": 4666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2842980858076325e-05,
+      "loss": 0.7566,
+      "step": 4667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2814350934634078e-05,
+      "loss": 0.7745,
+      "step": 4668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785750773366756e-05,
+      "loss": 0.8005,
+      "step": 4669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2757180384037504e-05,
+      "loss": 0.7497,
+      "step": 4670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2728639776399243e-05,
+      "loss": 0.6473,
+      "step": 4671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2700128960194767e-05,
+      "loss": 0.7453,
+      "step": 4672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267164794515665e-05,
+      "loss": 0.7473,
+      "step": 4673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2643196741007357e-05,
+      "loss": 0.7963,
+      "step": 4674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2614775357459174e-05,
+      "loss": 0.7521,
+      "step": 4675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2586383804214153e-05,
+      "loss": 0.7876,
+      "step": 4676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2558022090964172e-05,
+      "loss": 0.7224,
+      "step": 4677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2529690227390978e-05,
+      "loss": 0.8099,
+      "step": 4678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250138822316609e-05,
+      "loss": 0.76,
+      "step": 4679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247311608795082e-05,
+      "loss": 0.76,
+      "step": 4680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2444873831396275e-05,
+      "loss": 0.7301,
+      "step": 4681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2416661463143452e-05,
+      "loss": 0.7879,
+      "step": 4682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2388478992823038e-05,
+      "loss": 0.7156,
+      "step": 4683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2360326430055546e-05,
+      "loss": 0.7606,
+      "step": 4684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2332203784451334e-05,
+      "loss": 0.7564,
+      "step": 4685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2304111065610479e-05,
+      "loss": 0.7531,
+      "step": 4686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2276048283122843e-05,
+      "loss": 0.7576,
+      "step": 4687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2248015446568151e-05,
+      "loss": 0.7289,
+      "step": 4688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2220012565515792e-05,
+      "loss": 0.8055,
+      "step": 4689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2192039649525023e-05,
+      "loss": 0.7167,
+      "step": 4690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216409670814479e-05,
+      "loss": 0.8271,
+      "step": 4691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2136183750913887e-05,
+      "loss": 0.7555,
+      "step": 4692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2108300787360827e-05,
+      "loss": 0.7653,
+      "step": 4693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2080447827003871e-05,
+      "loss": 0.7436,
+      "step": 4694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2052624879351104e-05,
+      "loss": 0.7181,
+      "step": 4695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2024831953900284e-05,
+      "loss": 0.2554,
+      "step": 4696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1997069060138954e-05,
+      "loss": 0.7787,
+      "step": 4697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1969336207544445e-05,
+      "loss": 0.6922,
+      "step": 4698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1941633405583752e-05,
+      "loss": 0.7349,
+      "step": 4699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1913960663713696e-05,
+      "loss": 0.7923,
+      "step": 4700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188631799138078e-05,
+      "loss": 0.7787,
+      "step": 4701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1858705398021242e-05,
+      "loss": 0.7397,
+      "step": 4702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1831122893061108e-05,
+      "loss": 0.754,
+      "step": 4703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1803570485916083e-05,
+      "loss": 0.7115,
+      "step": 4704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1776048185991573e-05,
+      "loss": 0.2613,
+      "step": 4705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1748556002682786e-05,
+      "loss": 0.7702,
+      "step": 4706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1721093945374583e-05,
+      "loss": 0.7392,
+      "step": 4707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1693662023441576e-05,
+      "loss": 0.7774,
+      "step": 4708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.166626024624805e-05,
+      "loss": 0.704,
+      "step": 4709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.163888862314807e-05,
+      "loss": 0.2347,
+      "step": 4710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1611547163485347e-05,
+      "loss": 0.7248,
+      "step": 4711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1584235876593275e-05,
+      "loss": 0.7599,
+      "step": 4712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1556954771795048e-05,
+      "loss": 0.7299,
+      "step": 4713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1529703858403462e-05,
+      "loss": 0.7181,
+      "step": 4714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1502483145721043e-05,
+      "loss": 0.7218,
+      "step": 4715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1475292643039992e-05,
+      "loss": 0.7179,
+      "step": 4716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1448132359642228e-05,
+      "loss": 0.685,
+      "step": 4717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1421002304799333e-05,
+      "loss": 0.7331,
+      "step": 4718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1393902487772557e-05,
+      "loss": 0.7884,
+      "step": 4719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1366832917812864e-05,
+      "loss": 0.7388,
+      "step": 4720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1339793604160865e-05,
+      "loss": 0.8046,
+      "step": 4721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1312784556046829e-05,
+      "loss": 0.7691,
+      "step": 4722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1285805782690729e-05,
+      "loss": 0.7539,
+      "step": 4723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1258857293302206e-05,
+      "loss": 0.7367,
+      "step": 4724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1231939097080535e-05,
+      "loss": 0.2395,
+      "step": 4725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1205051203214656e-05,
+      "loss": 0.7092,
+      "step": 4726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1178193620883148e-05,
+      "loss": 0.7402,
+      "step": 4727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.11513663592543e-05,
+      "loss": 0.7721,
+      "step": 4728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.112456942748601e-05,
+      "loss": 0.7231,
+      "step": 4729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1097802834725778e-05,
+      "loss": 0.7667,
+      "step": 4730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1071066590110868e-05,
+      "loss": 0.8032,
+      "step": 4731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1044360702768064e-05,
+      "loss": 0.7704,
+      "step": 4732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1017685181813864e-05,
+      "loss": 0.2647,
+      "step": 4733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0991040036354327e-05,
+      "loss": 0.7334,
+      "step": 4734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0964425275485235e-05,
+      "loss": 0.7497,
+      "step": 4735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093784090829193e-05,
+      "loss": 0.7784,
+      "step": 4736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0911286943849386e-05,
+      "loss": 0.7282,
+      "step": 4737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088476339122223e-05,
+      "loss": 0.7307,
+      "step": 4738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0858270259464698e-05,
+      "loss": 0.7954,
+      "step": 4739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083180755762061e-05,
+      "loss": 0.7358,
+      "step": 4740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0805375294723408e-05,
+      "loss": 0.7496,
+      "step": 4741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778973479796195e-05,
+      "loss": 0.7671,
+      "step": 4742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0752602121851619e-05,
+      "loss": 0.7719,
+      "step": 4743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072626122989192e-05,
+      "loss": 0.7033,
+      "step": 4744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0699950812909043e-05,
+      "loss": 0.7503,
+      "step": 4745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06736708798844e-05,
+      "loss": 0.7758,
+      "step": 4746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0647421439789062e-05,
+      "loss": 0.7635,
+      "step": 4747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0621202501583693e-05,
+      "loss": 0.7186,
+      "step": 4748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0595014074218556e-05,
+      "loss": 0.7135,
+      "step": 4749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0568856166633455e-05,
+      "loss": 0.7214,
+      "step": 4750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0542728787757795e-05,
+      "loss": 0.7662,
+      "step": 4751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0516631946510546e-05,
+      "loss": 0.6611,
+      "step": 4752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.049056565180031e-05,
+      "loss": 0.6921,
+      "step": 4753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.04645299125252e-05,
+      "loss": 0.7645,
+      "step": 4754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0438524737572908e-05,
+      "loss": 0.8013,
+      "step": 4755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0412550135820731e-05,
+      "loss": 0.7503,
+      "step": 4756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038660611613549e-05,
+      "loss": 0.7741,
+      "step": 4757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0360692687373585e-05,
+      "loss": 0.723,
+      "step": 4758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0334809858380933e-05,
+      "loss": 0.7748,
+      "step": 4759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0308957637993088e-05,
+      "loss": 0.7283,
+      "step": 4760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0283136035035078e-05,
+      "loss": 0.7288,
+      "step": 4761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.025734505832151e-05,
+      "loss": 0.774,
+      "step": 4762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.023158471665655e-05,
+      "loss": 0.754,
+      "step": 4763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0205855018833876e-05,
+      "loss": 0.7218,
+      "step": 4764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0180155973636729e-05,
+      "loss": 0.2489,
+      "step": 4765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0154487589837858e-05,
+      "loss": 0.7609,
+      "step": 4766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0128849876199598e-05,
+      "loss": 0.7761,
+      "step": 4767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0103242841473758e-05,
+      "loss": 0.8222,
+      "step": 4768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0077666494401683e-05,
+      "loss": 0.7622,
+      "step": 4769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0052120843714285e-05,
+      "loss": 0.7617,
+      "step": 4770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0026605898131946e-05,
+      "loss": 0.7308,
+      "step": 4771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000112166636461e-05,
+      "loss": 0.7461,
+      "step": 4772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.975668157111695e-06,
+      "loss": 0.7056,
+      "step": 4773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.950245379062174e-06,
+      "loss": 0.7237,
+      "step": 4774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.924853340894492e-06,
+      "loss": 0.8076,
+      "step": 4775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.899492051276627e-06,
+      "loss": 0.7406,
+      "step": 4776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.874161518866022e-06,
+      "loss": 0.231,
+      "step": 4777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.848861752309668e-06,
+      "loss": 0.7318,
+      "step": 4778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.823592760244049e-06,
+      "loss": 0.7544,
+      "step": 4779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.798354551295075e-06,
+      "loss": 0.7196,
+      "step": 4780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.773147134078265e-06,
+      "loss": 0.7762,
+      "step": 4781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.747970517198524e-06,
+      "loss": 0.7818,
+      "step": 4782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.722824709250299e-06,
+      "loss": 0.7219,
+      "step": 4783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69770971881746e-06,
+      "loss": 0.7364,
+      "step": 4784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.672625554473457e-06,
+      "loss": 0.7894,
+      "step": 4785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.647572224781132e-06,
+      "loss": 0.7271,
+      "step": 4786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.622549738292808e-06,
+      "loss": 0.7508,
+      "step": 4787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597558103550341e-06,
+      "loss": 0.7061,
+      "step": 4788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.572597329084998e-06,
+      "loss": 0.758,
+      "step": 4789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547667423417516e-06,
+      "loss": 0.7677,
+      "step": 4790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.52276839505809e-06,
+      "loss": 0.7752,
+      "step": 4791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.497900252506431e-06,
+      "loss": 0.7822,
+      "step": 4792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.473063004251648e-06,
+      "loss": 0.7507,
+      "step": 4793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448256658772292e-06,
+      "loss": 0.7118,
+      "step": 4794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.423481224536446e-06,
+      "loss": 0.7385,
+      "step": 4795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.398736710001543e-06,
+      "loss": 0.6803,
+      "step": 4796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.37402312361455e-06,
+      "loss": 0.7038,
+      "step": 4797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.349340473811796e-06,
+      "loss": 0.7254,
+      "step": 4798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.324688769019118e-06,
+      "loss": 0.7379,
+      "step": 4799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.300068017651741e-06,
+      "loss": 0.7493,
+      "step": 4800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.275478228114342e-06,
+      "loss": 0.7447,
+      "step": 4801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.250919408801006e-06,
+      "loss": 0.2442,
+      "step": 4802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.226391568095305e-06,
+      "loss": 0.7357,
+      "step": 4803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.201894714370196e-06,
+      "loss": 0.7698,
+      "step": 4804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177428855988013e-06,
+      "loss": 0.8241,
+      "step": 4805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15299400130062e-06,
+      "loss": 0.6983,
+      "step": 4806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128590158649197e-06,
+      "loss": 0.783,
+      "step": 4807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.104217336364384e-06,
+      "loss": 0.7573,
+      "step": 4808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.079875542766203e-06,
+      "loss": 0.7407,
+      "step": 4809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.05556478616414e-06,
+      "loss": 0.7396,
+      "step": 4810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.031285074857021e-06,
+      "loss": 0.7904,
+      "step": 4811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007036417133097e-06,
+      "loss": 0.79,
+      "step": 4812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.982818821270056e-06,
+      "loss": 0.763,
+      "step": 4813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95863229553494e-06,
+      "loss": 0.2548,
+      "step": 4814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.934476848184182e-06,
+      "loss": 0.6388,
+      "step": 4815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.910352487463603e-06,
+      "loss": 0.7184,
+      "step": 4816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.886259221608462e-06,
+      "loss": 0.7572,
+      "step": 4817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.862197058843368e-06,
+      "loss": 0.7368,
+      "step": 4818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.838166007382277e-06,
+      "loss": 0.7104,
+      "step": 4819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814166075428587e-06,
+      "loss": 0.7232,
+      "step": 4820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.790197271175072e-06,
+      "loss": 0.794,
+      "step": 4821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.766259602803828e-06,
+      "loss": 0.7251,
+      "step": 4822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742353078486331e-06,
+      "loss": 0.712,
+      "step": 4823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.718477706383487e-06,
+      "loss": 0.724,
+      "step": 4824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.6946334946455e-06,
+      "loss": 0.7101,
+      "step": 4825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.670820451411965e-06,
+      "loss": 0.7564,
+      "step": 4826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.647038584811817e-06,
+      "loss": 0.7109,
+      "step": 4827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623287902963406e-06,
+      "loss": 0.7462,
+      "step": 4828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.599568413974369e-06,
+      "loss": 0.8558,
+      "step": 4829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.575880125941704e-06,
+      "loss": 0.699,
+      "step": 4830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.552223046951824e-06,
+      "loss": 0.7363,
+      "step": 4831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52859718508041e-06,
+      "loss": 0.7522,
+      "step": 4832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.505002548392526e-06,
+      "loss": 0.7034,
+      "step": 4833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.481439144942538e-06,
+      "loss": 0.8017,
+      "step": 4834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.457906982774221e-06,
+      "loss": 0.7744,
+      "step": 4835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.43440606992063e-06,
+      "loss": 0.8402,
+      "step": 4836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.410936414404147e-06,
+      "loss": 0.2389,
+      "step": 4837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.387498024236528e-06,
+      "loss": 0.6939,
+      "step": 4838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.364090907418832e-06,
+      "loss": 0.7871,
+      "step": 4839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34071507194143e-06,
+      "loss": 0.8057,
+      "step": 4840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.317370525784019e-06,
+      "loss": 0.6727,
+      "step": 4841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.294057276915635e-06,
+      "loss": 0.8002,
+      "step": 4842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.270775333294622e-06,
+      "loss": 0.7661,
+      "step": 4843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.247524702868615e-06,
+      "loss": 0.7289,
+      "step": 4844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.224305393574571e-06,
+      "loss": 0.727,
+      "step": 4845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20111741333881e-06,
+      "loss": 0.8182,
+      "step": 4846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.177960770076876e-06,
+      "loss": 0.7342,
+      "step": 4847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154835471693633e-06,
+      "loss": 0.7251,
+      "step": 4848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.131741526083303e-06,
+      "loss": 0.7735,
+      "step": 4849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.108678941129333e-06,
+      "loss": 0.7536,
+      "step": 4850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.085647724704493e-06,
+      "loss": 0.6831,
+      "step": 4851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.062647884670838e-06,
+      "loss": 0.7496,
+      "step": 4852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.039679428879754e-06,
+      "loss": 0.2226,
+      "step": 4853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.016742365171869e-06,
+      "loss": 0.7724,
+      "step": 4854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.993836701377078e-06,
+      "loss": 0.7685,
+      "step": 4855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.970962445314622e-06,
+      "loss": 0.7239,
+      "step": 4856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.948119604792959e-06,
+      "loss": 0.7046,
+      "step": 4857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92530818760987e-06,
+      "loss": 0.7756,
+      "step": 4858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.90252820155234e-06,
+      "loss": 0.6965,
+      "step": 4859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.879779654396724e-06,
+      "loss": 0.7855,
+      "step": 4860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.857062553908579e-06,
+      "loss": 0.682,
+      "step": 4861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.8343769078427e-06,
+      "loss": 0.7469,
+      "step": 4862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.811722723943238e-06,
+      "loss": 0.7768,
+      "step": 4863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.789100009943528e-06,
+      "loss": 0.8091,
+      "step": 4864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766508773566173e-06,
+      "loss": 0.7796,
+      "step": 4865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.743949022523066e-06,
+      "loss": 0.7609,
+      "step": 4866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.72142076451532e-06,
+      "loss": 0.725,
+      "step": 4867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.6989240072333e-06,
+      "loss": 0.2553,
+      "step": 4868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.676458758356609e-06,
+      "loss": 0.7813,
+      "step": 4869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.654025025554123e-06,
+      "loss": 0.8028,
+      "step": 4870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.631622816483964e-06,
+      "loss": 0.8273,
+      "step": 4871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.609252138793455e-06,
+      "loss": 0.7725,
+      "step": 4872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.586913000119167e-06,
+      "loss": 0.7642,
+      "step": 4873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564605408086933e-06,
+      "loss": 0.7938,
+      "step": 4874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5423293703117846e-06,
+      "loss": 0.7378,
+      "step": 4875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5200848943979825e-06,
+      "loss": 0.7305,
+      "step": 4876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.497871987939032e-06,
+      "loss": 0.7753,
+      "step": 4877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.475690658517664e-06,
+      "loss": 0.7697,
+      "step": 4878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.453540913705803e-06,
+      "loss": 0.7629,
+      "step": 4879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.43142276106461e-06,
+      "loss": 0.7933,
+      "step": 4880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40933620814448e-06,
+      "loss": 0.754,
+      "step": 4881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.387281262484991e-06,
+      "loss": 0.7329,
+      "step": 4882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.365257931614911e-06,
+      "loss": 0.8204,
+      "step": 4883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343266223052292e-06,
+      "loss": 0.6489,
+      "step": 4884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3213061443043296e-06,
+      "loss": 0.6841,
+      "step": 4885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.299377702867427e-06,
+      "loss": 0.735,
+      "step": 4886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2774809062271964e-06,
+      "loss": 0.6702,
+      "step": 4887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25561576185847e-06,
+      "loss": 0.7201,
+      "step": 4888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.233782277225232e-06,
+      "loss": 0.7686,
+      "step": 4889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2119804597806765e-06,
+      "loss": 0.7442,
+      "step": 4890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.190210316967228e-06,
+      "loss": 0.6918,
+      "step": 4891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168471856216441e-06,
+      "loss": 0.7542,
+      "step": 4892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146765084949081e-06,
+      "loss": 0.683,
+      "step": 4893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.125090010575064e-06,
+      "loss": 0.7587,
+      "step": 4894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103446640493538e-06,
+      "loss": 0.7953,
+      "step": 4895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081834982092817e-06,
+      "loss": 0.8003,
+      "step": 4896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060255042750363e-06,
+      "loss": 0.2517,
+      "step": 4897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.038706829832809e-06,
+      "loss": 0.7845,
+      "step": 4898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.01719035069599e-06,
+      "loss": 0.7161,
+      "step": 4899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.995705612684888e-06,
+      "loss": 0.2633,
+      "step": 4900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.974252623133637e-06,
+      "loss": 0.7614,
+      "step": 4901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.952831389365577e-06,
+      "loss": 0.2552,
+      "step": 4902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931441918693149e-06,
+      "loss": 0.254,
+      "step": 4903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.910084218418012e-06,
+      "loss": 0.742,
+      "step": 4904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8887582958308995e-06,
+      "loss": 0.7075,
+      "step": 4905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.867464158211789e-06,
+      "loss": 0.734,
+      "step": 4906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.846201812829767e-06,
+      "loss": 0.7692,
+      "step": 4907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.824971266943025e-06,
+      "loss": 0.8087,
+      "step": 4908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.803772527798979e-06,
+      "loss": 0.7362,
+      "step": 4909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.78260560263414e-06,
+      "loss": 0.7457,
+      "step": 4910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.76147049867416e-06,
+      "loss": 0.737,
+      "step": 4911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.740367223133814e-06,
+      "loss": 0.7764,
+      "step": 4912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.719295783217083e-06,
+      "loss": 0.7234,
+      "step": 4913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.698256186116991e-06,
+      "loss": 0.7787,
+      "step": 4914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.677248439015726e-06,
+      "loss": 0.8039,
+      "step": 4915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.656272549084641e-06,
+      "loss": 0.2343,
+      "step": 4916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.635328523484163e-06,
+      "loss": 0.7794,
+      "step": 4917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.614416369363841e-06,
+      "loss": 0.7184,
+      "step": 4918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593536093862396e-06,
+      "loss": 0.2563,
+      "step": 4919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5726877041076275e-06,
+      "loss": 0.7937,
+      "step": 4920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5518712072164645e-06,
+      "loss": 0.7018,
+      "step": 4921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.531086610294923e-06,
+      "loss": 0.7533,
+      "step": 4922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.510333920438138e-06,
+      "loss": 0.7887,
+      "step": 4923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489613144730389e-06,
+      "loss": 0.8169,
+      "step": 4924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.468924290245026e-06,
+      "loss": 0.7795,
+      "step": 4925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448267364044503e-06,
+      "loss": 0.7812,
+      "step": 4926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427642373180387e-06,
+      "loss": 0.7525,
+      "step": 4927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.40704932469336e-06,
+      "loss": 0.7652,
+      "step": 4928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386488225613141e-06,
+      "loss": 0.733,
+      "step": 4929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.365959082958595e-06,
+      "loss": 0.8157,
+      "step": 4930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.345461903737671e-06,
+      "loss": 0.6938,
+      "step": 4931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324996694947405e-06,
+      "loss": 0.7691,
+      "step": 4932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.304563463573876e-06,
+      "loss": 0.786,
+      "step": 4933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.284162216592338e-06,
+      "loss": 0.7235,
+      "step": 4934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.263792960967052e-06,
+      "loss": 0.7843,
+      "step": 4935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.243455703651369e-06,
+      "loss": 0.2487,
+      "step": 4936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223150451587711e-06,
+      "loss": 0.7464,
+      "step": 4937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.202877211707647e-06,
+      "loss": 0.7897,
+      "step": 4938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.182635990931718e-06,
+      "loss": 0.7578,
+      "step": 4939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.162426796169573e-06,
+      "loss": 0.8125,
+      "step": 4940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142249634319974e-06,
+      "loss": 0.7454,
+      "step": 4941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.122104512270688e-06,
+      "loss": 0.7325,
+      "step": 4942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101991436898546e-06,
+      "loss": 0.8117,
+      "step": 4943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08191041506947e-06,
+      "loss": 0.7374,
+      "step": 4944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.061861453638451e-06,
+      "loss": 0.2483,
+      "step": 4945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.041844559449506e-06,
+      "loss": 0.7979,
+      "step": 4946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.021859739335689e-06,
+      "loss": 0.7348,
+      "step": 4947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0019070001191446e-06,
+      "loss": 0.7146,
+      "step": 4948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.981986348611057e-06,
+      "loss": 0.7303,
+      "step": 4949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962097791611632e-06,
+      "loss": 0.2611,
+      "step": 4950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942241335910148e-06,
+      "loss": 0.7454,
+      "step": 4951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922416988284918e-06,
+      "loss": 0.7663,
+      "step": 4952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.902624755503283e-06,
+      "loss": 0.7262,
+      "step": 4953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.88286464432164e-06,
+      "loss": 0.7144,
+      "step": 4954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86313666148538e-06,
+      "loss": 0.748,
+      "step": 4955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.843440813728995e-06,
+      "loss": 0.7397,
+      "step": 4956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.823777107775963e-06,
+      "loss": 0.7129,
+      "step": 4957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.804145550338758e-06,
+      "loss": 0.7842,
+      "step": 4958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.784546148118953e-06,
+      "loss": 0.7822,
+      "step": 4959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.764978907807106e-06,
+      "loss": 0.7804,
+      "step": 4960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.745443836082787e-06,
+      "loss": 0.7336,
+      "step": 4961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.725940939614583e-06,
+      "loss": 0.7468,
+      "step": 4962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7064702250601455e-06,
+      "loss": 0.7513,
+      "step": 4963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.687031699066081e-06,
+      "loss": 0.7199,
+      "step": 4964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667625368268026e-06,
+      "loss": 0.7599,
+      "step": 4965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648251239290669e-06,
+      "loss": 0.7833,
+      "step": 4966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6289093187476325e-06,
+      "loss": 0.7169,
+      "step": 4967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609599613241579e-06,
+      "loss": 0.2655,
+      "step": 4968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5903221293642e-06,
+      "loss": 0.7306,
+      "step": 4969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.571076873696168e-06,
+      "loss": 0.7296,
+      "step": 4970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551863852807127e-06,
+      "loss": 0.7449,
+      "step": 4971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.532683073255762e-06,
+      "loss": 0.7805,
+      "step": 4972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.513534541589693e-06,
+      "loss": 0.6828,
+      "step": 4973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.494418264345614e-06,
+      "loss": 0.7598,
+      "step": 4974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.475334248049146e-06,
+      "loss": 0.7531,
+      "step": 4975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.456282499214893e-06,
+      "loss": 0.7869,
+      "step": 4976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437263024346495e-06,
+      "loss": 0.7611,
+      "step": 4977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.418275829936537e-06,
+      "loss": 0.7429,
+      "step": 4978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3993209224665885e-06,
+      "loss": 0.7435,
+      "step": 4979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.380398308407186e-06,
+      "loss": 0.744,
+      "step": 4980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.361507994217885e-06,
+      "loss": 0.7945,
+      "step": 4981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.342649986347181e-06,
+      "loss": 0.7527,
+      "step": 4982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3238242912325265e-06,
+      "loss": 0.7509,
+      "step": 4983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3050309153003905e-06,
+      "loss": 0.8044,
+      "step": 4984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286269864966175e-06,
+      "loss": 0.7597,
+      "step": 4985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.267541146634258e-06,
+      "loss": 0.7243,
+      "step": 4986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.248844766697969e-06,
+      "loss": 0.7705,
+      "step": 4987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230180731539613e-06,
+      "loss": 0.7265,
+      "step": 4988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211549047530462e-06,
+      "loss": 0.257,
+      "step": 4989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.192949721030693e-06,
+      "loss": 0.7016,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.174382758389518e-06,
+      "loss": 0.2446,
+      "step": 4991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.155848165945021e-06,
+      "loss": 0.7965,
+      "step": 4992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137345950024308e-06,
+      "loss": 0.7532,
+      "step": 4993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1188761169433715e-06,
+      "loss": 0.8038,
+      "step": 4994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100438673007191e-06,
+      "loss": 0.2608,
+      "step": 4995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.082033624509675e-06,
+      "loss": 0.7592,
+      "step": 4996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.063660977733664e-06,
+      "loss": 0.7284,
+      "step": 4997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045320738950943e-06,
+      "loss": 0.7104,
+      "step": 4998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027012914422246e-06,
+      "loss": 0.8125,
+      "step": 4999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.008737510397232e-06,
+      "loss": 0.7088,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990494533114476e-06,
+      "loss": 0.7181,
+      "step": 5001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972283988801529e-06,
+      "loss": 0.746,
+      "step": 5002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954105883674831e-06,
+      "loss": 0.6822,
+      "step": 5003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935960223939762e-06,
+      "loss": 0.7652,
+      "step": 5004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917847015790611e-06,
+      "loss": 0.748,
+      "step": 5005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899766265410632e-06,
+      "loss": 0.7694,
+      "step": 5006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.881717978971956e-06,
+      "loss": 0.671,
+      "step": 5007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.863702162635619e-06,
+      "loss": 0.8051,
+      "step": 5008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.845718822551637e-06,
+      "loss": 0.7431,
+      "step": 5009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.827767964858898e-06,
+      "loss": 0.7427,
+      "step": 5010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.80984959568519e-06,
+      "loss": 0.6987,
+      "step": 5011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7919637211472194e-06,
+      "loss": 0.8687,
+      "step": 5012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.774110347350635e-06,
+      "loss": 0.7824,
+      "step": 5013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7562894803899524e-06,
+      "loss": 0.7673,
+      "step": 5014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7385011263485825e-06,
+      "loss": 0.6631,
+      "step": 5015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.720745291298878e-06,
+      "loss": 0.6798,
+      "step": 5016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7030219813020584e-06,
+      "loss": 0.6953,
+      "step": 5017
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.685331202408261e-06,
+      "loss": 0.7733,
+      "step": 5018
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.667672960656488e-06,
+      "loss": 0.7244,
+      "step": 5019
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.650047262074686e-06,
+      "loss": 0.7789,
+      "step": 5020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.632454112679651e-06,
+      "loss": 0.7741,
+      "step": 5021
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.614893518477059e-06,
+      "loss": 0.7714,
+      "step": 5022
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597365485461491e-06,
+      "loss": 0.787,
+      "step": 5023
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.579870019616428e-06,
+      "loss": 0.7651,
+      "step": 5024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.562407126914225e-06,
+      "loss": 0.7371,
+      "step": 5025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.544976813316093e-06,
+      "loss": 0.7494,
+      "step": 5026
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.527579084772149e-06,
+      "loss": 0.7089,
+      "step": 5027
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.510213947221387e-06,
+      "loss": 0.7308,
+      "step": 5028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.492881406591654e-06,
+      "loss": 0.7338,
+      "step": 5029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.475581468799672e-06,
+      "loss": 0.7469,
+      "step": 5030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.458314139751074e-06,
+      "loss": 0.8229,
+      "step": 5031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.441079425340311e-06,
+      "loss": 0.7912,
+      "step": 5032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.423877331450699e-06,
+      "loss": 0.7195,
+      "step": 5033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.406707863954485e-06,
+      "loss": 0.7266,
+      "step": 5034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389571028712714e-06,
+      "loss": 0.7388,
+      "step": 5035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.372466831575283e-06,
+      "loss": 0.7518,
+      "step": 5036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.355395278381025e-06,
+      "loss": 0.8096,
+      "step": 5037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.338356374957553e-06,
+      "loss": 0.7883,
+      "step": 5038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.321350127121359e-06,
+      "loss": 0.7653,
+      "step": 5039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3043765406777905e-06,
+      "loss": 0.7477,
+      "step": 5040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287435621421032e-06,
+      "loss": 0.7308,
+      "step": 5041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.27052737513417e-06,
+      "loss": 0.796,
+      "step": 5042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2536518075890765e-06,
+      "loss": 0.7457,
+      "step": 5043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.236808924546465e-06,
+      "loss": 0.7478,
+      "step": 5044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.219998731755947e-06,
+      "loss": 0.2654,
+      "step": 5045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203221234955934e-06,
+      "loss": 0.7511,
+      "step": 5046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.186476439873677e-06,
+      "loss": 0.7655,
+      "step": 5047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.169764352225269e-06,
+      "loss": 0.7575,
+      "step": 5048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.15308497771566e-06,
+      "loss": 0.7338,
+      "step": 5049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.136438322038594e-06,
+      "loss": 0.7538,
+      "step": 5050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1198243908766695e-06,
+      "loss": 0.7821,
+      "step": 5051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103243189901318e-06,
+      "loss": 0.7425,
+      "step": 5052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.086694724772788e-06,
+      "loss": 0.7274,
+      "step": 5053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.070179001140151e-06,
+      "loss": 0.7318,
+      "step": 5054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.053696024641296e-06,
+      "loss": 0.7383,
+      "step": 5055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.037245800902967e-06,
+      "loss": 0.7204,
+      "step": 5056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.020828335540694e-06,
+      "loss": 0.7173,
+      "step": 5057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.004443634158806e-06,
+      "loss": 0.7319,
+      "step": 5058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.988091702350516e-06,
+      "loss": 0.7163,
+      "step": 5059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.971772545697794e-06,
+      "loss": 0.7653,
+      "step": 5060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.955486169771427e-06,
+      "loss": 0.6893,
+      "step": 5061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9392325801310355e-06,
+      "loss": 0.7518,
+      "step": 5062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.923011782325036e-06,
+      "loss": 0.7767,
+      "step": 5063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.906823781890645e-06,
+      "loss": 0.7121,
+      "step": 5064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.890668584353874e-06,
+      "loss": 0.7666,
+      "step": 5065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.874546195229556e-06,
+      "loss": 0.7462,
+      "step": 5066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8584566200213446e-06,
+      "loss": 0.7697,
+      "step": 5067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.842399864221647e-06,
+      "loss": 0.7829,
+      "step": 5068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.826375933311677e-06,
+      "loss": 0.7347,
+      "step": 5069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.810384832761471e-06,
+      "loss": 0.8227,
+      "step": 5070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.794426568029841e-06,
+      "loss": 0.7275,
+      "step": 5071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7785011445643614e-06,
+      "loss": 0.7201,
+      "step": 5072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.76260856780144e-06,
+      "loss": 0.7774,
+      "step": 5073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.746748843166259e-06,
+      "loss": 0.7939,
+      "step": 5074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7309219760727765e-06,
+      "loss": 0.7864,
+      "step": 5075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7151279719237264e-06,
+      "loss": 0.7082,
+      "step": 5076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.699366836110674e-06,
+      "loss": 0.7406,
+      "step": 5077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6836385740138923e-06,
+      "loss": 0.7354,
+      "step": 5078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6679431910024654e-06,
+      "loss": 0.7268,
+      "step": 5079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6522806924342954e-06,
+      "loss": 0.8082,
+      "step": 5080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.636651083655984e-06,
+      "loss": 0.6807,
+      "step": 5081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.621054370002952e-06,
+      "loss": 0.8182,
+      "step": 5082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6054905567993624e-06,
+      "loss": 0.2432,
+      "step": 5083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.589959649358199e-06,
+      "loss": 0.6683,
+      "step": 5084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5744616529811535e-06,
+      "loss": 0.7439,
+      "step": 5085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.558996572958706e-06,
+      "loss": 0.7423,
+      "step": 5086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.543564414570122e-06,
+      "loss": 0.2768,
+      "step": 5087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5281651830833984e-06,
+      "loss": 0.7278,
+      "step": 5088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.512798883755297e-06,
+      "loss": 0.8064,
+      "step": 5089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4974655218313333e-06,
+      "loss": 0.2403,
+      "step": 5090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482165102545798e-06,
+      "loss": 0.7341,
+      "step": 5091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466897631121746e-06,
+      "loss": 0.7594,
+      "step": 5092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451663112770942e-06,
+      "loss": 0.6946,
+      "step": 5093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4364615526939257e-06,
+      "loss": 0.7566,
+      "step": 5094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4212929560800023e-06,
+      "loss": 0.7664,
+      "step": 5095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4061573281071957e-06,
+      "loss": 0.7809,
+      "step": 5096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.391054673942273e-06,
+      "loss": 0.7356,
+      "step": 5097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3759849987407997e-06,
+      "loss": 0.7617,
+      "step": 5098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.360948307647016e-06,
+      "loss": 0.778,
+      "step": 5099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.345944605793927e-06,
+      "loss": 0.7643,
+      "step": 5100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.33097389830328e-06,
+      "loss": 0.6608,
+      "step": 5101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.316036190285576e-06,
+      "loss": 0.7132,
+      "step": 5102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3011314868400145e-06,
+      "loss": 0.7346,
+      "step": 5103
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2862597930545467e-06,
+      "loss": 0.7361,
+      "step": 5104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271421114005879e-06,
+      "loss": 0.2258,
+      "step": 5105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2566154547594153e-06,
+      "loss": 0.7225,
+      "step": 5106
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2418428203692788e-06,
+      "loss": 0.743,
+      "step": 5107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2271032158783464e-06,
+      "loss": 0.735,
+      "step": 5108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2123966463182276e-06,
+      "loss": 0.729,
+      "step": 5109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.197723116709228e-06,
+      "loss": 0.6851,
+      "step": 5110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1830826320603635e-06,
+      "loss": 0.7514,
+      "step": 5111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168475197369436e-06,
+      "loss": 0.7547,
+      "step": 5112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.153900817622879e-06,
+      "loss": 0.7372,
+      "step": 5113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1393594977959017e-06,
+      "loss": 0.7395,
+      "step": 5114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.124851242852389e-06,
+      "loss": 0.758,
+      "step": 5115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1103760577450012e-06,
+      "loss": 0.7059,
+      "step": 5116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0959339474150308e-06,
+      "loss": 0.7473,
+      "step": 5117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.081524916792544e-06,
+      "loss": 0.7138,
+      "step": 5118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.067148970796252e-06,
+      "loss": 0.7803,
+      "step": 5119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0528061143336282e-06,
+      "loss": 0.252,
+      "step": 5120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.038496352300835e-06,
+      "loss": 0.7655,
+      "step": 5121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0242196895827192e-06,
+      "loss": 0.754,
+      "step": 5122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0099761310528497e-06,
+      "loss": 0.7511,
+      "step": 5123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9957656815734927e-06,
+      "loss": 0.7624,
+      "step": 5124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9815883459955898e-06,
+      "loss": 0.7675,
+      "step": 5125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9674441291587806e-06,
+      "loss": 0.7757,
+      "step": 5126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9533330358914592e-06,
+      "loss": 0.743,
+      "step": 5127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.939255071010638e-06,
+      "loss": 0.7478,
+      "step": 5128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.925210239322029e-06,
+      "loss": 0.7339,
+      "step": 5129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9111985456200975e-06,
+      "loss": 0.7719,
+      "step": 5130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8972199946879384e-06,
+      "loss": 0.7602,
+      "step": 5131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8832745912973245e-06,
+      "loss": 0.7253,
+      "step": 5132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8693623402087476e-06,
+      "loss": 0.7992,
+      "step": 5133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8554832461713865e-06,
+      "loss": 0.7778,
+      "step": 5134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8416373139230734e-06,
+      "loss": 0.7659,
+      "step": 5135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8278245481903164e-06,
+      "loss": 0.7176,
+      "step": 5136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.814044953688344e-06,
+      "loss": 0.7683,
+      "step": 5137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8002985351210266e-06,
+      "loss": 0.7094,
+      "step": 5138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.78658529718091e-06,
+      "loss": 0.733,
+      "step": 5139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7729052445492176e-06,
+      "loss": 0.7172,
+      "step": 5140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7592583818958683e-06,
+      "loss": 0.7657,
+      "step": 5141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.745644713879414e-06,
+      "loss": 0.6847,
+      "step": 5142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7320642451471036e-06,
+      "loss": 0.7353,
+      "step": 5143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.718516980334818e-06,
+      "loss": 0.7419,
+      "step": 5144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705002924067157e-06,
+      "loss": 0.7897,
+      "step": 5145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6915220809573315e-06,
+      "loss": 0.7408,
+      "step": 5146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.678074455607238e-06,
+      "loss": 0.7796,
+      "step": 5147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6646600526074484e-06,
+      "loss": 0.7955,
+      "step": 5148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6512788765371666e-06,
+      "loss": 0.8054,
+      "step": 5149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.637930931964261e-06,
+      "loss": 0.7754,
+      "step": 5150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.624616223445253e-06,
+      "loss": 0.7101,
+      "step": 5151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6113347555253407e-06,
+      "loss": 0.6928,
+      "step": 5152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.598086532738364e-06,
+      "loss": 0.7273,
+      "step": 5153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.584871559606783e-06,
+      "loss": 0.7879,
+      "step": 5154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5716898406417667e-06,
+      "loss": 0.7687,
+      "step": 5155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5585413803430826e-06,
+      "loss": 0.7218,
+      "step": 5156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.545426183199151e-06,
+      "loss": 0.7024,
+      "step": 5157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532344253687069e-06,
+      "loss": 0.712,
+      "step": 5158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5192955962725416e-06,
+      "loss": 0.7559,
+      "step": 5159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5062802154099398e-06,
+      "loss": 0.8012,
+      "step": 5160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.493298115542253e-06,
+      "loss": 0.7533,
+      "step": 5161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480349301101137e-06,
+      "loss": 0.7845,
+      "step": 5162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4674337765068776e-06,
+      "loss": 0.8495,
+      "step": 5163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4545515461683486e-06,
+      "loss": 0.7979,
+      "step": 5164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4417026144831436e-06,
+      "loss": 0.7311,
+      "step": 5165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.428886985837431e-06,
+      "loss": 0.8378,
+      "step": 5166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.416104664606034e-06,
+      "loss": 0.764,
+      "step": 5167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4033556551523727e-06,
+      "loss": 0.2727,
+      "step": 5168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3906399618285224e-06,
+      "loss": 0.7176,
+      "step": 5169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3779575889752105e-06,
+      "loss": 0.7566,
+      "step": 5170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.36530854092174e-06,
+      "loss": 0.7952,
+      "step": 5171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3526928219860578e-06,
+      "loss": 0.7515,
+      "step": 5172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.340110436474741e-06,
+      "loss": 0.7844,
+      "step": 5173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.327561388682986e-06,
+      "loss": 0.7549,
+      "step": 5174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.315045682894601e-06,
+      "loss": 0.693,
+      "step": 5175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.302563323382012e-06,
+      "loss": 0.7332,
+      "step": 5176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2901143144062776e-06,
+      "loss": 0.2415,
+      "step": 5177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2776986602170535e-06,
+      "loss": 0.7486,
+      "step": 5178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2653163650526055e-06,
+      "loss": 0.7742,
+      "step": 5179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.252967433139841e-06,
+      "loss": 0.8186,
+      "step": 5180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2406518686942548e-06,
+      "loss": 0.7694,
+      "step": 5181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228369675919961e-06,
+      "loss": 0.738,
+      "step": 5182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2161208590096403e-06,
+      "loss": 0.719,
+      "step": 5183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2039054221446696e-06,
+      "loss": 0.6993,
+      "step": 5184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.191723369494947e-06,
+      "loss": 0.7635,
+      "step": 5185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1795747052190007e-06,
+      "loss": 0.7821,
+      "step": 5186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167459433463992e-06,
+      "loss": 0.756,
+      "step": 5187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1553775583656343e-06,
+      "loss": 0.783,
+      "step": 5188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.143329084048273e-06,
+      "loss": 0.7841,
+      "step": 5189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.13131401462483e-06,
+      "loss": 0.8213,
+      "step": 5190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.119332354196857e-06,
+      "loss": 0.7646,
+      "step": 5191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.107384106854471e-06,
+      "loss": 0.7418,
+      "step": 5192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.095469276676387e-06,
+      "loss": 0.771,
+      "step": 5193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.083587867729908e-06,
+      "loss": 0.7609,
+      "step": 5194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071739884070956e-06,
+      "loss": 0.7522,
+      "step": 5195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.059925329744017e-06,
+      "loss": 0.7683,
+      "step": 5196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0481442087821657e-06,
+      "loss": 0.7698,
+      "step": 5197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.036396525207085e-06,
+      "loss": 0.7123,
+      "step": 5198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0246822830290224e-06,
+      "loss": 0.7532,
+      "step": 5199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.013001486246824e-06,
+      "loss": 0.7864,
+      "step": 5200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0013541388478773e-06,
+      "loss": 0.7448,
+      "step": 5201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.989740244808225e-06,
+      "loss": 0.7799,
+      "step": 5202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.978159808092439e-06,
+      "loss": 0.2779,
+      "step": 5203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.966612832653669e-06,
+      "loss": 0.7663,
+      "step": 5204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9550993224336733e-06,
+      "loss": 0.7149,
+      "step": 5205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.943619281362752e-06,
+      "loss": 0.7455,
+      "step": 5206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9321727133598034e-06,
+      "loss": 0.7428,
+      "step": 5207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9207596223322798e-06,
+      "loss": 0.7368,
+      "step": 5208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9093800121762318e-06,
+      "loss": 0.7299,
+      "step": 5209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8980338867762515e-06,
+      "loss": 0.7873,
+      "step": 5210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8867212500055297e-06,
+      "loss": 0.7612,
+      "step": 5211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8754421057258e-06,
+      "loss": 0.6793,
+      "step": 5212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8641964577873815e-06,
+      "loss": 0.7333,
+      "step": 5213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8529843100291489e-06,
+      "loss": 0.8023,
+      "step": 5214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8418056662785289e-06,
+      "loss": 0.7141,
+      "step": 5215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8306605303515467e-06,
+      "loss": 0.7157,
+      "step": 5216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8195489060527592e-06,
+      "loss": 0.7764,
+      "step": 5217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.808470797175288e-06,
+      "loss": 0.7734,
+      "step": 5218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7974262075008186e-06,
+      "loss": 0.7877,
+      "step": 5219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.786415140799591e-06,
+      "loss": 0.6868,
+      "step": 5220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7754376008304207e-06,
+      "loss": 0.7727,
+      "step": 5221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7644935913406323e-06,
+      "loss": 0.7507,
+      "step": 5222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7535831160661598e-06,
+      "loss": 0.7381,
+      "step": 5223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7427061787314458e-06,
+      "loss": 0.6931,
+      "step": 5224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7318627830495094e-06,
+      "loss": 0.6827,
+      "step": 5225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7210529327219116e-06,
+      "loss": 0.7547,
+      "step": 5226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7102766314387675e-06,
+      "loss": 0.7288,
+      "step": 5227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6995338828787233e-06,
+      "loss": 0.7281,
+      "step": 5228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6888246907089899e-06,
+      "loss": 0.8075,
+      "step": 5229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6781490585853099e-06,
+      "loss": 0.7577,
+      "step": 5230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6675069901519902e-06,
+      "loss": 0.8052,
+      "step": 5231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6568984890418581e-06,
+      "loss": 0.7774,
+      "step": 5232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6463235588762837e-06,
+      "loss": 0.7695,
+      "step": 5233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6357822032652016e-06,
+      "loss": 0.2568,
+      "step": 5234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.625274425807044e-06,
+      "loss": 0.7764,
+      "step": 5235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6148002300888088e-06,
+      "loss": 0.7483,
+      "step": 5236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6043596196860467e-06,
+      "loss": 0.8091,
+      "step": 5237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5939525981627956e-06,
+      "loss": 0.8458,
+      "step": 5238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5835791690716695e-06,
+      "loss": 0.7438,
+      "step": 5239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5732393359537912e-06,
+      "loss": 0.7662,
+      "step": 5240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5629331023388372e-06,
+      "loss": 0.7635,
+      "step": 5241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5526604717449933e-06,
+      "loss": 0.7181,
+      "step": 5242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5424214476789768e-06,
+      "loss": 0.7616,
+      "step": 5243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5322160336360247e-06,
+      "loss": 0.725,
+      "step": 5244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5220442330999396e-06,
+      "loss": 0.8173,
+      "step": 5245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5119060495430105e-06,
+      "loss": 0.7799,
+      "step": 5246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5018014864260689e-06,
+      "loss": 0.7506,
+      "step": 5247
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4917305471984556e-06,
+      "loss": 0.7607,
+      "step": 5248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.481693235298054e-06,
+      "loss": 0.7594,
+      "step": 5249
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4716895541512454e-06,
+      "loss": 0.7587,
+      "step": 5250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4617195071729428e-06,
+      "loss": 0.7562,
+      "step": 5251
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4517830977665792e-06,
+      "loss": 0.7304,
+      "step": 5252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4418803293241079e-06,
+      "loss": 0.7295,
+      "step": 5253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4320112052259804e-06,
+      "loss": 0.6889,
+      "step": 5254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4221757288411907e-06,
+      "loss": 0.7739,
+      "step": 5255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4123739035272199e-06,
+      "loss": 0.7833,
+      "step": 5256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40260573263008e-06,
+      "loss": 0.745,
+      "step": 5257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.392871219484282e-06,
+      "loss": 0.7526,
+      "step": 5258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.383170367412856e-06,
+      "loss": 0.7371,
+      "step": 5259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3735031797273424e-06,
+      "loss": 0.767,
+      "step": 5260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638696597277679e-06,
+      "loss": 0.7536,
+      "step": 5261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3542698107027018e-06,
+      "loss": 0.7227,
+      "step": 5262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3447036359292008e-06,
+      "loss": 0.6966,
+      "step": 5263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3351711386728193e-06,
+      "loss": 0.7174,
+      "step": 5264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3256723221876322e-06,
+      "loss": 0.7287,
+      "step": 5265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.316207189716201e-06,
+      "loss": 0.7204,
+      "step": 5266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3067757444896077e-06,
+      "loss": 0.7671,
+      "step": 5267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2973779897274107e-06,
+      "loss": 0.7388,
+      "step": 5268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2880139286376658e-06,
+      "loss": 0.7297,
+      "step": 5269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2786835644169715e-06,
+      "loss": 0.7403,
+      "step": 5270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.26938690025038e-06,
+      "loss": 0.2618,
+      "step": 5271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2601239393114306e-06,
+      "loss": 0.6376,
+      "step": 5272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.250894684762216e-06,
+      "loss": 0.7839,
+      "step": 5273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2416991397532607e-06,
+      "loss": 0.7327,
+      "step": 5274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.232537307423609e-06,
+      "loss": 0.7241,
+      "step": 5275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2234091909008038e-06,
+      "loss": 0.6933,
+      "step": 5276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2143147933008747e-06,
+      "loss": 0.8457,
+      "step": 5277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.205254117728316e-06,
+      "loss": 0.692,
+      "step": 5278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1962271672761315e-06,
+      "loss": 0.7413,
+      "step": 5279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1872339450258341e-06,
+      "loss": 0.7799,
+      "step": 5280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1782744540473679e-06,
+      "loss": 0.6889,
+      "step": 5281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1693486973992197e-06,
+      "loss": 0.8703,
+      "step": 5282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1604566781283299e-06,
+      "loss": 0.7375,
+      "step": 5283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1515983992701151e-06,
+      "loss": 0.7446,
+      "step": 5284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.142773863848512e-06,
+      "loss": 0.7001,
+      "step": 5285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1339830748758773e-06,
+      "loss": 0.7746,
+      "step": 5286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1252260353530996e-06,
+      "loss": 0.7447,
+      "step": 5287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1165027482695433e-06,
+      "loss": 0.255,
+      "step": 5288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1078132166030264e-06,
+      "loss": 0.2689,
+      "step": 5289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0991574433198427e-06,
+      "loss": 0.7473,
+      "step": 5290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0905354313748062e-06,
+      "loss": 0.7345,
+      "step": 5291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0819471837111516e-06,
+      "loss": 0.7538,
+      "step": 5292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0733927032606118e-06,
+      "loss": 0.2526,
+      "step": 5293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0648719929433948e-06,
+      "loss": 0.7665,
+      "step": 5294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0563850556681742e-06,
+      "loss": 0.2295,
+      "step": 5295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0479318943321104e-06,
+      "loss": 0.7692,
+      "step": 5296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.039512511820806e-06,
+      "loss": 0.2488,
+      "step": 5297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0311269110083622e-06,
+      "loss": 0.7438,
+      "step": 5298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0227750947573223e-06,
+      "loss": 0.7506,
+      "step": 5299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.014457065918706e-06,
+      "loss": 0.7413,
+      "step": 5300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0061728273320081e-06,
+      "loss": 0.7328,
+      "step": 5301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.979223818251892e-07,
+      "loss": 0.2645,
+      "step": 5302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.89705732214652e-07,
+      "loss": 0.7522,
+      "step": 5303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.815228813052856e-07,
+      "loss": 0.76,
+      "step": 5304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.73373831890434e-07,
+      "loss": 0.7194,
+      "step": 5305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.652585867519048e-07,
+      "loss": 0.7843,
+      "step": 5306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.57177148659949e-07,
+      "loss": 0.7448,
+      "step": 5307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.491295203733152e-07,
+      "loss": 0.7489,
+      "step": 5308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.411157046391616e-07,
+      "loss": 0.7468,
+      "step": 5309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.331357041931443e-07,
+      "loss": 0.74,
+      "step": 5310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.251895217593287e-07,
+      "loss": 0.7541,
+      "step": 5311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.172771600503227e-07,
+      "loss": 0.7651,
+      "step": 5312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.093986217670991e-07,
+      "loss": 0.7,
+      "step": 5313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.015539095991177e-07,
+      "loss": 0.7272,
+      "step": 5314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.937430262243029e-07,
+      "loss": 0.795,
+      "step": 5315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.85965974309011e-07,
+      "loss": 0.7275,
+      "step": 5316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.782227565080736e-07,
+      "loss": 0.7231,
+      "step": 5317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.705133754647433e-07,
+      "loss": 0.2369,
+      "step": 5318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.62837833810759e-07,
+      "loss": 0.8197,
+      "step": 5319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.551961341662584e-07,
+      "loss": 0.2413,
+      "step": 5320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.47588279139877e-07,
+      "loss": 0.8198,
+      "step": 5321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.400142713286596e-07,
+      "loss": 0.7158,
+      "step": 5322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.324741133181158e-07,
+      "loss": 0.7321,
+      "step": 5323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.249678076821976e-07,
+      "loss": 0.7436,
+      "step": 5324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.174953569832999e-07,
+      "loss": 0.7631,
+      "step": 5325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.100567637722489e-07,
+      "loss": 0.7677,
+      "step": 5326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.026520305883467e-07,
+      "loss": 0.7225,
+      "step": 5327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.952811599592935e-07,
+      "loss": 0.7828,
+      "step": 5328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.879441544012544e-07,
+      "loss": 0.7346,
+      "step": 5329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.80641016418826e-07,
+      "loss": 0.788,
+      "step": 5330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.733717485050695e-07,
+      "loss": 0.7428,
+      "step": 5331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.661363531414334e-07,
+      "loss": 0.7396,
+      "step": 5332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.589348327978529e-07,
+      "loss": 0.7458,
+      "step": 5333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.517671899326728e-07,
+      "loss": 0.8117,
+      "step": 5334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.446334269926802e-07,
+      "loss": 0.7605,
+      "step": 5335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.375335464130828e-07,
+      "loss": 0.7831,
+      "step": 5336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.304675506175529e-07,
+      "loss": 0.7767,
+      "step": 5337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.234354420181722e-07,
+      "loss": 0.72,
+      "step": 5338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.164372230154537e-07,
+      "loss": 0.2368,
+      "step": 5339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.094728959983532e-07,
+      "loss": 0.7855,
+      "step": 5340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.02542463344269e-07,
+      "loss": 0.711,
+      "step": 5341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.956459274189753e-07,
+      "loss": 0.7383,
+      "step": 5342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.887832905767444e-07,
+      "loss": 0.7736,
+      "step": 5343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.819545551602246e-07,
+      "loss": 0.814,
+      "step": 5344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.75159723500518e-07,
+      "loss": 0.7587,
+      "step": 5345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.683987979171468e-07,
+      "loss": 0.7371,
+      "step": 5346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.616717807180539e-07,
+      "loss": 0.7707,
+      "step": 5347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.549786741996134e-07,
+      "loss": 0.7074,
+      "step": 5348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.483194806466308e-07,
+      "loss": 0.7649,
+      "step": 5349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.416942023322991e-07,
+      "loss": 0.7464,
+      "step": 5350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.35102841518287e-07,
+      "loss": 0.7414,
+      "step": 5351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.285454004546498e-07,
+      "loss": 0.7365,
+      "step": 5352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.22021881379864e-07,
+      "loss": 0.7449,
+      "step": 5353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.155322865208368e-07,
+      "loss": 0.7457,
+      "step": 5354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.090766180928964e-07,
+      "loss": 0.7731,
+      "step": 5355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.02654878299791e-07,
+      "loss": 0.7412,
+      "step": 5356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.962670693336669e-07,
+      "loss": 0.7671,
+      "step": 5357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.899131933751024e-07,
+      "loss": 0.7084,
+      "step": 5358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.835932525931064e-07,
+      "loss": 0.6917,
+      "step": 5359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.77307249145076e-07,
+      "loss": 0.2341,
+      "step": 5360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.710551851768498e-07,
+      "loss": 0.7574,
+      "step": 5361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.64837062822654e-07,
+      "loss": 0.7589,
+      "step": 5362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.586528842051464e-07,
+      "loss": 0.7973,
+      "step": 5363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.525026514353826e-07,
+      "loss": 0.7355,
+      "step": 5364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.463863666128499e-07,
+      "loss": 0.246,
+      "step": 5365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.403040318254449e-07,
+      "loss": 0.712,
+      "step": 5366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.34255649149451e-07,
+      "loss": 0.7692,
+      "step": 5367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.282412206495834e-07,
+      "loss": 0.705,
+      "step": 5368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.222607483789665e-07,
+      "loss": 0.7452,
+      "step": 5369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.163142343791227e-07,
+      "loss": 0.2657,
+      "step": 5370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.104016806799838e-07,
+      "loss": 0.7373,
+      "step": 5371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.045230892998909e-07,
+      "loss": 0.8045,
+      "step": 5372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.986784622455942e-07,
+      "loss": 0.724,
+      "step": 5373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.928678015122534e-07,
+      "loss": 0.7077,
+      "step": 5374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.870911090834263e-07,
+      "loss": 0.7421,
+      "step": 5375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.813483869310797e-07,
+      "loss": 0.7071,
+      "step": 5376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7563963701556804e-07,
+      "loss": 0.8019,
+      "step": 5377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.699648612856655e-07,
+      "loss": 0.7387,
+      "step": 5378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.64324061678556e-07,
+      "loss": 0.7343,
+      "step": 5379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5871724011981034e-07,
+      "loss": 0.2302,
+      "step": 5380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5314439852338633e-07,
+      "loss": 0.6862,
+      "step": 5381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.476055387916844e-07,
+      "loss": 0.7318,
+      "step": 5382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4210066281547e-07,
+      "loss": 0.7918,
+      "step": 5383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3662977247391767e-07,
+      "loss": 0.7102,
+      "step": 5384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3119286963460017e-07,
+      "loss": 0.7709,
+      "step": 5385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2578995615348835e-07,
+      "loss": 0.7267,
+      "step": 5386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2042103387495146e-07,
+      "loss": 0.7157,
+      "step": 5387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1508610463175666e-07,
+      "loss": 0.7254,
+      "step": 5388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0978517024506947e-07,
+      "loss": 0.7816,
+      "step": 5389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0451823252442014e-07,
+      "loss": 0.7527,
+      "step": 5390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.992852932678037e-07,
+      "loss": 0.7379,
+      "step": 5391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.940863542615247e-07,
+      "loss": 0.2665,
+      "step": 5392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8892141728034125e-07,
+      "loss": 0.7356,
+      "step": 5393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8379048408738736e-07,
+      "loss": 0.7415,
+      "step": 5394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7869355643417315e-07,
+      "loss": 0.7419,
+      "step": 5395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.73630636060629e-07,
+      "loss": 0.7424,
+      "step": 5396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6860172469505016e-07,
+      "loss": 0.7502,
+      "step": 5397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6360682405414127e-07,
+      "loss": 0.7687,
+      "step": 5398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.586459358429828e-07,
+      "loss": 0.7668,
+      "step": 5399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5371906175505343e-07,
+      "loss": 0.7622,
+      "step": 5400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4882620347221893e-07,
+      "loss": 0.7594,
+      "step": 5401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.439673626647322e-07,
+      "loss": 0.743,
+      "step": 5402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39142540991233e-07,
+      "loss": 0.7227,
+      "step": 5403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3435174009873724e-07,
+      "loss": 0.7718,
+      "step": 5404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.295949616226923e-07,
+      "loss": 0.7587,
+      "step": 5405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2487220718686593e-07,
+      "loss": 0.7377,
+      "step": 5406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.201834784034574e-07,
+      "loss": 0.7901,
+      "step": 5407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1552877687303085e-07,
+      "loss": 0.7816,
+      "step": 5408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.109081041845485e-07,
+      "loss": 0.2727,
+      "step": 5409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.063214619153487e-07,
+      "loss": 0.7747,
+      "step": 5410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0176885163114563e-07,
+      "loss": 0.7551,
+      "step": 5411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.972502748860517e-07,
+      "loss": 0.7564,
+      "step": 5412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.927657332225553e-07,
+      "loss": 0.8003,
+      "step": 5413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.883152281715096e-07,
+      "loss": 0.7962,
+      "step": 5414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8389876125216594e-07,
+      "loss": 0.7671,
+      "step": 5415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.79516333972174e-07,
+      "loss": 0.7386,
+      "step": 5416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7516794782752595e-07,
+      "loss": 0.7865,
+      "step": 5417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.708536043026011e-07,
+      "loss": 0.7966,
+      "step": 5418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6657330487019904e-07,
+      "loss": 0.7756,
+      "step": 5419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6232705099143993e-07,
+      "loss": 0.7692,
+      "step": 5420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5811484411585317e-07,
+      "loss": 0.7594,
+      "step": 5421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.539366856813552e-07,
+      "loss": 0.7346,
+      "step": 5422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4979257711420513e-07,
+      "loss": 0.7572,
+      "step": 5423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.456825198290713e-07,
+      "loss": 0.8036,
+      "step": 5424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4160651522897594e-07,
+      "loss": 0.7827,
+      "step": 5425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3756456470535038e-07,
+      "loss": 0.7136,
+      "step": 5426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3355666963794654e-07,
+      "loss": 0.744,
+      "step": 5427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2958283139493665e-07,
+      "loss": 0.7478,
+      "step": 5428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.256430513328578e-07,
+      "loss": 0.6901,
+      "step": 5429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2173733079660087e-07,
+      "loss": 0.7582,
+      "step": 5430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1786567111945488e-07,
+      "loss": 0.8018,
+      "step": 5431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1402807362306266e-07,
+      "loss": 0.7678,
+      "step": 5432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1022453961745402e-07,
+      "loss": 0.74,
+      "step": 5433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0645507040102375e-07,
+      "loss": 0.7601,
+      "step": 5434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027196672605314e-07,
+      "loss": 0.7182,
+      "step": 5435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.990183314711236e-07,
+      "loss": 0.7535,
+      "step": 5436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.953510642963119e-07,
+      "loss": 0.7245,
+      "step": 5437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9171786698797268e-07,
+      "loss": 0.7546,
+      "step": 5438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8811874078634717e-07,
+      "loss": 0.8007,
+      "step": 5439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8455368692005259e-07,
+      "loss": 0.7753,
+      "step": 5440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8102270660608213e-07,
+      "loss": 0.7627,
+      "step": 5441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7752580104978266e-07,
+      "loss": 0.7239,
+      "step": 5442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7406297144488827e-07,
+      "loss": 0.7238,
+      "step": 5443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7063421897347553e-07,
+      "loss": 0.7586,
+      "step": 5444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6723954480601934e-07,
+      "loss": 0.8205,
+      "step": 5445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6387895010133713e-07,
+      "loss": 0.7361,
+      "step": 5446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.605524360066113e-07,
+      "loss": 0.6978,
+      "step": 5447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726000365741123e-07,
+      "loss": 0.7892,
+      "step": 5448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.540016541776601e-07,
+      "loss": 0.7145,
+      "step": 5449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5077738867963486e-07,
+      "loss": 0.7493,
+      "step": 5450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4758720826398843e-07,
+      "loss": 0.7656,
+      "step": 5451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4443111401976072e-07,
+      "loss": 0.6991,
+      "step": 5452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.413091070243011e-07,
+      "loss": 0.6982,
+      "step": 5453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3822118834337927e-07,
+      "loss": 0.2727,
+      "step": 5454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3516735903109645e-07,
+      "loss": 0.7246,
+      "step": 5455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3214762012992988e-07,
+      "loss": 0.7931,
+      "step": 5456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.291619726706994e-07,
+      "loss": 0.7221,
+      "step": 5457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.262104176726231e-07,
+      "loss": 0.6925,
+      "step": 5458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2329295614323943e-07,
+      "loss": 0.7259,
+      "step": 5459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.20409589078474e-07,
+      "loss": 0.7636,
+      "step": 5460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.175603174626283e-07,
+      "loss": 0.8057,
+      "step": 5461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1474514226832434e-07,
+      "loss": 0.6788,
+      "step": 5462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1196406445657114e-07,
+      "loss": 0.7047,
+      "step": 5463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0921708497673155e-07,
+      "loss": 0.7463,
+      "step": 5464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0650420476653322e-07,
+      "loss": 0.7354,
+      "step": 5465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0382542475206869e-07,
+      "loss": 0.7308,
+      "step": 5466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0118074584776204e-07,
+      "loss": 0.2864,
+      "step": 5467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.857016895642446e-08,
+      "loss": 0.8002,
+      "step": 5468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.599369496922083e-08,
+      "loss": 0.7559,
+      "step": 5469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.345132476565876e-08,
+      "loss": 0.2581,
+      "step": 5470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.094305921363289e-08,
+      "loss": 0.7825,
+      "step": 5471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.84688991693583e-08,
+      "loss": 0.7516,
+      "step": 5472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.602884547744827e-08,
+      "loss": 0.7501,
+      "step": 5473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.362289897084763e-08,
+      "loss": 0.7168,
+      "step": 5474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.125106047085496e-08,
+      "loss": 0.7546,
+      "step": 5475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.89133307871337e-08,
+      "loss": 0.7524,
+      "step": 5476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.660971071772327e-08,
+      "loss": 0.7393,
+      "step": 5477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.434020104897243e-08,
+      "loss": 0.6872,
+      "step": 5478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.210480255563923e-08,
+      "loss": 0.7934,
+      "step": 5479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.990351600079104e-08,
+      "loss": 0.8088,
+      "step": 5480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.773634213588231e-08,
+      "loss": 0.8198,
+      "step": 5481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.560328170071017e-08,
+      "loss": 0.7665,
+      "step": 5482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.350433542342549e-08,
+      "loss": 0.7361,
+      "step": 5483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.143950402054399e-08,
+      "loss": 0.7339,
+      "step": 5484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9408788196912977e-08,
+      "loss": 0.7801,
+      "step": 5485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.74121886457557e-08,
+      "loss": 0.2744,
+      "step": 5486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.544970604864919e-08,
+      "loss": 0.8025,
+      "step": 5487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.35213410755131e-08,
+      "loss": 0.7602,
+      "step": 5488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.162709438463198e-08,
+      "loss": 0.7258,
+      "step": 5489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9766966622621925e-08,
+      "loss": 0.7657,
+      "step": 5490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.794095842446389e-08,
+      "loss": 0.7311,
+      "step": 5491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6149070413514796e-08,
+      "loss": 0.7776,
+      "step": 5492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4391303201440916e-08,
+      "loss": 0.7377,
+      "step": 5493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.266765738830669e-08,
+      "loss": 0.7464,
+      "step": 5494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.097813356248592e-08,
+      "loss": 0.7619,
+      "step": 5495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.932273230072836e-08,
+      "loss": 0.6988,
+      "step": 5496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.770145416813753e-08,
+      "loss": 0.7362,
+      "step": 5497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6114299718159605e-08,
+      "loss": 0.8243,
+      "step": 5498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.456126949259453e-08,
+      "loss": 0.7657,
+      "step": 5499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.304236402158489e-08,
+      "loss": 0.7568,
+      "step": 5500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.155758382363816e-08,
+      "loss": 0.7641,
+      "step": 5501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0106929405615544e-08,
+      "loss": 0.8019,
+      "step": 5502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.869040126269873e-08,
+      "loss": 0.7203,
+      "step": 5503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.730799987846755e-08,
+      "loss": 0.7917,
+      "step": 5504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5959725724811202e-08,
+      "loss": 0.7897,
+      "step": 5505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4645579261994844e-08,
+      "loss": 0.7784,
+      "step": 5506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3365560938604092e-08,
+      "loss": 0.7598,
+      "step": 5507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2119671191611624e-08,
+      "loss": 0.7639,
+      "step": 5508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0907910446310576e-08,
+      "loss": 0.7442,
+      "step": 5509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.973027911635894e-08,
+      "loss": 0.776,
+      "step": 5510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8586777603746276e-08,
+      "loss": 0.721,
+      "step": 5511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7477406298849197e-08,
+      "loss": 0.79,
+      "step": 5512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.640216558035368e-08,
+      "loss": 0.7304,
+      "step": 5513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5361055815310553e-08,
+      "loss": 0.7777,
+      "step": 5514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4354077359124419e-08,
+      "loss": 0.2439,
+      "step": 5515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.338123055553142e-08,
+      "loss": 0.751,
+      "step": 5516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2442515736643678e-08,
+      "loss": 0.2774,
+      "step": 5517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.153793322289376e-08,
+      "loss": 0.7496,
+      "step": 5518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0667483323079097e-08,
+      "loss": 0.7706,
+      "step": 5519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.831166334339782e-09,
+      "loss": 0.7515,
+      "step": 5520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.028982542169661e-09,
+      "loss": 0.765,
+      "step": 5521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.260932220405249e-09,
+      "loss": 0.7728,
+      "step": 5522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.52701563122571e-09,
+      "loss": 0.8626,
+      "step": 5523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.827233025175073e-09,
+      "loss": 0.706,
+      "step": 5524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1615846411289255e-09,
+      "loss": 0.806,
+      "step": 5525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.530070706316615e-09,
+      "loss": 0.7348,
+      "step": 5526
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.932691436310144e-09,
+      "loss": 0.685,
+      "step": 5527
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3694470350463814e-09,
+      "loss": 0.707,
+      "step": 5528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.84033769479375e-09,
+      "loss": 0.7058,
+      "step": 5529
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3453635961744332e-09,
+      "loss": 0.7415,
+      "step": 5530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8845249081421722e-09,
+      "loss": 0.7312,
+      "step": 5531
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.457821788026671e-09,
+      "loss": 0.6871,
+      "step": 5532
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.065254381478088e-09,
+      "loss": 0.6962,
+      "step": 5533
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7068228225114447e-09,
+      "loss": 0.7557,
+      "step": 5534
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3825272334844208e-09,
+      "loss": 0.7778,
+      "step": 5535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0923677250973541e-09,
+      "loss": 0.7202,
+      "step": 5536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.363443964043427e-10,
+      "loss": 0.7776,
+      "step": 5537
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.144573347910409e-10,
+      "loss": 0.7467,
+      "step": 5538
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2670661601906804e-10,
+      "loss": 0.7403,
+      "step": 5539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7309230417049694e-10,
+      "loss": 0.7293,
+      "step": 5540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5361445168116107e-10,
+      "loss": 0.7723,
+      "step": 5541
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.827309935175663e-11,
+      "loss": 0.2507,
+      "step": 5542
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.706827629233132e-11,
+      "loss": 0.7268,
+      "step": 5543
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7824,
+      "step": 5544
+    },
+    {
+      "epoch": 1.0,
+      "step": 5544,
+      "total_flos": 3.200558990229504e+16,
+      "train_loss": 0.7664979397874288,
+      "train_runtime": 200252.6174,
+      "train_samples_per_second": 3.322,
+      "train_steps_per_second": 0.028
+    }
+  ],
+  "max_steps": 5544,
+  "num_train_epochs": 1,
+  "total_flos": 3.200558990229504e+16,
+  "trial_name": null,
+  "trial_params": null
+}