diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,43213 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998958152457024,
+  "global_step": 7198,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.259259259259259e-08,
+      "loss": 1.3534,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8518518518518518e-07,
+      "loss": 1.8937,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7777777777777776e-07,
+      "loss": 1.7758,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7037037037037036e-07,
+      "loss": 1.8755,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6296296296296297e-07,
+      "loss": 1.879,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 1.7936,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.481481481481481e-07,
+      "loss": 1.8185,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 0.2265,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.7357,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 1.7015,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0185185185185185e-06,
+      "loss": 1.6423,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 1.6487,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2037037037037037e-06,
+      "loss": 1.7433,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2962962962962962e-06,
+      "loss": 1.5067,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3888888888888892e-06,
+      "loss": 1.5557,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 0.2223,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5740740740740742e-06,
+      "loss": 1.4639,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.5154,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7592592592592594e-06,
+      "loss": 1.4327,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 1.4087,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.944444444444445e-06,
+      "loss": 1.2898,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.037037037037037e-06,
+      "loss": 1.2455,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1296296296296298e-06,
+      "loss": 1.2719,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.2507,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.314814814814815e-06,
+      "loss": 1.2723,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4074074074074075e-06,
+      "loss": 0.1912,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.1699,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 1.1391,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6851851851851856e-06,
+      "loss": 1.1421,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 1.1666,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8703703703703706e-06,
+      "loss": 1.2253,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 1.1476,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.055555555555556e-06,
+      "loss": 1.1211,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1481481481481483e-06,
+      "loss": 1.0739,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240740740740741e-06,
+      "loss": 1.0676,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0936,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4259259259259265e-06,
+      "loss": 1.0916,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5185185185185187e-06,
+      "loss": 1.1425,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6111111111111115e-06,
+      "loss": 1.0767,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 1.0875,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.796296296296297e-06,
+      "loss": 1.0672,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 0.2444,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9814814814814814e-06,
+      "loss": 1.0603,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.074074074074074e-06,
+      "loss": 1.0319,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.1951,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2592592592592596e-06,
+      "loss": 1.0914,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.351851851851852e-06,
+      "loss": 0.9852,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.0457,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.537037037037038e-06,
+      "loss": 1.0576,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.62962962962963e-06,
+      "loss": 0.9499,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722222222222222e-06,
+      "loss": 1.0815,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 1.0311,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.907407407407408e-06,
+      "loss": 0.969,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.2272,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.092592592592593e-06,
+      "loss": 1.0208,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 1.0091,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2777777777777785e-06,
+      "loss": 0.9681,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.370370370370371e-06,
+      "loss": 1.0579,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.462962962962963e-06,
+      "loss": 1.0138,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.9509,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6481481481481485e-06,
+      "loss": 1.005,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.740740740740741e-06,
+      "loss": 0.9846,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.9514,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 0.916,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.018518518518519e-06,
+      "loss": 0.9985,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.9207,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.203703703703704e-06,
+      "loss": 0.995,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.296296296296297e-06,
+      "loss": 0.9948,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3888888888888885e-06,
+      "loss": 0.981,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.481481481481482e-06,
+      "loss": 1.0054,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.574074074074075e-06,
+      "loss": 1.0161,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9232,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.75925925925926e-06,
+      "loss": 0.962,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.851851851851853e-06,
+      "loss": 0.995,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.9777,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 0.9734,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.129629629629629e-06,
+      "loss": 0.9874,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 0.9355,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.314814814814816e-06,
+      "loss": 0.8827,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.9589,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.9179,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.592592592592594e-06,
+      "loss": 1.0399,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.685185185185185e-06,
+      "loss": 0.9125,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.8964,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.870370370370372e-06,
+      "loss": 0.2138,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.962962962962963e-06,
+      "loss": 0.9621,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.055555555555557e-06,
+      "loss": 0.9314,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 0.9548,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.240740740740741e-06,
+      "loss": 1.01,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.9176,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.425925925925926e-06,
+      "loss": 0.9063,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.518518518518519e-06,
+      "loss": 0.949,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.611111111111112e-06,
+      "loss": 0.9319,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.703703703703705e-06,
+      "loss": 0.8925,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.796296296296297e-06,
+      "loss": 0.8841,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.9323,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.981481481481483e-06,
+      "loss": 0.9178,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.074074074074075e-06,
+      "loss": 0.8806,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.975,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.25925925925926e-06,
+      "loss": 1.0333,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.351851851851854e-06,
+      "loss": 0.9636,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 0.8416,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.537037037037037e-06,
+      "loss": 0.84,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.9929,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.8332,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.814814814814815e-06,
+      "loss": 0.9571,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.907407407407408e-06,
+      "loss": 0.9144,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.8743,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0092592592592594e-05,
+      "loss": 0.8747,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0185185185185186e-05,
+      "loss": 0.9771,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0277777777777777e-05,
+      "loss": 0.8556,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.2197,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0462962962962964e-05,
+      "loss": 0.9214,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 0.918,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.064814814814815e-05,
+      "loss": 0.9087,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0740740740740742e-05,
+      "loss": 0.949,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.8346,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0925925925925926e-05,
+      "loss": 0.9372,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1018518518518519e-05,
+      "loss": 0.9157,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.8588,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1203703703703706e-05,
+      "loss": 0.9519,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1296296296296297e-05,
+      "loss": 0.8975,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.138888888888889e-05,
+      "loss": 0.875,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1481481481481482e-05,
+      "loss": 0.9433,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1574074074074075e-05,
+      "loss": 0.9508,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.8746,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1759259259259259e-05,
+      "loss": 0.9065,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.8875,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1944444444444444e-05,
+      "loss": 0.8745,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2037037037037039e-05,
+      "loss": 0.8957,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2129629629629631e-05,
+      "loss": 0.8811,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 0.9265,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2314814814814815e-05,
+      "loss": 0.8515,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2407407407407408e-05,
+      "loss": 0.9456,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0062,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.8826,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2685185185185188e-05,
+      "loss": 0.9238,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 0.8975,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2870370370370371e-05,
+      "loss": 0.8133,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2962962962962964e-05,
+      "loss": 0.8756,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3055555555555557e-05,
+      "loss": 0.9016,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.314814814814815e-05,
+      "loss": 0.2294,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3240740740740742e-05,
+      "loss": 0.9211,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8347,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3425925925925926e-05,
+      "loss": 0.8944,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.351851851851852e-05,
+      "loss": 0.9579,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3611111111111113e-05,
+      "loss": 0.8671,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3703703703703706e-05,
+      "loss": 0.9236,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3796296296296297e-05,
+      "loss": 0.8628,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.8648,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3981481481481482e-05,
+      "loss": 0.8237,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.8916,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 0.855,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4259259259259259e-05,
+      "loss": 0.925,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4351851851851853e-05,
+      "loss": 0.8491,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.8812,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4537037037037039e-05,
+      "loss": 0.8828,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4629629629629631e-05,
+      "loss": 0.9377,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4722222222222224e-05,
+      "loss": 0.9583,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.9424,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4907407407407408e-05,
+      "loss": 0.8936,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.939,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5092592592592595e-05,
+      "loss": 0.8612,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5185185185185187e-05,
+      "loss": 0.8742,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5277777777777777e-05,
+      "loss": 0.8592,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.537037037037037e-05,
+      "loss": 0.8993,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5462962962962966e-05,
+      "loss": 0.8941,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.8878,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.564814814814815e-05,
+      "loss": 0.8688,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5740740740740744e-05,
+      "loss": 0.8771,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.8716,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5925925925925926e-05,
+      "loss": 0.9166,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.601851851851852e-05,
+      "loss": 0.928,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 0.9055,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6203703703703707e-05,
+      "loss": 0.8852,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.8473,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.638888888888889e-05,
+      "loss": 0.8684,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6481481481481482e-05,
+      "loss": 0.8992,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6574074074074075e-05,
+      "loss": 0.1736,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9305,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.675925925925926e-05,
+      "loss": 0.9154,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6851851851851853e-05,
+      "loss": 0.8753,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6944444444444446e-05,
+      "loss": 0.9658,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.963,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.712962962962963e-05,
+      "loss": 0.8651,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 0.8518,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7314814814814816e-05,
+      "loss": 0.8643,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 0.8392,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.8574,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7592592592592595e-05,
+      "loss": 0.8516,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7685185185185187e-05,
+      "loss": 0.9082,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.8926,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7870370370370373e-05,
+      "loss": 0.9113,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7962962962962965e-05,
+      "loss": 0.8609,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8055555555555558e-05,
+      "loss": 0.8472,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 0.7832,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8240740740740744e-05,
+      "loss": 0.8838,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.8202,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8425925925925926e-05,
+      "loss": 0.8584,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.8838,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8611111111111114e-05,
+      "loss": 0.9035,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8703703703703707e-05,
+      "loss": 0.8492,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8796296296296296e-05,
+      "loss": 0.8326,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.7941,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8981481481481482e-05,
+      "loss": 0.9185,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9074074074074075e-05,
+      "loss": 0.9285,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.9616,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.8704,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9351851851851853e-05,
+      "loss": 0.8919,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.877,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9537037037037038e-05,
+      "loss": 0.8643,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.962962962962963e-05,
+      "loss": 0.8853,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9722222222222224e-05,
+      "loss": 0.8424,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9814814814814816e-05,
+      "loss": 0.7993,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.990740740740741e-05,
+      "loss": 0.8996,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8646,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998987698096e-05,
+      "loss": 0.9007,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999595079259e-05,
+      "loss": 0.9521,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999908892841e-05,
+      "loss": 0.8026,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983803173647e-05,
+      "loss": 0.8312,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999974692462666e-05,
+      "loss": 0.8236,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963557152998e-05,
+      "loss": 0.9286,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999950397246907e-05,
+      "loss": 0.846,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999935212747047e-05,
+      "loss": 0.8433,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999180036565e-05,
+      "loss": 0.8731,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989876997875e-05,
+      "loss": 0.1818,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987751171768e-05,
+      "loss": 0.792,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999854228877605e-05,
+      "loss": 0.8066,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999982892146324e-05,
+      "loss": 0.786,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999801589479705e-05,
+      "loss": 0.2305,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999772232932527e-05,
+      "loss": 0.8324,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999740851827662e-05,
+      "loss": 0.8963,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999707446171458e-05,
+      "loss": 0.8811,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999672015970675e-05,
+      "loss": 0.8612,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999634561232488e-05,
+      "loss": 0.8266,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999595081964482e-05,
+      "loss": 0.8316,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999955357817465e-05,
+      "loss": 0.8165,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999510049871394e-05,
+      "loss": 0.8553,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999464497063527e-05,
+      "loss": 0.9156,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999416919760268e-05,
+      "loss": 0.9187,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999367317971256e-05,
+      "loss": 0.9046,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999931569170653e-05,
+      "loss": 0.8113,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999262040976543e-05,
+      "loss": 0.8681,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999206365792154e-05,
+      "loss": 0.9601,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999914866616464e-05,
+      "loss": 0.9064,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999908894210568e-05,
+      "loss": 0.795,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999027193627364e-05,
+      "loss": 0.8553,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99989634207422e-05,
+      "loss": 0.8363,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998897623463095e-05,
+      "loss": 0.8119,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998829801803368e-05,
+      "loss": 0.8428,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998759955776755e-05,
+      "loss": 0.8785,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998688085397393e-05,
+      "loss": 0.8543,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998614190679836e-05,
+      "loss": 0.8235,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998538271639044e-05,
+      "loss": 0.8116,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999846032829039e-05,
+      "loss": 0.7971,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998380360649648e-05,
+      "loss": 0.8357,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999829836873301e-05,
+      "loss": 0.8905,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998214352557084e-05,
+      "loss": 0.9021,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998128312138872e-05,
+      "loss": 0.8326,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998040247495794e-05,
+      "loss": 0.7825,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997950158645684e-05,
+      "loss": 0.7549,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997858045606773e-05,
+      "loss": 0.8416,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997763908397722e-05,
+      "loss": 0.8623,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997667747037584e-05,
+      "loss": 0.8384,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997569561545824e-05,
+      "loss": 0.9049,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999746935194233e-05,
+      "loss": 0.8715,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999736711824738e-05,
+      "loss": 0.9777,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999726286048168e-05,
+      "loss": 0.8652,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997156578666335e-05,
+      "loss": 0.9211,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997048272822864e-05,
+      "loss": 0.824,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996937942973194e-05,
+      "loss": 0.8179,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996825589139665e-05,
+      "loss": 0.207,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996711211345015e-05,
+      "loss": 0.8991,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996594809612412e-05,
+      "loss": 0.8426,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999647638396542e-05,
+      "loss": 0.8625,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999635593442801e-05,
+      "loss": 0.8382,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996233461024574e-05,
+      "loss": 0.9084,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996108963779907e-05,
+      "loss": 0.9066,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999598244271921e-05,
+      "loss": 0.8701,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995853897868108e-05,
+      "loss": 0.8964,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995723329252615e-05,
+      "loss": 0.8672,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995590736899174e-05,
+      "loss": 0.9029,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995456120834625e-05,
+      "loss": 0.8615,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999531948108623e-05,
+      "loss": 0.9196,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995180817681646e-05,
+      "loss": 0.8914,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995040130648947e-05,
+      "loss": 0.8222,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999489742001662e-05,
+      "loss": 0.2181,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994752685813554e-05,
+      "loss": 0.866,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994605928069054e-05,
+      "loss": 0.8663,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994457146812838e-05,
+      "loss": 0.855,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994306342075017e-05,
+      "loss": 0.89,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994153513886135e-05,
+      "loss": 0.8762,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993998662277125e-05,
+      "loss": 0.8236,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993841787279344e-05,
+      "loss": 0.9142,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993682888924546e-05,
+      "loss": 0.8763,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993521967244905e-05,
+      "loss": 0.8951,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993359022273008e-05,
+      "loss": 0.9033,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993194054041834e-05,
+      "loss": 0.8165,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999302706258479e-05,
+      "loss": 0.9553,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992858047935683e-05,
+      "loss": 0.8012,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992687010128736e-05,
+      "loss": 0.8393,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992513949198566e-05,
+      "loss": 0.8401,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999233886518022e-05,
+      "loss": 0.8852,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992161758109143e-05,
+      "loss": 0.7513,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991982628021193e-05,
+      "loss": 0.8544,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991801474952637e-05,
+      "loss": 0.9085,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999161829894015e-05,
+      "loss": 0.8717,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999143310002082e-05,
+      "loss": 0.8878,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999124587823214e-05,
+      "loss": 0.8519,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991056633612014e-05,
+      "loss": 0.8162,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999086536619876e-05,
+      "loss": 0.8603,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990672076031102e-05,
+      "loss": 0.8098,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990476763148173e-05,
+      "loss": 0.9012,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990279427589513e-05,
+      "loss": 0.8445,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990080069395077e-05,
+      "loss": 0.7736,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989878688605226e-05,
+      "loss": 0.8456,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989675285260738e-05,
+      "loss": 0.8105,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998946985940278e-05,
+      "loss": 0.8305,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998926241107296e-05,
+      "loss": 0.841,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989052940313267e-05,
+      "loss": 0.8598,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998884144716611e-05,
+      "loss": 0.8587,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988627931674315e-05,
+      "loss": 0.8205,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988412393881104e-05,
+      "loss": 0.9311,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998819483383012e-05,
+      "loss": 0.878,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987975251565405e-05,
+      "loss": 0.9339,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987753647131417e-05,
+      "loss": 0.848,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987530020573028e-05,
+      "loss": 0.8714,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987304371935506e-05,
+      "loss": 0.8112,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998707670126454e-05,
+      "loss": 0.9234,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986847008606222e-05,
+      "loss": 0.7897,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986615294007057e-05,
+      "loss": 0.8329,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986381557513962e-05,
+      "loss": 0.8908,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998614579917425e-05,
+      "loss": 0.8461,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998590801903566e-05,
+      "loss": 0.7766,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998566821714633e-05,
+      "loss": 0.9303,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985426393554812e-05,
+      "loss": 0.8656,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985182548310065e-05,
+      "loss": 0.8302,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998493668146146e-05,
+      "loss": 0.8423,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998468879305877e-05,
+      "loss": 0.744,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998443888315219e-05,
+      "loss": 0.8553,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984186951792308e-05,
+      "loss": 0.8873,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983932999030138e-05,
+      "loss": 0.8646,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998367702491709e-05,
+      "loss": 0.8082,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983419029504995e-05,
+      "loss": 0.8792,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998315901284608e-05,
+      "loss": 0.8198,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982896974992992e-05,
+      "loss": 0.8686,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998263291599878e-05,
+      "loss": 0.876,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998236683591691e-05,
+      "loss": 0.7719,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998209873480125e-05,
+      "loss": 0.9394,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998182861270608e-05,
+      "loss": 0.88,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981556469686086e-05,
+      "loss": 0.8197,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981282305796375e-05,
+      "loss": 0.8022,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981006121092447e-05,
+      "loss": 0.9067,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998072791563022e-05,
+      "loss": 0.853,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998044768946602e-05,
+      "loss": 0.9186,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998016544265658e-05,
+      "loss": 0.8334,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997988117525904e-05,
+      "loss": 0.8657,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979594887330968e-05,
+      "loss": 0.8485,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997930657893031e-05,
+      "loss": 0.8917,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979016250115446e-05,
+      "loss": 0.8786,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978723900945153e-05,
+      "loss": 0.798,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978429531478616e-05,
+      "loss": 0.9006,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997813314177544e-05,
+      "loss": 0.8871,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977834731895633e-05,
+      "loss": 0.8879,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977534301899604e-05,
+      "loss": 0.8224,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977231851848184e-05,
+      "loss": 0.951,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976927381802604e-05,
+      "loss": 0.8349,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997662089182451e-05,
+      "loss": 0.9103,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997631238197595e-05,
+      "loss": 0.2465,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976001852319388e-05,
+      "loss": 0.8394,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975689302917694e-05,
+      "loss": 0.8384,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975374733834147e-05,
+      "loss": 0.8053,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997505814513243e-05,
+      "loss": 0.8719,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997473953687665e-05,
+      "loss": 0.8595,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99744189091313e-05,
+      "loss": 0.8374,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974096261961302e-05,
+      "loss": 0.8651,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997377159543198e-05,
+      "loss": 0.8893,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973444909609063e-05,
+      "loss": 0.885,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973116204558695e-05,
+      "loss": 0.8883,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997278548034742e-05,
+      "loss": 0.8736,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99724527370422e-05,
+      "loss": 0.8342,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972117974710406e-05,
+      "loss": 0.8068,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997178119341981e-05,
+      "loss": 0.8897,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971442393238595e-05,
+      "loss": 0.8547,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997110157423536e-05,
+      "loss": 0.8448,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99707587364791e-05,
+      "loss": 0.8991,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970413880039236e-05,
+      "loss": 0.8916,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997006700498558e-05,
+      "loss": 0.8243,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969718111388362e-05,
+      "loss": 0.8617,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996936719931822e-05,
+      "loss": 0.8083,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99690142688462e-05,
+      "loss": 0.8288,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968659320043754e-05,
+      "loss": 0.9021,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996830235298275e-05,
+      "loss": 0.7753,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967943367735458e-05,
+      "loss": 0.8205,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967582364374552e-05,
+      "loss": 0.8076,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967219342973133e-05,
+      "loss": 0.8294,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966854303604687e-05,
+      "loss": 0.7946,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966487246343128e-05,
+      "loss": 0.7921,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996611817126276e-05,
+      "loss": 0.9026,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996574707843832e-05,
+      "loss": 0.8367,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996537396794493e-05,
+      "loss": 0.8333,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964998839858135e-05,
+      "loss": 0.8235,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964621694253878e-05,
+      "loss": 0.7803,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964242531208523e-05,
+      "loss": 0.8709,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996386135079883e-05,
+      "loss": 0.8808,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963478153101974e-05,
+      "loss": 0.8465,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996309293819554e-05,
+      "loss": 0.8667,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962705706157516e-05,
+      "loss": 0.7437,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962316457066304e-05,
+      "loss": 0.8482,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961925191000706e-05,
+      "loss": 0.8221,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961531908039943e-05,
+      "loss": 0.9171,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996113660826364e-05,
+      "loss": 0.8368,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960739291751825e-05,
+      "loss": 0.7859,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960339958584943e-05,
+      "loss": 0.8052,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995993860884384e-05,
+      "loss": 0.8353,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959535242609772e-05,
+      "loss": 0.8763,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959129859964412e-05,
+      "loss": 0.8176,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958722460989827e-05,
+      "loss": 0.827,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99583130457685e-05,
+      "loss": 0.8858,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957901614383327e-05,
+      "loss": 0.7823,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99574881669176e-05,
+      "loss": 0.8686,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957072703455027e-05,
+      "loss": 0.8616,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956655224079725e-05,
+      "loss": 0.836,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956235728876218e-05,
+      "loss": 0.9135,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995581421792943e-05,
+      "loss": 0.8957,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955390691324708e-05,
+      "loss": 0.8707,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99549651491478e-05,
+      "loss": 0.8185,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954537591484853e-05,
+      "loss": 0.822,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995410801842244e-05,
+      "loss": 0.8862,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953676430047527e-05,
+      "loss": 0.8157,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953242826447498e-05,
+      "loss": 0.8281,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995280720771013e-05,
+      "loss": 0.8328,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995236957392363e-05,
+      "loss": 0.8414,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99519299251766e-05,
+      "loss": 0.8144,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951488261558042e-05,
+      "loss": 0.8518,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995104458315739e-05,
+      "loss": 0.8707,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995059889006446e-05,
+      "loss": 0.8738,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950151182369487e-05,
+      "loss": 0.7858,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949701460163123e-05,
+      "loss": 0.8687,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949249723536414e-05,
+      "loss": 0.7571,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948795972580815e-05,
+      "loss": 0.897,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948340207388197e-05,
+      "loss": 0.7868,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947882428050834e-05,
+      "loss": 0.8502,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994742263466141e-05,
+      "loss": 0.9571,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994696082731301e-05,
+      "loss": 0.8428,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946497006099133e-05,
+      "loss": 0.8299,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994603117111369e-05,
+      "loss": 0.8689,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945563322450986e-05,
+      "loss": 0.8757,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945093460205744e-05,
+      "loss": 0.8904,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99446215844731e-05,
+      "loss": 0.8264,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994414769534858e-05,
+      "loss": 0.7949,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943671792928132e-05,
+      "loss": 0.8482,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994319387730811e-05,
+      "loss": 0.8949,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994271394858527e-05,
+      "loss": 0.8399,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994223200685678e-05,
+      "loss": 0.8649,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994174805222021e-05,
+      "loss": 0.8857,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941262084773545e-05,
+      "loss": 0.7763,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940774104615175e-05,
+      "loss": 0.8646,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940284111843896e-05,
+      "loss": 0.8662,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993979210655891e-05,
+      "loss": 0.7559,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993929808885983e-05,
+      "loss": 0.8659,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938802058846677e-05,
+      "loss": 0.2263,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938304016619873e-05,
+      "loss": 0.8592,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993780396228026e-05,
+      "loss": 0.2239,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993730189592907e-05,
+      "loss": 0.8909,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936797817667955e-05,
+      "loss": 0.8694,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993629172759897e-05,
+      "loss": 0.771,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935783625824584e-05,
+      "loss": 0.8089,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935273512447657e-05,
+      "loss": 0.8795,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934761387571476e-05,
+      "loss": 0.7981,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934247251299722e-05,
+      "loss": 0.8997,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933731103736488e-05,
+      "loss": 0.8347,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993321294498627e-05,
+      "loss": 0.769,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993269277515398e-05,
+      "loss": 0.7861,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993217059434493e-05,
+      "loss": 0.8148,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993164640266484e-05,
+      "loss": 0.8294,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993112020021984e-05,
+      "loss": 0.8205,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930591987116463e-05,
+      "loss": 0.8216,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930061763461652e-05,
+      "loss": 0.8382,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929529529362756e-05,
+      "loss": 0.8617,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992899528492753e-05,
+      "loss": 0.857,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992845903026414e-05,
+      "loss": 0.8997,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927920765481158e-05,
+      "loss": 0.8021,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927380490687552e-05,
+      "loss": 0.7988,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992683820599272e-05,
+      "loss": 0.8672,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992629391150644e-05,
+      "loss": 0.8279,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992574760733892e-05,
+      "loss": 0.8548,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925199293600757e-05,
+      "loss": 0.7884,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924648970402968e-05,
+      "loss": 0.8832,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924096637856972e-05,
+      "loss": 0.8325,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992354229607459e-05,
+      "loss": 0.9139,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992298594516806e-05,
+      "loss": 0.87,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922427585250017e-05,
+      "loss": 0.9189,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992186721643351e-05,
+      "loss": 0.2298,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921304838831984e-05,
+      "loss": 0.8228,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920740452559306e-05,
+      "loss": 0.8464,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920174057729742e-05,
+      "loss": 0.1906,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991960565445796e-05,
+      "loss": 0.8599,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991903524285904e-05,
+      "loss": 0.9339,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991846282304847e-05,
+      "loss": 0.8381,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991788839514214e-05,
+      "loss": 0.8169,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991731195925635e-05,
+      "loss": 0.8463,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916733515507805e-05,
+      "loss": 0.8623,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916153064013616e-05,
+      "loss": 0.7782,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915570604891304e-05,
+      "loss": 0.801,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914986138258796e-05,
+      "loss": 0.7987,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991439966423442e-05,
+      "loss": 0.8702,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991381118293691e-05,
+      "loss": 0.8233,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913220694485416e-05,
+      "loss": 0.8143,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912628198999484e-05,
+      "loss": 0.8066,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912033696599075e-05,
+      "loss": 0.2214,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991143718740455e-05,
+      "loss": 0.8947,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910838671536682e-05,
+      "loss": 0.7849,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991023814911664e-05,
+      "loss": 0.7991,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909635620266014e-05,
+      "loss": 0.8899,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909031085106784e-05,
+      "loss": 0.8131,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908424543761345e-05,
+      "loss": 0.9523,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907815996352507e-05,
+      "loss": 0.8305,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907205443003468e-05,
+      "loss": 0.789,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906592883837845e-05,
+      "loss": 0.8725,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905978318979654e-05,
+      "loss": 0.9112,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905361748553323e-05,
+      "loss": 0.7832,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990474317268368e-05,
+      "loss": 0.8569,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904122591495967e-05,
+      "loss": 0.8242,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903500005115818e-05,
+      "loss": 0.784,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902875413669293e-05,
+      "loss": 0.8446,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902248817282838e-05,
+      "loss": 0.7916,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901620216083323e-05,
+      "loss": 0.8247,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900989610198007e-05,
+      "loss": 0.7819,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900356999754563e-05,
+      "loss": 0.828,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899722384881074e-05,
+      "loss": 0.8382,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899085765706024e-05,
+      "loss": 0.838,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898447142358298e-05,
+      "loss": 0.7468,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98978065149672e-05,
+      "loss": 0.8909,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897163883662425e-05,
+      "loss": 0.9358,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989651924857408e-05,
+      "loss": 0.7742,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895872609832684e-05,
+      "loss": 0.8302,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989522396756915e-05,
+      "loss": 0.8551,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894573321914804e-05,
+      "loss": 0.8451,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989392067300138e-05,
+      "loss": 0.8635,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893266020961008e-05,
+      "loss": 0.8208,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989260936592623e-05,
+      "loss": 0.8363,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891950708029994e-05,
+      "loss": 0.7912,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891290047405652e-05,
+      "loss": 0.7321,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890627384186965e-05,
+      "loss": 0.8109,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889962718508088e-05,
+      "loss": 0.8018,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98892960505036e-05,
+      "loss": 0.8286,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888627380308464e-05,
+      "loss": 0.7406,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887956708058066e-05,
+      "loss": 0.7488,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887284033888187e-05,
+      "loss": 0.9106,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886609357935017e-05,
+      "loss": 0.8066,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988593268033516e-05,
+      "loss": 0.8538,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98852540012256e-05,
+      "loss": 0.8251,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988457332074376e-05,
+      "loss": 0.8106,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988389063902744e-05,
+      "loss": 0.8411,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883205956214857e-05,
+      "loss": 0.818,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882519272444634e-05,
+      "loss": 0.7974,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98818305878558e-05,
+      "loss": 0.7846,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881139902587777e-05,
+      "loss": 0.8251,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880447216780413e-05,
+      "loss": 0.9129,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879752530573945e-05,
+      "loss": 0.2179,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879055844109014e-05,
+      "loss": 0.199,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987835715752668e-05,
+      "loss": 0.8181,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877656470968395e-05,
+      "loss": 0.9096,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987695378457602e-05,
+      "loss": 0.8985,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876249098491817e-05,
+      "loss": 0.8616,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987554241285847e-05,
+      "loss": 0.8367,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874833727819042e-05,
+      "loss": 0.8607,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987412304351702e-05,
+      "loss": 0.8688,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873410360096287e-05,
+      "loss": 0.8072,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872695677701134e-05,
+      "loss": 0.8894,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871978996476253e-05,
+      "loss": 0.8231,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987126031656675e-05,
+      "loss": 0.8324,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987053963811812e-05,
+      "loss": 0.9188,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869816961276287e-05,
+      "loss": 0.8295,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869092286187548e-05,
+      "loss": 0.7895,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986836561299863e-05,
+      "loss": 0.8152,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986763694185665e-05,
+      "loss": 0.8539,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866906272909144e-05,
+      "loss": 0.7676,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866173606304035e-05,
+      "loss": 0.8411,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986543894218966e-05,
+      "loss": 0.7896,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864702280714764e-05,
+      "loss": 0.8839,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863963622028484e-05,
+      "loss": 0.8599,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986322296628038e-05,
+      "loss": 0.8292,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862480313620397e-05,
+      "loss": 0.7572,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861735664198896e-05,
+      "loss": 0.835,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986098901816664e-05,
+      "loss": 0.8177,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986024037567479e-05,
+      "loss": 0.8206,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985948973687492e-05,
+      "loss": 0.8961,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858737101919007e-05,
+      "loss": 0.8698,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857982470959425e-05,
+      "loss": 0.7974,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985722584414896e-05,
+      "loss": 0.7885,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98564672216408e-05,
+      "loss": 0.7848,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855706603588533e-05,
+      "loss": 0.8653,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854943990146153e-05,
+      "loss": 0.8083,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985417938146806e-05,
+      "loss": 0.7671,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853412777709065e-05,
+      "loss": 0.8145,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985264417902436e-05,
+      "loss": 0.8957,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985187358556957e-05,
+      "loss": 0.8202,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851100997500704e-05,
+      "loss": 0.8934,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850326414974173e-05,
+      "loss": 0.8735,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984954983814681e-05,
+      "loss": 0.8474,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984877126717584e-05,
+      "loss": 0.8259,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847990702218885e-05,
+      "loss": 0.848,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847208143433988e-05,
+      "loss": 0.8013,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846423590979583e-05,
+      "loss": 0.7948,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845637045014502e-05,
+      "loss": 0.8397,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844848505698004e-05,
+      "loss": 0.8717,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844057973189724e-05,
+      "loss": 0.7985,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843265447649724e-05,
+      "loss": 0.7815,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984247092923845e-05,
+      "loss": 0.8715,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984167441811677e-05,
+      "loss": 0.7937,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840875914445935e-05,
+      "loss": 0.8217,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984007541838762e-05,
+      "loss": 0.8177,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983927293010389e-05,
+      "loss": 0.802,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838468449757218e-05,
+      "loss": 0.8233,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837661977510473e-05,
+      "loss": 0.8229,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836853513526946e-05,
+      "loss": 0.8363,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836043057970307e-05,
+      "loss": 0.7814,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835230611004646e-05,
+      "loss": 0.873,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834416172794455e-05,
+      "loss": 0.8597,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983359974350462e-05,
+      "loss": 0.8824,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832781323300437e-05,
+      "loss": 0.8273,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831960912347603e-05,
+      "loss": 0.8451,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831138510812222e-05,
+      "loss": 0.8263,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830314118860797e-05,
+      "loss": 0.8208,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982948773666023e-05,
+      "loss": 0.8174,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828659364377833e-05,
+      "loss": 0.8537,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827829002181323e-05,
+      "loss": 0.8723,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982699665023881e-05,
+      "loss": 0.9268,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826162308718816e-05,
+      "loss": 0.7626,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982532597779026e-05,
+      "loss": 0.841,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824487657622466e-05,
+      "loss": 0.8417,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982364734838516e-05,
+      "loss": 0.8423,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982280505024847e-05,
+      "loss": 0.7632,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821960763382933e-05,
+      "loss": 0.8094,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982111448795948e-05,
+      "loss": 0.8555,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982026622414945e-05,
+      "loss": 0.7842,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819415972124582e-05,
+      "loss": 0.9244,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818563732057016e-05,
+      "loss": 0.8634,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817709504119302e-05,
+      "loss": 0.806,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981685328848438e-05,
+      "loss": 0.7883,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815995085325607e-05,
+      "loss": 0.7493,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815134894816735e-05,
+      "loss": 0.8721,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981427271713191e-05,
+      "loss": 0.7884,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813408552445697e-05,
+      "loss": 0.7905,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812542400933056e-05,
+      "loss": 0.7918,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811674262769344e-05,
+      "loss": 0.8952,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810804138130322e-05,
+      "loss": 0.7503,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809932027192162e-05,
+      "loss": 0.8923,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980905793013143e-05,
+      "loss": 0.859,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808181847125093e-05,
+      "loss": 0.8945,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807303778350526e-05,
+      "loss": 0.8274,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806423723985505e-05,
+      "loss": 0.8084,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98055416842082e-05,
+      "loss": 0.8512,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804657659197194e-05,
+      "loss": 0.7835,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980377164913147e-05,
+      "loss": 0.8669,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802883654190403e-05,
+      "loss": 0.7648,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980199367455378e-05,
+      "loss": 0.7979,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801101710401785e-05,
+      "loss": 0.9274,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800207761915006e-05,
+      "loss": 0.8984,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799311829274433e-05,
+      "loss": 0.8306,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979841391266146e-05,
+      "loss": 0.7786,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979751401225787e-05,
+      "loss": 0.8197,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796612128245866e-05,
+      "loss": 0.8597,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795708260808044e-05,
+      "loss": 0.83,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794802410127394e-05,
+      "loss": 0.7373,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793894576387324e-05,
+      "loss": 0.8051,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792984759771628e-05,
+      "loss": 0.9022,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979207296046451e-05,
+      "loss": 0.9089,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979115917865057e-05,
+      "loss": 0.8569,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790243414514817e-05,
+      "loss": 0.8754,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789325668242658e-05,
+      "loss": 0.8463,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788405940019894e-05,
+      "loss": 0.8521,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787484230032738e-05,
+      "loss": 0.8437,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97865605384678e-05,
+      "loss": 0.8244,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785634865512088e-05,
+      "loss": 0.831,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978470721135302e-05,
+      "loss": 0.8314,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97837775761784e-05,
+      "loss": 0.8046,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978284596017645e-05,
+      "loss": 0.7952,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978191236353578e-05,
+      "loss": 0.8322,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780976786445413e-05,
+      "loss": 0.8214,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780039229094762e-05,
+      "loss": 0.8111,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779099691673643e-05,
+      "loss": 0.8336,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977815817437228e-05,
+      "loss": 0.894,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977721467738129e-05,
+      "loss": 0.8235,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776269200891693e-05,
+      "loss": 0.7597,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775321745094915e-05,
+      "loss": 0.8561,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774372310182772e-05,
+      "loss": 0.8212,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773420896347494e-05,
+      "loss": 0.8316,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772467503781696e-05,
+      "loss": 0.7851,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977151213267841e-05,
+      "loss": 0.7743,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977055478323106e-05,
+      "loss": 0.8485,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769595455633463e-05,
+      "loss": 0.2324,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768634150079854e-05,
+      "loss": 0.8108,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976767086676486e-05,
+      "loss": 0.8953,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766705605883496e-05,
+      "loss": 0.8063,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765738367631205e-05,
+      "loss": 0.7901,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97647691522038e-05,
+      "loss": 0.8299,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976379795979752e-05,
+      "loss": 0.8452,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762824790608985e-05,
+      "loss": 0.9012,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976184964483523e-05,
+      "loss": 0.7694,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760872522673677e-05,
+      "loss": 0.79,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759893424322156e-05,
+      "loss": 0.8123,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97589123499789e-05,
+      "loss": 0.7527,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975792929984253e-05,
+      "loss": 0.8359,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756944274112087e-05,
+      "loss": 0.7677,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755957272986987e-05,
+      "loss": 0.7997,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754968296667065e-05,
+      "loss": 0.7945,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753977345352546e-05,
+      "loss": 0.8782,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975298441924406e-05,
+      "loss": 0.8834,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751989518542637e-05,
+      "loss": 0.8765,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750992643449706e-05,
+      "loss": 0.7869,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974999379416709e-05,
+      "loss": 0.8236,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974899297089702e-05,
+      "loss": 0.8466,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974799017384212e-05,
+      "loss": 0.8172,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974698540320542e-05,
+      "loss": 0.7868,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745978659190347e-05,
+      "loss": 0.8404,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974496994200072e-05,
+      "loss": 0.7979,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743959251840768e-05,
+      "loss": 0.8414,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742946588915123e-05,
+      "loss": 0.7388,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97419319534288e-05,
+      "loss": 0.7045,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740915345587224e-05,
+      "loss": 0.8644,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739896765596218e-05,
+      "loss": 0.7838,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738876213662007e-05,
+      "loss": 0.8961,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973785368999121e-05,
+      "loss": 0.824,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973682919479085e-05,
+      "loss": 0.8113,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735802728268337e-05,
+      "loss": 0.8391,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734774290631504e-05,
+      "loss": 0.847,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973374388208856e-05,
+      "loss": 0.7487,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732711502848128e-05,
+      "loss": 0.7735,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731677153119216e-05,
+      "loss": 0.8206,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730640833111242e-05,
+      "loss": 0.8025,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729602543034026e-05,
+      "loss": 0.883,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728562283097773e-05,
+      "loss": 0.7643,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972752005351309e-05,
+      "loss": 0.8155,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726475854491004e-05,
+      "loss": 0.8014,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972542968624291e-05,
+      "loss": 0.8101,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724381548980615e-05,
+      "loss": 0.8251,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723331442916336e-05,
+      "loss": 0.8278,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972227936826267e-05,
+      "loss": 0.792,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721225325232617e-05,
+      "loss": 0.8478,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972016931403959e-05,
+      "loss": 0.7955,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719111334897382e-05,
+      "loss": 0.832,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718051388020192e-05,
+      "loss": 0.7638,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971698947362262e-05,
+      "loss": 0.7738,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971592559191966e-05,
+      "loss": 0.8227,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714859743126705e-05,
+      "loss": 0.8324,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971379192745955e-05,
+      "loss": 0.8208,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971272214513438e-05,
+      "loss": 0.8481,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971165039636779e-05,
+      "loss": 0.8449,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710576681376764e-05,
+      "loss": 0.7468,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709501000378686e-05,
+      "loss": 0.7755,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708423353591338e-05,
+      "loss": 0.8471,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97073437412329e-05,
+      "loss": 0.7643,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706262163521958e-05,
+      "loss": 0.791,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705178620677482e-05,
+      "loss": 0.8017,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704093112918844e-05,
+      "loss": 0.9096,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703005640465824e-05,
+      "loss": 0.8386,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701916203538584e-05,
+      "loss": 0.8853,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97008248023577e-05,
+      "loss": 0.9052,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699731437144135e-05,
+      "loss": 0.8677,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969863610811925e-05,
+      "loss": 0.7517,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697538815504802e-05,
+      "loss": 0.8269,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696439559522958e-05,
+      "loss": 0.8157,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695338340396265e-05,
+      "loss": 0.7995,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694235158347688e-05,
+      "loss": 0.8727,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693130013600566e-05,
+      "loss": 0.8004,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969202290637865e-05,
+      "loss": 0.8695,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969091383690609e-05,
+      "loss": 0.7859,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689802805407425e-05,
+      "loss": 0.7825,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688689812107593e-05,
+      "loss": 0.8689,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687574857231935e-05,
+      "loss": 0.8235,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686457941006182e-05,
+      "loss": 0.8729,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968533906365647e-05,
+      "loss": 0.8465,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968421822540932e-05,
+      "loss": 0.7452,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683095426491668e-05,
+      "loss": 0.8175,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681970667130824e-05,
+      "loss": 0.7086,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680843947554512e-05,
+      "loss": 0.8467,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679715267990852e-05,
+      "loss": 0.8751,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678584628668356e-05,
+      "loss": 0.7961,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967745202981593e-05,
+      "loss": 0.7834,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967631747166288e-05,
+      "loss": 0.8684,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675180954438913e-05,
+      "loss": 0.8307,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674042478374125e-05,
+      "loss": 0.8445,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672902043699017e-05,
+      "loss": 0.8411,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671759650644477e-05,
+      "loss": 0.8839,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670615299441795e-05,
+      "loss": 0.7655,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669468990322663e-05,
+      "loss": 0.8361,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668320723519153e-05,
+      "loss": 0.8607,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667170499263752e-05,
+      "loss": 0.7659,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666018317789327e-05,
+      "loss": 0.8141,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664864179329157e-05,
+      "loss": 0.8535,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966370808411691e-05,
+      "loss": 0.8328,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966255003238664e-05,
+      "loss": 0.781,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661390024372815e-05,
+      "loss": 0.7976,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660228060310286e-05,
+      "loss": 0.8858,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659064140434306e-05,
+      "loss": 0.8509,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657898264980524e-05,
+      "loss": 0.8004,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656730434184982e-05,
+      "loss": 0.7923,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965556064828412e-05,
+      "loss": 0.8962,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654388907514776e-05,
+      "loss": 0.7538,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653215212114175e-05,
+      "loss": 0.7702,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965203956231995e-05,
+      "loss": 0.7498,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965086195837012e-05,
+      "loss": 0.7108,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649682400503103e-05,
+      "loss": 0.8273,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648500888957717e-05,
+      "loss": 0.7813,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647317423973164e-05,
+      "loss": 0.8073,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646132005789056e-05,
+      "loss": 0.7735,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644944634645388e-05,
+      "loss": 0.7869,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643755310782558e-05,
+      "loss": 0.7704,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642564034441358e-05,
+      "loss": 0.7987,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641370805862968e-05,
+      "loss": 0.2109,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964017562528898e-05,
+      "loss": 0.7738,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638978492961364e-05,
+      "loss": 0.7696,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637779409122493e-05,
+      "loss": 0.7778,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963657837401513e-05,
+      "loss": 0.7734,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635375387882445e-05,
+      "loss": 0.8046,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634170450967992e-05,
+      "loss": 0.8298,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632963563515718e-05,
+      "loss": 0.8549,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631754725769977e-05,
+      "loss": 0.8924,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630543937975506e-05,
+      "loss": 0.8406,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629331200377445e-05,
+      "loss": 0.8039,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962811651322132e-05,
+      "loss": 0.8026,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626899876753064e-05,
+      "loss": 0.7975,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625681291218993e-05,
+      "loss": 0.8199,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962446075686582e-05,
+      "loss": 0.8301,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623238273940663e-05,
+      "loss": 0.7636,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962201384269102e-05,
+      "loss": 0.8928,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962078746336479e-05,
+      "loss": 0.8503,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619559136210273e-05,
+      "loss": 0.8001,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618328861476145e-05,
+      "loss": 0.8796,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617096639411495e-05,
+      "loss": 0.875,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615862470265803e-05,
+      "loss": 0.8734,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614626354288933e-05,
+      "loss": 0.8138,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961338829173115e-05,
+      "loss": 0.8002,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961214828284311e-05,
+      "loss": 0.8451,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610906327875878e-05,
+      "loss": 0.815,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609662427080886e-05,
+      "loss": 0.7685,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960841658070998e-05,
+      "loss": 0.8031,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607168789015402e-05,
+      "loss": 0.8244,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960591905224977e-05,
+      "loss": 0.8325,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960466737066611e-05,
+      "loss": 0.7283,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603413744517835e-05,
+      "loss": 0.8223,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602158174058757e-05,
+      "loss": 0.7952,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600900659543086e-05,
+      "loss": 0.8474,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959964120122541e-05,
+      "loss": 0.8375,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598379799360718e-05,
+      "loss": 0.7969,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597116454204402e-05,
+      "loss": 0.2601,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595851166012233e-05,
+      "loss": 0.8283,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594583935040386e-05,
+      "loss": 0.8321,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593314761545425e-05,
+      "loss": 0.8785,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95920436457843e-05,
+      "loss": 0.2218,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959077058801437e-05,
+      "loss": 0.804,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958949558849337e-05,
+      "loss": 0.7638,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588218647479455e-05,
+      "loss": 0.8171,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958693976523113e-05,
+      "loss": 0.835,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585658942007336e-05,
+      "loss": 0.894,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584376178067386e-05,
+      "loss": 0.8068,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583091473670983e-05,
+      "loss": 0.2229,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958180482907823e-05,
+      "loss": 0.7762,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580516244549634e-05,
+      "loss": 0.7903,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957922572034606e-05,
+      "loss": 0.8467,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577933256728806e-05,
+      "loss": 0.7509,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576638853959538e-05,
+      "loss": 0.7775,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575342512300323e-05,
+      "loss": 0.8449,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574044232013617e-05,
+      "loss": 0.848,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572744013362267e-05,
+      "loss": 0.8137,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571441856609525e-05,
+      "loss": 0.8217,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957013776201902e-05,
+      "loss": 0.7855,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956883172985478e-05,
+      "loss": 0.858,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567523760381224e-05,
+      "loss": 0.6915,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566213853863168e-05,
+      "loss": 0.7994,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956490201056581e-05,
+      "loss": 0.8413,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563588230754754e-05,
+      "loss": 0.9324,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956227251469598e-05,
+      "loss": 0.8168,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560954862655876e-05,
+      "loss": 0.7986,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559635274901206e-05,
+      "loss": 0.8236,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558313751699146e-05,
+      "loss": 0.8936,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955699029331724e-05,
+      "loss": 0.8008,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555664900023445e-05,
+      "loss": 0.8395,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955433757208609e-05,
+      "loss": 0.8319,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955300830977392e-05,
+      "loss": 0.8333,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551677113356046e-05,
+      "loss": 0.2163,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550343983101993e-05,
+      "loss": 0.2371,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549008919281656e-05,
+      "loss": 0.7722,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547671922165342e-05,
+      "loss": 0.8484,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546332992023734e-05,
+      "loss": 0.8285,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544992129127914e-05,
+      "loss": 0.7348,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543649333749356e-05,
+      "loss": 0.7769,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954230460615992e-05,
+      "loss": 0.7933,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954095794663186e-05,
+      "loss": 0.8119,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953960935543782e-05,
+      "loss": 0.8206,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953825883285084e-05,
+      "loss": 0.8166,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536906379144347e-05,
+      "loss": 0.8324,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953555199459216e-05,
+      "loss": 0.8352,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953419567946848e-05,
+      "loss": 0.8444,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953283743404792e-05,
+      "loss": 0.8256,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953147725860546e-05,
+      "loss": 0.8587,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530115153416487e-05,
+      "loss": 0.8002,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952875111875677e-05,
+      "loss": 0.8648,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952738515490248e-05,
+      "loss": 0.8075,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952601726213016e-05,
+      "loss": 0.8228,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952464744071676e-05,
+      "loss": 0.7887,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952327569093961e-05,
+      "loss": 0.884,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521902013076443e-05,
+      "loss": 0.8205,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520526407405363e-05,
+      "loss": 0.7276,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951914887420489e-05,
+      "loss": 0.75,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517769413753907e-05,
+      "loss": 0.8053,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951638802633171e-05,
+      "loss": 0.7797,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515004712217967e-05,
+      "loss": 0.8202,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513619471692747e-05,
+      "loss": 0.8314,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951223230503651e-05,
+      "loss": 0.8777,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510843212530098e-05,
+      "loss": 0.7382,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950945219445475e-05,
+      "loss": 0.8616,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508059251092087e-05,
+      "loss": 0.8067,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950666438272413e-05,
+      "loss": 0.7021,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505267589633286e-05,
+      "loss": 0.8479,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503868872102348e-05,
+      "loss": 0.7829,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502468230414494e-05,
+      "loss": 0.852,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950106566485331e-05,
+      "loss": 0.8346,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499661175702754e-05,
+      "loss": 0.7513,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949825476324718e-05,
+      "loss": 0.8452,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496846427771325e-05,
+      "loss": 0.7471,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495436169560334e-05,
+      "loss": 0.8776,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494023988899717e-05,
+      "loss": 0.8514,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492609886075394e-05,
+      "loss": 0.7862,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491193861373654e-05,
+      "loss": 0.7611,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489775915081195e-05,
+      "loss": 0.8123,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948835604748509e-05,
+      "loss": 0.7611,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486934258872813e-05,
+      "loss": 0.8533,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485510549532207e-05,
+      "loss": 0.7977,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948408491975153e-05,
+      "loss": 0.7981,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482657369819404e-05,
+      "loss": 0.222,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481227900024863e-05,
+      "loss": 0.8487,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479796510657307e-05,
+      "loss": 0.7906,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947836320200654e-05,
+      "loss": 0.823,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947692797436276e-05,
+      "loss": 0.774,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475490828016525e-05,
+      "loss": 0.8614,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474051763258814e-05,
+      "loss": 0.8734,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472610780380976e-05,
+      "loss": 0.8332,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947116787967475e-05,
+      "loss": 0.8811,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469723061432275e-05,
+      "loss": 0.8143,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946827632594606e-05,
+      "loss": 0.8457,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466827673509017e-05,
+      "loss": 0.8141,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946537710441444e-05,
+      "loss": 0.8397,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463924618956015e-05,
+      "loss": 0.8139,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94624702174278e-05,
+      "loss": 0.7987,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946101390012427e-05,
+      "loss": 0.8253,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459555667340264e-05,
+      "loss": 0.81,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458095519371014e-05,
+      "loss": 0.8421,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945663345651214e-05,
+      "loss": 0.7682,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455169479059665e-05,
+      "loss": 0.8187,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453703587309974e-05,
+      "loss": 0.8651,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452235781559854e-05,
+      "loss": 0.7741,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450766062106482e-05,
+      "loss": 0.8782,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449294429247418e-05,
+      "loss": 0.8134,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447820883280603e-05,
+      "loss": 0.7689,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446345424504375e-05,
+      "loss": 0.8064,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444868053217454e-05,
+      "loss": 0.8148,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443388769718953e-05,
+      "loss": 0.8132,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441907574308372e-05,
+      "loss": 0.8226,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440424467285583e-05,
+      "loss": 0.8264,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438939448950865e-05,
+      "loss": 0.7705,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437452519604877e-05,
+      "loss": 0.2461,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435963679548652e-05,
+      "loss": 0.763,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943447292908363e-05,
+      "loss": 0.8227,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432980268511633e-05,
+      "loss": 0.7555,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431485698134853e-05,
+      "loss": 0.8063,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429989218255893e-05,
+      "loss": 0.7554,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942849082917772e-05,
+      "loss": 0.7753,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942699053120371e-05,
+      "loss": 0.8038,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425488324637608e-05,
+      "loss": 0.7251,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423984209783547e-05,
+      "loss": 0.8035,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422478186946062e-05,
+      "loss": 0.7999,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942097025643005e-05,
+      "loss": 0.8706,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941946041854082e-05,
+      "loss": 0.7871,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417948673584043e-05,
+      "loss": 0.8125,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416435021865795e-05,
+      "loss": 0.8202,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414919463692525e-05,
+      "loss": 0.8625,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413401999371076e-05,
+      "loss": 0.8103,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411882629208678e-05,
+      "loss": 0.7507,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410361353512937e-05,
+      "loss": 0.7496,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940883817259186e-05,
+      "loss": 0.8617,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407313086753818e-05,
+      "loss": 0.8543,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405786096307593e-05,
+      "loss": 0.815,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404257201562327e-05,
+      "loss": 0.8447,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402726402827573e-05,
+      "loss": 0.7512,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401193700413252e-05,
+      "loss": 0.8013,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399659094629675e-05,
+      "loss": 0.9335,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939812258578754e-05,
+      "loss": 0.8143,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939658417419793e-05,
+      "loss": 0.791,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395043860172306e-05,
+      "loss": 0.8409,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393501644022527e-05,
+      "loss": 0.7362,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391957526060833e-05,
+      "loss": 0.7595,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390411506599838e-05,
+      "loss": 0.7817,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388863585952557e-05,
+      "loss": 0.6734,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387313764432383e-05,
+      "loss": 0.7715,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385762042353087e-05,
+      "loss": 0.8259,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384208420028834e-05,
+      "loss": 0.7514,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382652897774173e-05,
+      "loss": 0.8404,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381095475904038e-05,
+      "loss": 0.7946,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379536154733737e-05,
+      "loss": 0.7949,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937797493457898e-05,
+      "loss": 0.7859,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376411815755845e-05,
+      "loss": 0.851,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374846798580805e-05,
+      "loss": 0.6945,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373279883370714e-05,
+      "loss": 0.7905,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371711070442807e-05,
+      "loss": 0.8375,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370140360114708e-05,
+      "loss": 0.7911,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368567752704427e-05,
+      "loss": 0.8259,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366993248530353e-05,
+      "loss": 0.7352,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365416847911263e-05,
+      "loss": 0.7941,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363838551166307e-05,
+      "loss": 0.866,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362258358615035e-05,
+      "loss": 0.8292,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360676270577374e-05,
+      "loss": 0.7625,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935909228737363e-05,
+      "loss": 0.2392,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93575064093245e-05,
+      "loss": 0.8159,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935591863675106e-05,
+      "loss": 0.8134,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935432896997477e-05,
+      "loss": 0.877,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352737409317477e-05,
+      "loss": 0.7235,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351143955101405e-05,
+      "loss": 0.8802,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934954860764917e-05,
+      "loss": 0.788,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347951367283767e-05,
+      "loss": 0.7625,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934635223432857e-05,
+      "loss": 0.8267,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344751209107336e-05,
+      "loss": 0.84,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934314829194422e-05,
+      "loss": 0.802,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341543483163746e-05,
+      "loss": 0.8211,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933993678309082e-05,
+      "loss": 0.8267,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338328192050735e-05,
+      "loss": 0.8054,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336717710369172e-05,
+      "loss": 0.7764,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335105338372187e-05,
+      "loss": 0.8052,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333491076386223e-05,
+      "loss": 0.7461,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93318749247381e-05,
+      "loss": 0.7955,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330256883755027e-05,
+      "loss": 0.8125,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328636953764596e-05,
+      "loss": 0.7848,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327015135094773e-05,
+      "loss": 0.7598,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325391428073916e-05,
+      "loss": 0.9055,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323765833030765e-05,
+      "loss": 0.2321,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322138350294426e-05,
+      "loss": 0.7715,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320508980194415e-05,
+      "loss": 0.7981,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93188777230606e-05,
+      "loss": 0.8229,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317244579223263e-05,
+      "loss": 0.8388,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315609549013038e-05,
+      "loss": 0.8133,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313972632760952e-05,
+      "loss": 0.852,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312333830798427e-05,
+      "loss": 0.8337,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931069314345725e-05,
+      "loss": 0.8648,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309050571069594e-05,
+      "loss": 0.7589,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307406113968016e-05,
+      "loss": 0.7973,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930575977248545e-05,
+      "loss": 0.8097,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930411154695522e-05,
+      "loss": 0.7489,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930246143771102e-05,
+      "loss": 0.7803,
+      "step": 1051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300809445086945e-05,
+      "loss": 0.845,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929915556941744e-05,
+      "loss": 0.7688,
+      "step": 1053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297499811037364e-05,
+      "loss": 0.8736,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295842170281934e-05,
+      "loss": 0.711,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294182647486764e-05,
+      "loss": 0.8036,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292521242987833e-05,
+      "loss": 0.838,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290857957121515e-05,
+      "loss": 0.7459,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289192790224555e-05,
+      "loss": 0.8082,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287525742634093e-05,
+      "loss": 0.8055,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928585681468763e-05,
+      "loss": 0.8028,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928418600672306e-05,
+      "loss": 0.8398,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282513319078656e-05,
+      "loss": 0.8434,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280838752093074e-05,
+      "loss": 0.7569,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279162306105347e-05,
+      "loss": 0.823,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277483981454884e-05,
+      "loss": 0.8895,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275803778481484e-05,
+      "loss": 0.7798,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927412169752532e-05,
+      "loss": 0.7978,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272437738926942e-05,
+      "loss": 0.8202,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270751903027295e-05,
+      "loss": 0.8598,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926906419016769e-05,
+      "loss": 0.8189,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267374600689813e-05,
+      "loss": 0.8016,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265683134935755e-05,
+      "loss": 0.8096,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263989793247958e-05,
+      "loss": 0.7639,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926229457596926e-05,
+      "loss": 0.8636,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260597483442874e-05,
+      "loss": 0.7709,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92588985160124e-05,
+      "loss": 0.8477,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925719767402181e-05,
+      "loss": 0.8126,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925549495781545e-05,
+      "loss": 0.8041,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925379036773806e-05,
+      "loss": 0.7741,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925208390413475e-05,
+      "loss": 0.73,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925037556735101e-05,
+      "loss": 0.7694,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924866535773271e-05,
+      "loss": 0.747,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246953275626102e-05,
+      "loss": 0.826,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245239321377812e-05,
+      "loss": 0.8807,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924352349533485e-05,
+      "loss": 0.825,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241805797844605e-05,
+      "loss": 0.8121,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240086229254838e-05,
+      "loss": 0.8895,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238364789913694e-05,
+      "loss": 0.801,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92366414801697e-05,
+      "loss": 0.8287,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234916300371756e-05,
+      "loss": 0.8174,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923318925086914e-05,
+      "loss": 0.8129,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923146033201152e-05,
+      "loss": 0.7869,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922972954414892e-05,
+      "loss": 0.2463,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227996887631764e-05,
+      "loss": 0.2259,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226262362810844e-05,
+      "loss": 0.8063,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224525970037334e-05,
+      "loss": 0.7517,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922278770966279e-05,
+      "loss": 0.7784,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922104758203913e-05,
+      "loss": 0.8413,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921930558751867e-05,
+      "loss": 0.8529,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217561726454087e-05,
+      "loss": 0.7825,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921581599919845e-05,
+      "loss": 0.7429,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92140684061052e-05,
+      "loss": 0.7893,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212318947528146e-05,
+      "loss": 0.7129,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210567623821495e-05,
+      "loss": 0.831,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208814435339817e-05,
+      "loss": 0.7604,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920705938243806e-05,
+      "loss": 0.7425,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920530246547156e-05,
+      "loss": 0.7846,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920354368479602e-05,
+      "loss": 0.7493,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201783040767516e-05,
+      "loss": 0.7933,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200020533742515e-05,
+      "loss": 0.8364,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198256164077856e-05,
+      "loss": 0.8742,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196489932130754e-05,
+      "loss": 0.7886,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.91947218382588e-05,
+      "loss": 0.8054,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919295188281996e-05,
+      "loss": 0.8676,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191180066172584e-05,
+      "loss": 0.7548,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189406388675392e-05,
+      "loss": 0.7715,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187630850687486e-05,
+      "loss": 0.7874,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918585345256834e-05,
+      "loss": 0.8651,
+      "step": 1119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184074194677804e-05,
+      "loss": 0.7992,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182293077376113e-05,
+      "loss": 0.8547,
+      "step": 1121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918051010102387e-05,
+      "loss": 0.8333,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178725265982055e-05,
+      "loss": 0.7703,
+      "step": 1123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917693857261203e-05,
+      "loss": 0.8103,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917515002127553e-05,
+      "loss": 0.8743,
+      "step": 1125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917335961233466e-05,
+      "loss": 0.8273,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917156734615191e-05,
+      "loss": 0.7828,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916977322309015e-05,
+      "loss": 0.8179,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167977243512608e-05,
+      "loss": 0.7556,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91661794077829e-05,
+      "loss": 0.749,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164379716265028e-05,
+      "loss": 0.8258,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916257816932334e-05,
+      "loss": 0.2225,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160774767322597e-05,
+      "loss": 0.2532,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158969510627906e-05,
+      "loss": 0.7925,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157162399604757e-05,
+      "loss": 0.7599,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155353434619023e-05,
+      "loss": 0.8077,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915354261603695e-05,
+      "loss": 0.8476,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151729944225154e-05,
+      "loss": 0.7783,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914991541955063e-05,
+      "loss": 0.7631,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148099042380746e-05,
+      "loss": 0.7292,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146280813083245e-05,
+      "loss": 0.778,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144460732026253e-05,
+      "loss": 0.7865,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142638799578258e-05,
+      "loss": 0.7858,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914081501610813e-05,
+      "loss": 0.801,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138989381985114e-05,
+      "loss": 0.8612,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137161897578832e-05,
+      "loss": 0.7589,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135332563259272e-05,
+      "loss": 0.8165,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133501379396803e-05,
+      "loss": 0.8124,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913166834636217e-05,
+      "loss": 0.8472,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129833464526487e-05,
+      "loss": 0.7893,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127996734261243e-05,
+      "loss": 0.7771,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126158155938307e-05,
+      "loss": 0.8771,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124317729929916e-05,
+      "loss": 0.8161,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122475456608682e-05,
+      "loss": 0.751,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120631336347597e-05,
+      "loss": 0.8556,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911878536952002e-05,
+      "loss": 0.8336,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116937556499682e-05,
+      "loss": 0.7596,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91150878976607e-05,
+      "loss": 0.8036,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911323639337755e-05,
+      "loss": 0.7769,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911138304402509e-05,
+      "loss": 0.7515,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109527849978553e-05,
+      "loss": 0.8524,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107670811613537e-05,
+      "loss": 0.7862,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105811929306022e-05,
+      "loss": 0.8279,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910395120343236e-05,
+      "loss": 0.7572,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102088634369268e-05,
+      "loss": 0.7991,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100224222493847e-05,
+      "loss": 0.8024,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098357968183567e-05,
+      "loss": 0.8402,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909648987181627e-05,
+      "loss": 0.8507,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909461993377017e-05,
+      "loss": 0.8254,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092748154423853e-05,
+      "loss": 0.7353,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090874534156286e-05,
+      "loss": 0.811,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90889990733468e-05,
+      "loss": 0.8206,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90871217723751e-05,
+      "loss": 0.7355,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085242631621267e-05,
+      "loss": 0.7756,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083361651465756e-05,
+      "loss": 0.776,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081478832289383e-05,
+      "loss": 0.789,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907959417447335e-05,
+      "loss": 0.8189,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077707678399222e-05,
+      "loss": 0.8204,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075819344448946e-05,
+      "loss": 0.7775,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073929173004826e-05,
+      "loss": 0.8464,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072037164449554e-05,
+      "loss": 0.7665,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070143319166182e-05,
+      "loss": 0.7558,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068247637538145e-05,
+      "loss": 0.7478,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066350119949234e-05,
+      "loss": 0.227,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906445076678363e-05,
+      "loss": 0.7737,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062549578425874e-05,
+      "loss": 0.8211,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060646555260876e-05,
+      "loss": 0.7491,
+      "step": 1187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905874169767393e-05,
+      "loss": 0.7779,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905683500605069e-05,
+      "loss": 0.7878,
+      "step": 1189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905492648077719e-05,
+      "loss": 0.8326,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053016122239823e-05,
+      "loss": 0.2247,
+      "step": 1191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905110393082537e-05,
+      "loss": 0.8345,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904918990692097e-05,
+      "loss": 0.8254,
+      "step": 1193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904727405091413e-05,
+      "loss": 0.8293,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045356363192747e-05,
+      "loss": 0.7928,
+      "step": 1195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904343684414507e-05,
+      "loss": 0.8306,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041515494159724e-05,
+      "loss": 0.8582,
+      "step": 1197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039592313625714e-05,
+      "loss": 0.7553,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037667302932405e-05,
+      "loss": 0.8094,
+      "step": 1199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035740462469527e-05,
+      "loss": 0.8381,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90338117926272e-05,
+      "loss": 0.7004,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031881293795896e-05,
+      "loss": 0.7871,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029948966366466e-05,
+      "loss": 0.8188,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902801481073013e-05,
+      "loss": 0.7651,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902607882727848e-05,
+      "loss": 0.8142,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024141016403476e-05,
+      "loss": 0.8532,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022201378497447e-05,
+      "loss": 0.8162,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020259913953093e-05,
+      "loss": 0.7497,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901831662316348e-05,
+      "loss": 0.8042,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901637150652205e-05,
+      "loss": 0.7985,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014424564422615e-05,
+      "loss": 0.8133,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901247579725935e-05,
+      "loss": 0.7813,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010525205426803e-05,
+      "loss": 0.8037,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008572789319896e-05,
+      "loss": 0.8141,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900661854933391e-05,
+      "loss": 0.7981,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004662485864503e-05,
+      "loss": 0.7773,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90027045993077e-05,
+      "loss": 0.773,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90007448900599e-05,
+      "loss": 0.8037,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998783358517864e-05,
+      "loss": 0.7909,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996820005078724e-05,
+      "loss": 0.7661,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994854830139977e-05,
+      "loss": 0.7552,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89928878340995e-05,
+      "loss": 0.8507,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990919017355526e-05,
+      "loss": 0.7927,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988948380306665e-05,
+      "loss": 0.7971,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986975923351894e-05,
+      "loss": 0.8558,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898500164689056e-05,
+      "loss": 0.8599,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898302555132237e-05,
+      "loss": 0.2618,
+      "step": 1227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981047637047407e-05,
+      "loss": 0.8148,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979067904466117e-05,
+      "loss": 0.6802,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977086353979324e-05,
+      "loss": 0.7999,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897510298598821e-05,
+      "loss": 0.8049,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897311780089433e-05,
+      "loss": 0.7748,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89711307990996e-05,
+      "loss": 0.7522,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896914198100632e-05,
+      "loss": 0.8387,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967151347017133e-05,
+      "loss": 0.78,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896515889753508e-05,
+      "loss": 0.8452,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963164632963537e-05,
+      "loss": 0.787,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961168553706268e-05,
+      "loss": 0.8028,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959170660167404e-05,
+      "loss": 0.7513,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895717095275144e-05,
+      "loss": 0.815,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955169431863238e-05,
+      "loss": 0.8251,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953166097908017e-05,
+      "loss": 0.8104,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951160951291383e-05,
+      "loss": 0.2187,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949153992419295e-05,
+      "loss": 0.7824,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947145221698084e-05,
+      "loss": 0.7688,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894513463953444e-05,
+      "loss": 0.7999,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943122246335438e-05,
+      "loss": 0.8856,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89411080425085e-05,
+      "loss": 0.8662,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939092028461423e-05,
+      "loss": 0.8459,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893707420460237e-05,
+      "loss": 0.7691,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935054571339876e-05,
+      "loss": 0.8026,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933033129082824e-05,
+      "loss": 0.8343,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893100987824049e-05,
+      "loss": 0.7413,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928984819222495e-05,
+      "loss": 0.7442,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926957952438834e-05,
+      "loss": 0.8153,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892492927829987e-05,
+      "loss": 0.8608,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922898797216323e-05,
+      "loss": 0.7572,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892086650959929e-05,
+      "loss": 0.8139,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918832415860228e-05,
+      "loss": 0.841,
+      "step": 1259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916796516410964e-05,
+      "loss": 0.773,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914758811663676e-05,
+      "loss": 0.7947,
+      "step": 1261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912719302030927e-05,
+      "loss": 0.8093,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910677987925635e-05,
+      "loss": 0.7505,
+      "step": 1263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908634869761088e-05,
+      "loss": 0.725,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906589947950932e-05,
+      "loss": 0.7741,
+      "step": 1265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904543222909188e-05,
+      "loss": 0.6615,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890249469505023e-05,
+      "loss": 0.7786,
+      "step": 1267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900444364788806e-05,
+      "loss": 0.8051,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898392232540028e-05,
+      "loss": 0.8172,
+      "step": 1269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896338298719374e-05,
+      "loss": 0.7812,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889428256374268e-05,
+      "loss": 0.7978,
+      "step": 1271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889222502802615e-05,
+      "loss": 0.7855,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890165691986356e-05,
+      "loss": 0.8194,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888104556040234e-05,
+      "loss": 0.8208,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886041620605083e-05,
+      "loss": 0.8072,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888397688609856e-05,
+      "loss": 0.7227,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881910352938694e-05,
+      "loss": 0.7883,
+      "step": 1277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879842021543876e-05,
+      "loss": 0.7224,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887777189233286e-05,
+      "loss": 0.7336,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887569996572477e-05,
+      "loss": 0.743,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873626242139085e-05,
+      "loss": 0.8229,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871550721995655e-05,
+      "loss": 0.808,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869473405714686e-05,
+      "loss": 0.8105,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867394293716757e-05,
+      "loss": 0.8273,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865313386422803e-05,
+      "loss": 0.7461,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863230684254124e-05,
+      "loss": 0.7767,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886114618763239e-05,
+      "loss": 0.8828,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859059896979626e-05,
+      "loss": 0.7234,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885697181271822e-05,
+      "loss": 0.745,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854881935270927e-05,
+      "loss": 0.7708,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885279026506087e-05,
+      "loss": 0.795,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850696802511525e-05,
+      "loss": 0.7729,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884860154804674e-05,
+      "loss": 0.8226,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846504502090713e-05,
+      "loss": 0.8352,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844405665068016e-05,
+      "loss": 0.8369,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884230503740358e-05,
+      "loss": 0.8011,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88402026195227e-05,
+      "loss": 0.8391,
+      "step": 1297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838098411851037e-05,
+      "loss": 0.7647,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835992414814605e-05,
+      "loss": 0.8344,
+      "step": 1299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833884628839783e-05,
+      "loss": 0.8174,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831775054353315e-05,
+      "loss": 0.7794,
+      "step": 1301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882966369178231e-05,
+      "loss": 0.8267,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882755054155423e-05,
+      "loss": 0.8661,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882543560409691e-05,
+      "loss": 0.824,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823318879838536e-05,
+      "loss": 0.7704,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882120036920766e-05,
+      "loss": 0.8115,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819080072633203e-05,
+      "loss": 0.8592,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816957990544437e-05,
+      "loss": 0.8198,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814834123370997e-05,
+      "loss": 0.8859,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812708471542888e-05,
+      "loss": 0.9075,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810581035490464e-05,
+      "loss": 0.7827,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880845181564445e-05,
+      "loss": 0.7552,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806320812435932e-05,
+      "loss": 0.6878,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804188026296347e-05,
+      "loss": 0.8012,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802053457657504e-05,
+      "loss": 0.7685,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799917106951566e-05,
+      "loss": 0.866,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879777897461106e-05,
+      "loss": 0.8048,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795639061068873e-05,
+      "loss": 0.8165,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793497366758257e-05,
+      "loss": 0.7576,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791353892112812e-05,
+      "loss": 0.747,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789208637566515e-05,
+      "loss": 0.7727,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787061603553687e-05,
+      "loss": 0.8268,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784912790509026e-05,
+      "loss": 0.8481,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782762198867576e-05,
+      "loss": 0.7982,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780609829064748e-05,
+      "loss": 0.8155,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778455681536313e-05,
+      "loss": 0.8209,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776299756718397e-05,
+      "loss": 0.8175,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774142055047487e-05,
+      "loss": 0.847,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771982576960443e-05,
+      "loss": 0.7407,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769821322894467e-05,
+      "loss": 0.8419,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767658293287125e-05,
+      "loss": 0.8504,
+      "step": 1331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876549348857635e-05,
+      "loss": 0.8812,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876332690920042e-05,
+      "loss": 0.7855,
+      "step": 1333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876115855559799e-05,
+      "loss": 0.7395,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758988428208065e-05,
+      "loss": 0.7943,
+      "step": 1335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756816527470012e-05,
+      "loss": 0.8066,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875464285382355e-05,
+      "loss": 0.7373,
+      "step": 1337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875246740770876e-05,
+      "loss": 0.816,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750290189566087e-05,
+      "loss": 0.7137,
+      "step": 1339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874811119983633e-05,
+      "loss": 0.7659,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745930438960654e-05,
+      "loss": 0.8023,
+      "step": 1341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874374790738057e-05,
+      "loss": 0.2654,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741563605537954e-05,
+      "loss": 0.8136,
+      "step": 1343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739377533875044e-05,
+      "loss": 0.8582,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737189692834432e-05,
+      "loss": 0.7835,
+      "step": 1345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873500008285907e-05,
+      "loss": 0.2263,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732808704392263e-05,
+      "loss": 0.7864,
+      "step": 1347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730615557877682e-05,
+      "loss": 0.8493,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728420643759355e-05,
+      "loss": 0.7734,
+      "step": 1349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726223962481658e-05,
+      "loss": 0.8456,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724025514489338e-05,
+      "loss": 0.8243,
+      "step": 1351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872182530022749e-05,
+      "loss": 0.8132,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871962332014157e-05,
+      "loss": 0.7129,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717419574677394e-05,
+      "loss": 0.7986,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715214064281135e-05,
+      "loss": 0.8303,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713006789399316e-05,
+      "loss": 0.7593,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710797750478827e-05,
+      "loss": 0.8042,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870858694796691e-05,
+      "loss": 0.7754,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706374382311163e-05,
+      "loss": 0.2221,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704160053959543e-05,
+      "loss": 0.7463,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870194396336037e-05,
+      "loss": 0.7826,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699726110962303e-05,
+      "loss": 0.861,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869750649721438e-05,
+      "loss": 0.777,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869528512256598e-05,
+      "loss": 0.8394,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869306198746684e-05,
+      "loss": 0.8574,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690837092367064e-05,
+      "loss": 0.8169,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688610437717103e-05,
+      "loss": 0.8002,
+      "step": 1367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868638202396776e-05,
+      "loss": 0.7397,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684151851570214e-05,
+      "loss": 0.767,
+      "step": 1369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681919920975976e-05,
+      "loss": 0.82,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679686232636923e-05,
+      "loss": 0.869,
+      "step": 1371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677450787005293e-05,
+      "loss": 0.7895,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867521358453367e-05,
+      "loss": 0.8059,
+      "step": 1373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672974625675003e-05,
+      "loss": 0.7987,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670733910882593e-05,
+      "loss": 0.7512,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668491440610098e-05,
+      "loss": 0.8253,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666247215311523e-05,
+      "loss": 0.8138,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664001235441236e-05,
+      "loss": 0.791,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866175350145396e-05,
+      "loss": 0.7728,
+      "step": 1379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659504013804778e-05,
+      "loss": 0.749,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657252772949112e-05,
+      "loss": 0.7502,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654999779342754e-05,
+      "loss": 0.7615,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865274503344185e-05,
+      "loss": 0.8006,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650488535702885e-05,
+      "loss": 0.8212,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648230286582726e-05,
+      "loss": 0.7222,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864597028653857e-05,
+      "loss": 0.8064,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643708536027976e-05,
+      "loss": 0.7717,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641445035508862e-05,
+      "loss": 0.7743,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639179785439495e-05,
+      "loss": 0.762,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86369127862785e-05,
+      "loss": 0.81,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634644038484854e-05,
+      "loss": 0.7305,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863237354251789e-05,
+      "loss": 0.7562,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630101298837295e-05,
+      "loss": 0.7922,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627827307903102e-05,
+      "loss": 0.7401,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862555157017571e-05,
+      "loss": 0.8221,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623274086115865e-05,
+      "loss": 0.8749,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862099485618466e-05,
+      "loss": 0.7233,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618713880843562e-05,
+      "loss": 0.7754,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861643116055437e-05,
+      "loss": 0.824,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614146695779237e-05,
+      "loss": 0.7741,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861186048698069e-05,
+      "loss": 0.7824,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860957253462159e-05,
+      "loss": 0.7703,
+      "step": 1402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860728283916516e-05,
+      "loss": 0.8474,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604991401074967e-05,
+      "loss": 0.7815,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860269822081494e-05,
+      "loss": 0.7767,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600403298849355e-05,
+      "loss": 0.7264,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598106635642845e-05,
+      "loss": 0.8133,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859580823166039e-05,
+      "loss": 0.7566,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593508087367335e-05,
+      "loss": 0.8777,
+      "step": 1409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591206203229353e-05,
+      "loss": 0.796,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588902579712496e-05,
+      "loss": 0.8129,
+      "step": 1411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586597217283153e-05,
+      "loss": 0.7697,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858429011640807e-05,
+      "loss": 0.7566,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858198127755434e-05,
+      "loss": 0.7042,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579670701189417e-05,
+      "loss": 0.8367,
+      "step": 1415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577358387781093e-05,
+      "loss": 0.787,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575044337797527e-05,
+      "loss": 0.2355,
+      "step": 1417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857272855170722e-05,
+      "loss": 0.9187,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570411029979028e-05,
+      "loss": 0.7665,
+      "step": 1419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568091773082154e-05,
+      "loss": 0.7957,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565770781486162e-05,
+      "loss": 0.7657,
+      "step": 1421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563448055660955e-05,
+      "loss": 0.7514,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856112359607679e-05,
+      "loss": 0.8253,
+      "step": 1423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558797403204286e-05,
+      "loss": 0.7887,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556469477514403e-05,
+      "loss": 0.8763,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554139819478453e-05,
+      "loss": 0.7143,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551808429568096e-05,
+      "loss": 0.8451,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549475308255348e-05,
+      "loss": 0.8236,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547140456012578e-05,
+      "loss": 0.2094,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544803873312497e-05,
+      "loss": 0.7951,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854246556062817e-05,
+      "loss": 0.2168,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854012551843301e-05,
+      "loss": 0.7671,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853778374720079e-05,
+      "loss": 0.8683,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535440247405623e-05,
+      "loss": 0.7406,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533095019521975e-05,
+      "loss": 0.8152,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853074806402466e-05,
+      "loss": 0.6948,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528399381388844e-05,
+      "loss": 0.8251,
+      "step": 1437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852604897209004e-05,
+      "loss": 0.81,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523696836604118e-05,
+      "loss": 0.8031,
+      "step": 1439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852134297540729e-05,
+      "loss": 0.7448,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851898738897612e-05,
+      "loss": 0.7496,
+      "step": 1441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851663007778752e-05,
+      "loss": 0.8035,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851427104231875e-05,
+      "loss": 0.7719,
+      "step": 1443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851191028304743e-05,
+      "loss": 0.7677,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509547800451507e-05,
+      "loss": 0.2418,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507183595009297e-05,
+      "loss": 0.756,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850481766719946e-05,
+      "loss": 0.8076,
+      "step": 1447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502450017501e-05,
+      "loss": 0.7659,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500080646393273e-05,
+      "loss": 0.7606,
+      "step": 1449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849770955435598e-05,
+      "loss": 0.8304,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849533674186918e-05,
+      "loss": 0.7719,
+      "step": 1451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492962209413267e-05,
+      "loss": 0.7862,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490585957468992e-05,
+      "loss": 0.8666,
+      "step": 1453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488207986517454e-05,
+      "loss": 0.8442,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848582829704009e-05,
+      "loss": 0.7425,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483446889518703e-05,
+      "loss": 0.7625,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848106376443543e-05,
+      "loss": 0.849,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478678922272756e-05,
+      "loss": 0.7608,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476292363513523e-05,
+      "loss": 0.7671,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847390408864091e-05,
+      "loss": 0.7735,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847151409813845e-05,
+      "loss": 0.8419,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846912239249002e-05,
+      "loss": 0.7759,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466728972179843e-05,
+      "loss": 0.8139,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464333837692497e-05,
+      "loss": 0.8027,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461936989512903e-05,
+      "loss": 0.7949,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845953842812632e-05,
+      "loss": 0.7946,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457138154018373e-05,
+      "loss": 0.8139,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845473616767501e-05,
+      "loss": 0.7793,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452332469582547e-05,
+      "loss": 0.7937,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449927060227636e-05,
+      "loss": 0.8035,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844751994009727e-05,
+      "loss": 0.8293,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445111109678807e-05,
+      "loss": 0.8112,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442700569459934e-05,
+      "loss": 0.7771,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440288319928682e-05,
+      "loss": 0.7145,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437874361573453e-05,
+      "loss": 0.7773,
+      "step": 1475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435458694882965e-05,
+      "loss": 0.7962,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433041320346295e-05,
+      "loss": 0.8398,
+      "step": 1477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430622238452873e-05,
+      "loss": 0.7957,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428201449692462e-05,
+      "loss": 0.8013,
+      "step": 1479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842577895455518e-05,
+      "loss": 0.7578,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842335475353148e-05,
+      "loss": 0.8034,
+      "step": 1481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420928847112167e-05,
+      "loss": 0.7492,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418501235788398e-05,
+      "loss": 0.7502,
+      "step": 1483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416071920051665e-05,
+      "loss": 0.8526,
+      "step": 1484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413640900393806e-05,
+      "loss": 0.7637,
+      "step": 1485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411208177307004e-05,
+      "loss": 0.8461,
+      "step": 1486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408773751283796e-05,
+      "loss": 0.8824,
+      "step": 1487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406337622817053e-05,
+      "loss": 0.7183,
+      "step": 1488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403899792399995e-05,
+      "loss": 0.7625,
+      "step": 1489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401460260526186e-05,
+      "loss": 0.7243,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399019027689534e-05,
+      "loss": 0.7959,
+      "step": 1491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396576094384292e-05,
+      "loss": 0.8137,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394131461105057e-05,
+      "loss": 0.9005,
+      "step": 1493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839168512834677e-05,
+      "loss": 0.8518,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389237096604722e-05,
+      "loss": 0.7737,
+      "step": 1495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386787366374533e-05,
+      "loss": 0.71,
+      "step": 1496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384335938152183e-05,
+      "loss": 0.7826,
+      "step": 1497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381882812433988e-05,
+      "loss": 0.7697,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379427989716605e-05,
+      "loss": 0.8285,
+      "step": 1499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376971470497045e-05,
+      "loss": 0.7104,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837451325527265e-05,
+      "loss": 0.7642,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372053344541112e-05,
+      "loss": 0.8237,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836959173880047e-05,
+      "loss": 0.7807,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367128438549095e-05,
+      "loss": 0.8154,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364663444285714e-05,
+      "loss": 0.7778,
+      "step": 1505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836219675650939e-05,
+      "loss": 0.7958,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359728375719522e-05,
+      "loss": 0.782,
+      "step": 1507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357258302415868e-05,
+      "loss": 0.2399,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354786537098514e-05,
+      "loss": 0.7876,
+      "step": 1509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352313080267898e-05,
+      "loss": 0.7596,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349837932424798e-05,
+      "loss": 0.7194,
+      "step": 1511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347361094070334e-05,
+      "loss": 0.75,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834488256570596e-05,
+      "loss": 0.7848,
+      "step": 1513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342402347833485e-05,
+      "loss": 0.8255,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339920440955058e-05,
+      "loss": 0.7372,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833743684557316e-05,
+      "loss": 0.7684,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833495156219063e-05,
+      "loss": 0.7382,
+      "step": 1517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332464591310627e-05,
+      "loss": 0.7874,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329975933436675e-05,
+      "loss": 0.842,
+      "step": 1519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327485589072623e-05,
+      "loss": 0.8207,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324993558722674e-05,
+      "loss": 0.7627,
+      "step": 1521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322499842891353e-05,
+      "loss": 0.738,
+      "step": 1522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320004442083546e-05,
+      "loss": 0.8056,
+      "step": 1523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317507356804475e-05,
+      "loss": 0.8304,
+      "step": 1524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315008587559696e-05,
+      "loss": 0.8115,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312508134855116e-05,
+      "loss": 0.7009,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831000599919697e-05,
+      "loss": 0.7649,
+      "step": 1527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307502181091852e-05,
+      "loss": 0.7821,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304996681046678e-05,
+      "loss": 0.7159,
+      "step": 1529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302489499568712e-05,
+      "loss": 0.809,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299980637165562e-05,
+      "loss": 0.7927,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297470094345173e-05,
+      "loss": 0.7563,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829495787161583e-05,
+      "loss": 0.8303,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829244396948616e-05,
+      "loss": 0.7594,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289928388465124e-05,
+      "loss": 0.743,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287411129062035e-05,
+      "loss": 0.7442,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828489219178653e-05,
+      "loss": 0.779,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82823715771486e-05,
+      "loss": 0.8605,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279849285658566e-05,
+      "loss": 0.827,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277325317827098e-05,
+      "loss": 0.7439,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827479967416519e-05,
+      "loss": 0.8362,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272272355184193e-05,
+      "loss": 0.8107,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269743361395785e-05,
+      "loss": 0.7936,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267212693311985e-05,
+      "loss": 0.7734,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826468035144516e-05,
+      "loss": 0.7249,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262146336308003e-05,
+      "loss": 0.7682,
+      "step": 1546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259610648413555e-05,
+      "loss": 0.7325,
+      "step": 1547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257073288275186e-05,
+      "loss": 0.8023,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254534256406623e-05,
+      "loss": 0.8255,
+      "step": 1549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251993553321906e-05,
+      "loss": 0.7194,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249451179535434e-05,
+      "loss": 0.8239,
+      "step": 1551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824690713556194e-05,
+      "loss": 0.6989,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244361421916484e-05,
+      "loss": 0.7178,
+      "step": 1553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241814039114478e-05,
+      "loss": 0.7453,
+      "step": 1554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239264987671663e-05,
+      "loss": 0.7299,
+      "step": 1555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823671426810412e-05,
+      "loss": 0.7855,
+      "step": 1556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234161880928277e-05,
+      "loss": 0.7639,
+      "step": 1557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823160782666088e-05,
+      "loss": 0.8077,
+      "step": 1558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822905210581903e-05,
+      "loss": 0.7636,
+      "step": 1559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822649471892016e-05,
+      "loss": 0.8137,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223935666482033e-05,
+      "loss": 0.7085,
+      "step": 1561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221374949022762e-05,
+      "loss": 0.7424,
+      "step": 1562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821881256706079e-05,
+      "loss": 0.7419,
+      "step": 1563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216248521114898e-05,
+      "loss": 0.7676,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82136828117042e-05,
+      "loss": 0.8368,
+      "step": 1565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211115439348153e-05,
+      "loss": 0.7772,
+      "step": 1566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820854640456655e-05,
+      "loss": 0.8585,
+      "step": 1567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205975707879518e-05,
+      "loss": 0.7629,
+      "step": 1568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820340334980752e-05,
+      "loss": 0.6841,
+      "step": 1569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200829330871354e-05,
+      "loss": 0.7476,
+      "step": 1570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198253651592166e-05,
+      "loss": 0.799,
+      "step": 1571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195676312491417e-05,
+      "loss": 0.2256,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193097314090925e-05,
+      "loss": 0.7421,
+      "step": 1573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819051665691283e-05,
+      "loss": 0.8379,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187934341479617e-05,
+      "loss": 0.7649,
+      "step": 1575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185350368314097e-05,
+      "loss": 0.7984,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182764737939423e-05,
+      "loss": 0.7674,
+      "step": 1577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818017745087909e-05,
+      "loss": 0.7369,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817758850765692e-05,
+      "loss": 0.7152,
+      "step": 1579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817499790879706e-05,
+      "loss": 0.7793,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172405654824013e-05,
+      "loss": 0.7804,
+      "step": 1581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816981174626261e-05,
+      "loss": 0.7955,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167216183638005e-05,
+      "loss": 0.7717,
+      "step": 1583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164618967475708e-05,
+      "loss": 0.7519,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162020098301543e-05,
+      "loss": 0.7848,
+      "step": 1585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815941957664168e-05,
+      "loss": 0.7265,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815681740302263e-05,
+      "loss": 0.8275,
+      "step": 1587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815421357797122e-05,
+      "loss": 0.7596,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815160810201462e-05,
+      "loss": 0.758,
+      "step": 1589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149000975680347e-05,
+      "loss": 0.8132,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146392199496233e-05,
+      "loss": 0.7495,
+      "step": 1591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143781773990456e-05,
+      "loss": 0.8262,
+      "step": 1592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141169699691516e-05,
+      "loss": 0.7976,
+      "step": 1593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813855597712826e-05,
+      "loss": 0.7469,
+      "step": 1594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135940606829867e-05,
+      "loss": 0.8499,
+      "step": 1595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133323589325834e-05,
+      "loss": 0.7906,
+      "step": 1596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130704925146014e-05,
+      "loss": 0.7708,
+      "step": 1597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812808461482058e-05,
+      "loss": 0.7732,
+      "step": 1598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125462658880045e-05,
+      "loss": 0.2461,
+      "step": 1599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812283905785524e-05,
+      "loss": 0.8002,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812021381227735e-05,
+      "loss": 0.8169,
+      "step": 1601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117586922677884e-05,
+      "loss": 0.7395,
+      "step": 1602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114958389588674e-05,
+      "loss": 0.7889,
+      "step": 1603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81123282135419e-05,
+      "loss": 0.7877,
+      "step": 1604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810969639507007e-05,
+      "loss": 0.7272,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810706293470602e-05,
+      "loss": 0.8089,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104427832982923e-05,
+      "loss": 0.2283,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810179109043428e-05,
+      "loss": 0.8257,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809915270759393e-05,
+      "loss": 0.8515,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096512684996037e-05,
+      "loss": 0.7554,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093871023175106e-05,
+      "loss": 0.7837,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091227722665964e-05,
+      "loss": 0.7876,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088582784003778e-05,
+      "loss": 0.834,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085936207724043e-05,
+      "loss": 0.8077,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083287994362586e-05,
+      "loss": 0.8542,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080638144455565e-05,
+      "loss": 0.8054,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077986658539467e-05,
+      "loss": 0.6962,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8075333537151117e-05,
+      "loss": 0.7608,
+      "step": 1618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072678780827667e-05,
+      "loss": 0.818,
+      "step": 1619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070022390106592e-05,
+      "loss": 0.7761,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067364365525715e-05,
+      "loss": 0.7579,
+      "step": 1621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064704707623177e-05,
+      "loss": 0.7446,
+      "step": 1622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062043416937455e-05,
+      "loss": 0.7716,
+      "step": 1623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059380494007352e-05,
+      "loss": 0.816,
+      "step": 1624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056715939372007e-05,
+      "loss": 0.8546,
+      "step": 1625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805404975357089e-05,
+      "loss": 0.8587,
+      "step": 1626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051381937143784e-05,
+      "loss": 0.2508,
+      "step": 1627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048712490630836e-05,
+      "loss": 0.7883,
+      "step": 1628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046041414572485e-05,
+      "loss": 0.7058,
+      "step": 1629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804336870950953e-05,
+      "loss": 0.7534,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040694375983087e-05,
+      "loss": 0.7323,
+      "step": 1631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038018414534596e-05,
+      "loss": 0.2406,
+      "step": 1632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803534082570584e-05,
+      "loss": 0.8159,
+      "step": 1633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032661610038917e-05,
+      "loss": 0.774,
+      "step": 1634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029980768076267e-05,
+      "loss": 0.8161,
+      "step": 1635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027298300360655e-05,
+      "loss": 0.8901,
+      "step": 1636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024614207435176e-05,
+      "loss": 0.8419,
+      "step": 1637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802192848984325e-05,
+      "loss": 0.6807,
+      "step": 1638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019241148128624e-05,
+      "loss": 0.7622,
+      "step": 1639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016552182835384e-05,
+      "loss": 0.7708,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801386159450794e-05,
+      "loss": 0.7821,
+      "step": 1641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011169383691024e-05,
+      "loss": 0.7895,
+      "step": 1642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008475550929704e-05,
+      "loss": 0.7436,
+      "step": 1643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800578009676938e-05,
+      "loss": 0.7446,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003083021755764e-05,
+      "loss": 0.7513,
+      "step": 1645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000384326434915e-05,
+      "loss": 0.8222,
+      "step": 1646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997684011353208e-05,
+      "loss": 0.8219,
+      "step": 1647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994982077057356e-05,
+      "loss": 0.8329,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992278524094386e-05,
+      "loss": 0.7603,
+      "step": 1649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798957335301166e-05,
+      "loss": 0.8692,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798686656435688e-05,
+      "loss": 0.8246,
+      "step": 1651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984158158678048e-05,
+      "loss": 0.8094,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981448136523522e-05,
+      "loss": 0.8184,
+      "step": 1653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978736498441965e-05,
+      "loss": 0.7972,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976023244982377e-05,
+      "loss": 0.7885,
+      "step": 1655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973308376694086e-05,
+      "loss": 0.8275,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970591894126748e-05,
+      "loss": 0.7907,
+      "step": 1657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967873797830342e-05,
+      "loss": 0.9018,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965154088355172e-05,
+      "loss": 0.7397,
+      "step": 1659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962432766251873e-05,
+      "loss": 0.7808,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795970983207141e-05,
+      "loss": 0.7726,
+      "step": 1661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956985286365055e-05,
+      "loss": 0.7681,
+      "step": 1662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954259129684434e-05,
+      "loss": 0.8516,
+      "step": 1663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951531362581483e-05,
+      "loss": 0.7649,
+      "step": 1664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948801985608465e-05,
+      "loss": 0.843,
+      "step": 1665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794607099931797e-05,
+      "loss": 0.7784,
+      "step": 1666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943338404262915e-05,
+      "loss": 0.7069,
+      "step": 1667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940604200996544e-05,
+      "loss": 0.7921,
+      "step": 1668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937868390072424e-05,
+      "loss": 0.7645,
+      "step": 1669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793513097204445e-05,
+      "loss": 0.8652,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932391947466836e-05,
+      "loss": 0.782,
+      "step": 1671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929651316894126e-05,
+      "loss": 0.7991,
+      "step": 1672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926909080881195e-05,
+      "loss": 0.7373,
+      "step": 1673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924165239983233e-05,
+      "loss": 0.7612,
+      "step": 1674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792141979475576e-05,
+      "loss": 0.7891,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918672745754622e-05,
+      "loss": 0.7698,
+      "step": 1676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915924093535983e-05,
+      "loss": 0.833,
+      "step": 1677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791317383865634e-05,
+      "loss": 0.7651,
+      "step": 1678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910421981672508e-05,
+      "loss": 0.8107,
+      "step": 1679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790766852314163e-05,
+      "loss": 0.8174,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790491346362117e-05,
+      "loss": 0.7284,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902156803668925e-05,
+      "loss": 0.8098,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899398543843004e-05,
+      "loss": 0.8873,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789663868470185e-05,
+      "loss": 0.8409,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893877226804217e-05,
+      "loss": 0.7138,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891114170709198e-05,
+      "loss": 0.789,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78883495169762e-05,
+      "loss": 0.796,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885583266164956e-05,
+      "loss": 0.7462,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788281541883552e-05,
+      "loss": 0.7836,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7880045975548274e-05,
+      "loss": 0.8085,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877274936863923e-05,
+      "loss": 0.8244,
+      "step": 1691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874502303343486e-05,
+      "loss": 0.8268,
+      "step": 1692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871728075548316e-05,
+      "loss": 0.7853,
+      "step": 1693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868952254040085e-05,
+      "loss": 0.2163,
+      "step": 1694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866174839380785e-05,
+      "loss": 0.7426,
+      "step": 1695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786339583213273e-05,
+      "loss": 0.7036,
+      "step": 1696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860615232858564e-05,
+      "loss": 0.8198,
+      "step": 1697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785783304212124e-05,
+      "loss": 0.743,
+      "step": 1698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855049260484053e-05,
+      "loss": 0.772,
+      "step": 1699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78522638885106e-05,
+      "loss": 0.7269,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784947692676481e-05,
+      "loss": 0.7577,
+      "step": 1701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846688375810935e-05,
+      "loss": 0.7577,
+      "step": 1702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843898236213543e-05,
+      "loss": 0.7069,
+      "step": 1703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841106508537532e-05,
+      "loss": 0.8313,
+      "step": 1704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783831319334811e-05,
+      "loss": 0.7855,
+      "step": 1705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835518291210813e-05,
+      "loss": 0.223,
+      "step": 1706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78327218026915e-05,
+      "loss": 0.7701,
+      "step": 1707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782992372835635e-05,
+      "loss": 0.8158,
+      "step": 1708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827124068771865e-05,
+      "loss": 0.7966,
+      "step": 1709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824322824504857e-05,
+      "loss": 0.7652,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821519996122473e-05,
+      "loss": 0.7105,
+      "step": 1711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818715584192173e-05,
+      "loss": 0.7841,
+      "step": 1712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815909589281737e-05,
+      "loss": 0.8128,
+      "step": 1713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781310201195927e-05,
+      "loss": 0.7708,
+      "step": 1714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78102928527932e-05,
+      "loss": 0.8358,
+      "step": 1715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807482112352262e-05,
+      "loss": 0.7954,
+      "step": 1716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804669791205525e-05,
+      "loss": 0.2612,
+      "step": 1717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801855889922372e-05,
+      "loss": 0.6658,
+      "step": 1718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799040409072505e-05,
+      "loss": 0.8068,
+      "step": 1719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796223349225945e-05,
+      "loss": 0.7923,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779340471095304e-05,
+      "loss": 0.7474,
+      "step": 1721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790584494824452e-05,
+      "loss": 0.7567,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778776270141116e-05,
+      "loss": 0.8059,
+      "step": 1723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778493933128447e-05,
+      "loss": 0.7847,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782114385015992e-05,
+      "loss": 0.7734,
+      "step": 1725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779287863177675e-05,
+      "loss": 0.75,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777645976634178e-05,
+      "loss": 0.9068,
+      "step": 1727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777363009508088e-05,
+      "loss": 0.7698,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770798849967866e-05,
+      "loss": 0.8677,
+      "step": 1729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767966031575966e-05,
+      "loss": 0.8124,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77651316404787e-05,
+      "loss": 0.7709,
+      "step": 1731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776229567724993e-05,
+      "loss": 0.7512,
+      "step": 1732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775945814246382e-05,
+      "loss": 0.2454,
+      "step": 1733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756619036694864e-05,
+      "loss": 0.7531,
+      "step": 1734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753778360517862e-05,
+      "loss": 0.8267,
+      "step": 1735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750936114507943e-05,
+      "loss": 0.7912,
+      "step": 1736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748092299240543e-05,
+      "loss": 0.7527,
+      "step": 1737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745246915291433e-05,
+      "loss": 0.8075,
+      "step": 1738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774239996323668e-05,
+      "loss": 0.7655,
+      "step": 1739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739551443652684e-05,
+      "loss": 0.8137,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773670135711616e-05,
+      "loss": 0.7785,
+      "step": 1741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733849704204128e-05,
+      "loss": 0.8179,
+      "step": 1742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730996485493945e-05,
+      "loss": 0.7374,
+      "step": 1743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728141701563268e-05,
+      "loss": 0.7449,
+      "step": 1744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772528535299008e-05,
+      "loss": 0.7983,
+      "step": 1745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722427440352682e-05,
+      "loss": 0.7674,
+      "step": 1746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771956796422968e-05,
+      "loss": 0.784,
+      "step": 1747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716706925200013e-05,
+      "loss": 0.7935,
+      "step": 1748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713844323842922e-05,
+      "loss": 0.7183,
+      "step": 1749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710980160737975e-05,
+      "loss": 0.7468,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770811443646505e-05,
+      "loss": 0.7963,
+      "step": 1751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770524715160434e-05,
+      "loss": 0.8008,
+      "step": 1752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702378306736358e-05,
+      "loss": 0.8341,
+      "step": 1753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769950790244193e-05,
+      "loss": 0.7927,
+      "step": 1754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696635939302206e-05,
+      "loss": 0.8105,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693762417898635e-05,
+      "loss": 0.6789,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690887338812997e-05,
+      "loss": 0.7932,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688010702627384e-05,
+      "loss": 0.7672,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7685132509924194e-05,
+      "loss": 0.7916,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682252761286152e-05,
+      "loss": 0.7227,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767937145729629e-05,
+      "loss": 0.7843,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676488598537963e-05,
+      "loss": 0.753,
+      "step": 1762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767360418559483e-05,
+      "loss": 0.7032,
+      "step": 1763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670718219050868e-05,
+      "loss": 0.8431,
+      "step": 1764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766783069949038e-05,
+      "loss": 0.8608,
+      "step": 1765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664941627497965e-05,
+      "loss": 0.7889,
+      "step": 1766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662051003658556e-05,
+      "loss": 0.8582,
+      "step": 1767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659158828557384e-05,
+      "loss": 0.7849,
+      "step": 1768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656265102779998e-05,
+      "loss": 0.2587,
+      "step": 1769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653369826912264e-05,
+      "loss": 0.8219,
+      "step": 1770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765047300154036e-05,
+      "loss": 0.7515,
+      "step": 1771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647574627250783e-05,
+      "loss": 0.8058,
+      "step": 1772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644674704630333e-05,
+      "loss": 0.7959,
+      "step": 1773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641773234266136e-05,
+      "loss": 0.7698,
+      "step": 1774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638870216745617e-05,
+      "loss": 0.7614,
+      "step": 1775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635965652656527e-05,
+      "loss": 0.7,
+      "step": 1776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633059542586924e-05,
+      "loss": 0.7711,
+      "step": 1777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763015188712518e-05,
+      "loss": 0.7701,
+      "step": 1778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627242686859984e-05,
+      "loss": 0.8245,
+      "step": 1779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624331942380327e-05,
+      "loss": 0.7548,
+      "step": 1780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762141965427552e-05,
+      "loss": 0.823,
+      "step": 1781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761850582313519e-05,
+      "loss": 0.7955,
+      "step": 1782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615590449549273e-05,
+      "loss": 0.7172,
+      "step": 1783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612673534108014e-05,
+      "loss": 0.7176,
+      "step": 1784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760975507740197e-05,
+      "loss": 0.8277,
+      "step": 1785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760683508002202e-05,
+      "loss": 0.7822,
+      "step": 1786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603913542559343e-05,
+      "loss": 0.7224,
+      "step": 1787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600990465605435e-05,
+      "loss": 0.8101,
+      "step": 1788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598065849752104e-05,
+      "loss": 0.8588,
+      "step": 1789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595139695591466e-05,
+      "loss": 0.74,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592212003715957e-05,
+      "loss": 0.7452,
+      "step": 1791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589282774718314e-05,
+      "loss": 0.7705,
+      "step": 1792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586352009191595e-05,
+      "loss": 0.8238,
+      "step": 1793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583419707729157e-05,
+      "loss": 0.7446,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580485870924678e-05,
+      "loss": 0.7954,
+      "step": 1795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577550499372146e-05,
+      "loss": 0.6762,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757461359366585e-05,
+      "loss": 0.7985,
+      "step": 1797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757167515440041e-05,
+      "loss": 0.7939,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756873518217073e-05,
+      "loss": 0.7445,
+      "step": 1799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565793677572047e-05,
+      "loss": 0.2432,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562850641199893e-05,
+      "loss": 0.8015,
+      "step": 1801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559906073650122e-05,
+      "loss": 0.7455,
+      "step": 1802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556959975518886e-05,
+      "loss": 0.7824,
+      "step": 1803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755401234740266e-05,
+      "loss": 0.8048,
+      "step": 1804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551063189898218e-05,
+      "loss": 0.8157,
+      "step": 1805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548112503602648e-05,
+      "loss": 0.8532,
+      "step": 1806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545160289113344e-05,
+      "loss": 0.7835,
+      "step": 1807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754220654702802e-05,
+      "loss": 0.7288,
+      "step": 1808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539251277944686e-05,
+      "loss": 0.743,
+      "step": 1809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536294482461666e-05,
+      "loss": 0.7347,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533336161177595e-05,
+      "loss": 0.7043,
+      "step": 1811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753037631469142e-05,
+      "loss": 0.73,
+      "step": 1812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752741494360239e-05,
+      "loss": 0.7638,
+      "step": 1813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524452048510065e-05,
+      "loss": 0.7415,
+      "step": 1814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752148763001431e-05,
+      "loss": 0.8363,
+      "step": 1815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518521688715313e-05,
+      "loss": 0.7583,
+      "step": 1816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515554225213548e-05,
+      "loss": 0.7565,
+      "step": 1817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512585240109813e-05,
+      "loss": 0.7823,
+      "step": 1818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509614734005214e-05,
+      "loss": 0.7071,
+      "step": 1819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506642707501153e-05,
+      "loss": 0.824,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503669161199357e-05,
+      "loss": 0.2276,
+      "step": 1821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750069409570184e-05,
+      "loss": 0.8545,
+      "step": 1822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497717511610946e-05,
+      "loss": 0.7752,
+      "step": 1823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494739409529307e-05,
+      "loss": 0.7829,
+      "step": 1824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491759790059877e-05,
+      "loss": 0.746,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748877865380591e-05,
+      "loss": 0.7862,
+      "step": 1826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485796001370957e-05,
+      "loss": 0.8362,
+      "step": 1827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482811833358903e-05,
+      "loss": 0.7443,
+      "step": 1828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479826150373914e-05,
+      "loss": 0.7752,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476838953020475e-05,
+      "loss": 0.796,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747385024190338e-05,
+      "loss": 0.2209,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470860017627714e-05,
+      "loss": 0.7756,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467868280798884e-05,
+      "loss": 0.8104,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464875032022607e-05,
+      "loss": 0.7203,
+      "step": 1834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746188027190488e-05,
+      "loss": 0.7287,
+      "step": 1835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458884001052038e-05,
+      "loss": 0.805,
+      "step": 1836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74558862200707e-05,
+      "loss": 0.7014,
+      "step": 1837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74528869295678e-05,
+      "loss": 0.7819,
+      "step": 1838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744988613015058e-05,
+      "loss": 0.8241,
+      "step": 1839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446883822426566e-05,
+      "loss": 0.7674,
+      "step": 1840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443880007003627e-05,
+      "loss": 0.7908,
+      "step": 1841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440874684489906e-05,
+      "loss": 0.8329,
+      "step": 1842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437867855493865e-05,
+      "loss": 0.7774,
+      "step": 1843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743485952062427e-05,
+      "loss": 0.8512,
+      "step": 1844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743184968049018e-05,
+      "loss": 0.2139,
+      "step": 1845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742883833570098e-05,
+      "loss": 0.7881,
+      "step": 1846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425825486866336e-05,
+      "loss": 0.7274,
+      "step": 1847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742281113459624e-05,
+      "loss": 0.753,
+      "step": 1848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419795279500976e-05,
+      "loss": 0.746,
+      "step": 1849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416777922191136e-05,
+      "loss": 0.6869,
+      "step": 1850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413759063277614e-05,
+      "loss": 0.7682,
+      "step": 1851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741073870337161e-05,
+      "loss": 0.8229,
+      "step": 1852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407716843084625e-05,
+      "loss": 0.7189,
+      "step": 1853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404693483028474e-05,
+      "loss": 0.7833,
+      "step": 1854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401668623815255e-05,
+      "loss": 0.6918,
+      "step": 1855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739864226605739e-05,
+      "loss": 0.8075,
+      "step": 1856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395614410367597e-05,
+      "loss": 0.7414,
+      "step": 1857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392585057358892e-05,
+      "loss": 0.7771,
+      "step": 1858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389554207644606e-05,
+      "loss": 0.7412,
+      "step": 1859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386521861838363e-05,
+      "loss": 0.7797,
+      "step": 1860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7383488020554086e-05,
+      "loss": 0.6972,
+      "step": 1861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380452684406016e-05,
+      "loss": 0.7931,
+      "step": 1862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737741585400869e-05,
+      "loss": 0.7667,
+      "step": 1863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374377529976936e-05,
+      "loss": 0.7204,
+      "step": 1864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371337712925906e-05,
+      "loss": 0.7652,
+      "step": 1865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368296403471032e-05,
+      "loss": 0.8189,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736525360222806e-05,
+      "loss": 0.8102,
+      "step": 1867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362209309813047e-05,
+      "loss": 0.8363,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735916352684233e-05,
+      "loss": 0.8379,
+      "step": 1869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356116253932568e-05,
+      "loss": 0.7072,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353067491700706e-05,
+      "loss": 0.7439,
+      "step": 1871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350017240764004e-05,
+      "loss": 0.7437,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346965501740006e-05,
+      "loss": 0.7351,
+      "step": 1873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343912275246583e-05,
+      "loss": 0.7709,
+      "step": 1874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734085756190188e-05,
+      "loss": 0.7956,
+      "step": 1875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337801362324365e-05,
+      "loss": 0.8174,
+      "step": 1876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733474367713279e-05,
+      "loss": 0.8033,
+      "step": 1877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733168450694622e-05,
+      "loss": 0.8378,
+      "step": 1878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328623852384012e-05,
+      "loss": 0.2283,
+      "step": 1879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325561714065826e-05,
+      "loss": 0.8333,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732249809261163e-05,
+      "loss": 0.8011,
+      "step": 1881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319432988641684e-05,
+      "loss": 0.8168,
+      "step": 1882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316366402776546e-05,
+      "loss": 0.767,
+      "step": 1883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313298335637083e-05,
+      "loss": 0.7273,
+      "step": 1884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310228787844456e-05,
+      "loss": 0.7601,
+      "step": 1885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730715776002012e-05,
+      "loss": 0.2308,
+      "step": 1886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304085252785848e-05,
+      "loss": 0.8376,
+      "step": 1887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301011266763695e-05,
+      "loss": 0.8031,
+      "step": 1888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297935802576018e-05,
+      "loss": 0.2338,
+      "step": 1889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729485886084548e-05,
+      "loss": 0.8211,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291780442195042e-05,
+      "loss": 0.7916,
+      "step": 1891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728870054724796e-05,
+      "loss": 0.7742,
+      "step": 1892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285619176627794e-05,
+      "loss": 0.7675,
+      "step": 1893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728253633095839e-05,
+      "loss": 0.7949,
+      "step": 1894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279452010863914e-05,
+      "loss": 0.8325,
+      "step": 1895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276366216968807e-05,
+      "loss": 0.7262,
+      "step": 1896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273278949897827e-05,
+      "loss": 0.7577,
+      "step": 1897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270190210276023e-05,
+      "loss": 0.8769,
+      "step": 1898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267099998728742e-05,
+      "loss": 0.8155,
+      "step": 1899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726400831588163e-05,
+      "loss": 0.7165,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260915162360624e-05,
+      "loss": 0.7995,
+      "step": 1901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257820538791973e-05,
+      "loss": 0.7627,
+      "step": 1902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254724445802217e-05,
+      "loss": 0.1989,
+      "step": 1903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725162688401818e-05,
+      "loss": 0.7714,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724852785406701e-05,
+      "loss": 0.7599,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245427356576127e-05,
+      "loss": 0.7272,
+      "step": 1906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242325392173267e-05,
+      "loss": 0.7508,
+      "step": 1907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723922196148645e-05,
+      "loss": 0.8128,
+      "step": 1908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236117065144e-05,
+      "loss": 0.7597,
+      "step": 1909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233010703774532e-05,
+      "loss": 0.7743,
+      "step": 1910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229902878006965e-05,
+      "loss": 0.8434,
+      "step": 1911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226793588470506e-05,
+      "loss": 0.7755,
+      "step": 1912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223682835794666e-05,
+      "loss": 0.8078,
+      "step": 1913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220570620609254e-05,
+      "loss": 0.2301,
+      "step": 1914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721745694354436e-05,
+      "loss": 0.7937,
+      "step": 1915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721434180523039e-05,
+      "loss": 0.7098,
+      "step": 1916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721122520629803e-05,
+      "loss": 0.7276,
+      "step": 1917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720810714737827e-05,
+      "loss": 0.767,
+      "step": 1918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720498762910239e-05,
+      "loss": 0.8067,
+      "step": 1919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201866652101978e-05,
+      "loss": 0.7817,
+      "step": 1920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198744217008895e-05,
+      "loss": 0.7593,
+      "step": 1921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195620324455323e-05,
+      "loss": 0.2146,
+      "step": 1922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192494975073716e-05,
+      "loss": 0.7272,
+      "step": 1923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189368169496843e-05,
+      "loss": 0.7808,
+      "step": 1924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718623990835775e-05,
+      "loss": 0.7294,
+      "step": 1925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718311019228979e-05,
+      "loss": 0.754,
+      "step": 1926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717997902192661e-05,
+      "loss": 0.7228,
+      "step": 1927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717684639790214e-05,
+      "loss": 0.8354,
+      "step": 1928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173712320850616e-05,
+      "loss": 0.7906,
+      "step": 1929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170576791406565e-05,
+      "loss": 0.7334,
+      "step": 1930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716743981020481e-05,
+      "loss": 0.8738,
+      "step": 1931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716430137788046e-05,
+      "loss": 0.762,
+      "step": 1932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161161495068925e-05,
+      "loss": 0.7792,
+      "step": 1933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715802016240591e-05,
+      "loss": 0.7586,
+      "step": 1934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154877380527405e-05,
+      "loss": 0.8239,
+      "step": 1935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151733150069706e-05,
+      "loss": 0.7726,
+      "step": 1936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148587471669387e-05,
+      "loss": 0.8694,
+      "step": 1937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145440345963326e-05,
+      "loss": 0.771,
+      "step": 1938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142291773588698e-05,
+      "loss": 0.8193,
+      "step": 1939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139141755182956e-05,
+      "loss": 0.7648,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135990291383856e-05,
+      "loss": 0.228,
+      "step": 1941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132837382829443e-05,
+      "loss": 0.6744,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129683030158062e-05,
+      "loss": 0.7549,
+      "step": 1943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712652723400834e-05,
+      "loss": 0.7643,
+      "step": 1944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71233699950192e-05,
+      "loss": 0.7553,
+      "step": 1945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120211313829857e-05,
+      "loss": 0.7838,
+      "step": 1946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117051191079827e-05,
+      "loss": 0.7518,
+      "step": 1947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113889627408897e-05,
+      "loss": 0.7076,
+      "step": 1948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110726623457167e-05,
+      "loss": 0.8165,
+      "step": 1949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710756217986502e-05,
+      "loss": 0.8153,
+      "step": 1950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104396297273125e-05,
+      "loss": 0.77,
+      "step": 1951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101228976322452e-05,
+      "loss": 0.8716,
+      "step": 1952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709806021765426e-05,
+      "loss": 0.8111,
+      "step": 1953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094890021910093e-05,
+      "loss": 0.7694,
+      "step": 1954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091718389731792e-05,
+      "loss": 0.7256,
+      "step": 1955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708854532176148e-05,
+      "loss": 0.7993,
+      "step": 1956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708537081864159e-05,
+      "loss": 0.8548,
+      "step": 1957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082194881014826e-05,
+      "loss": 0.765,
+      "step": 1958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707901750952419e-05,
+      "loss": 0.7233,
+      "step": 1959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075838704812975e-05,
+      "loss": 0.7872,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707265846752476e-05,
+      "loss": 0.7358,
+      "step": 1961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706947679830342e-05,
+      "loss": 0.7225,
+      "step": 1962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706629369779312e-05,
+      "loss": 0.6899,
+      "step": 1963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70631091666383e-05,
+      "loss": 0.7645,
+      "step": 1964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059923205483714e-05,
+      "loss": 0.8356,
+      "step": 1965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056735814974388e-05,
+      "loss": 0.6769,
+      "step": 1966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705354699575564e-05,
+      "loss": 0.7462,
+      "step": 1967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050356748473086e-05,
+      "loss": 0.7901,
+      "step": 1968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047165073772622e-05,
+      "loss": 0.6971,
+      "step": 1969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043971972300432e-05,
+      "loss": 0.7389,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040777444702992e-05,
+      "loss": 0.8231,
+      "step": 1971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037581491627073e-05,
+      "loss": 0.8375,
+      "step": 1972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703438411371973e-05,
+      "loss": 0.7142,
+      "step": 1973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031185311628293e-05,
+      "loss": 0.8319,
+      "step": 1974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702798508600041e-05,
+      "loss": 0.7856,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024783437483986e-05,
+      "loss": 0.7936,
+      "step": 1976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021580366727233e-05,
+      "loss": 0.7274,
+      "step": 1977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018375874378645e-05,
+      "loss": 0.8153,
+      "step": 1978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701516996108701e-05,
+      "loss": 0.7713,
+      "step": 1979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701196262750139e-05,
+      "loss": 0.7558,
+      "step": 1980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700875387427115e-05,
+      "loss": 0.8398,
+      "step": 1981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700554370204593e-05,
+      "loss": 0.7468,
+      "step": 1982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002332111475667e-05,
+      "loss": 0.766,
+      "step": 1983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999119103210578e-05,
+      "loss": 0.8103,
+      "step": 1984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699590467790117e-05,
+      "loss": 0.7726,
+      "step": 1985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992688836198242e-05,
+      "loss": 0.8083,
+      "step": 1986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989471578752867e-05,
+      "loss": 0.7405,
+      "step": 1987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986252906216416e-05,
+      "loss": 0.7064,
+      "step": 1988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983032819240543e-05,
+      "loss": 0.7661,
+      "step": 1989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979811318477185e-05,
+      "loss": 0.793,
+      "step": 1990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976588404578572e-05,
+      "loss": 0.7737,
+      "step": 1991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973364078197215e-05,
+      "loss": 0.6945,
+      "step": 1992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970138339985912e-05,
+      "loss": 0.777,
+      "step": 1993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966911190597746e-05,
+      "loss": 0.7785,
+      "step": 1994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963682630686088e-05,
+      "loss": 0.7926,
+      "step": 1995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960452660904596e-05,
+      "loss": 0.7227,
+      "step": 1996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957221281907207e-05,
+      "loss": 0.7931,
+      "step": 1997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953988494348148e-05,
+      "loss": 0.794,
+      "step": 1998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950754298881933e-05,
+      "loss": 0.6976,
+      "step": 1999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947518696163358e-05,
+      "loss": 0.7501,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69442816868475e-05,
+      "loss": 0.7227,
+      "step": 2001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941043271589727e-05,
+      "loss": 0.2198,
+      "step": 2002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937803451045692e-05,
+      "loss": 0.2894,
+      "step": 2003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934562225871332e-05,
+      "loss": 0.7998,
+      "step": 2004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693131959672286e-05,
+      "loss": 0.8369,
+      "step": 2005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928075564256788e-05,
+      "loss": 0.7638,
+      "step": 2006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924830129129898e-05,
+      "loss": 0.7722,
+      "step": 2007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692158329199926e-05,
+      "loss": 0.8055,
+      "step": 2008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691833505352224e-05,
+      "loss": 0.8204,
+      "step": 2009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915085414356465e-05,
+      "loss": 0.8007,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691183437515987e-05,
+      "loss": 0.7615,
+      "step": 2011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908581936590653e-05,
+      "loss": 0.7906,
+      "step": 2012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690532809930731e-05,
+      "loss": 0.8,
+      "step": 2013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690207286396861e-05,
+      "loss": 0.8349,
+      "step": 2014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898816231233613e-05,
+      "loss": 0.8203,
+      "step": 2015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895558201761652e-05,
+      "loss": 0.7404,
+      "step": 2016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892298776212352e-05,
+      "loss": 0.7532,
+      "step": 2017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688903795524562e-05,
+      "loss": 0.7442,
+      "step": 2018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885775739521642e-05,
+      "loss": 0.8012,
+      "step": 2019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882512129700882e-05,
+      "loss": 0.802,
+      "step": 2020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68792471264441e-05,
+      "loss": 0.7799,
+      "step": 2021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875980730412327e-05,
+      "loss": 0.7556,
+      "step": 2022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872712942266875e-05,
+      "loss": 0.7116,
+      "step": 2023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6869443762669346e-05,
+      "loss": 0.7996,
+      "step": 2024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866173192281616e-05,
+      "loss": 0.7873,
+      "step": 2025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686290123176585e-05,
+      "loss": 0.8112,
+      "step": 2026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859627881784484e-05,
+      "loss": 0.8046,
+      "step": 2027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685635314300025e-05,
+      "loss": 0.8406,
+      "step": 2028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853077016076147e-05,
+      "loss": 0.7621,
+      "step": 2029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849799501675463e-05,
+      "loss": 0.7048,
+      "step": 2030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846520600461765e-05,
+      "loss": 0.758,
+      "step": 2031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843240313098896e-05,
+      "loss": 0.2348,
+      "step": 2032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683995864025099e-05,
+      "loss": 0.7271,
+      "step": 2033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836675582582457e-05,
+      "loss": 0.8262,
+      "step": 2034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833391140757983e-05,
+      "loss": 0.6829,
+      "step": 2035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830105315442534e-05,
+      "loss": 0.7659,
+      "step": 2036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682681810730136e-05,
+      "loss": 0.8117,
+      "step": 2037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823529517000002e-05,
+      "loss": 0.818,
+      "step": 2038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820239545204254e-05,
+      "loss": 0.7978,
+      "step": 2039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681694819258021e-05,
+      "loss": 0.2386,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813655459794246e-05,
+      "loss": 0.7969,
+      "step": 2041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810361347512998e-05,
+      "loss": 0.764,
+      "step": 2042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807065856403403e-05,
+      "loss": 0.765,
+      "step": 2043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680376898713266e-05,
+      "loss": 0.7236,
+      "step": 2044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680047074036826e-05,
+      "loss": 0.762,
+      "step": 2045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797171116777962e-05,
+      "loss": 0.7311,
+      "step": 2046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793870117029816e-05,
+      "loss": 0.8093,
+      "step": 2047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679056774179214e-05,
+      "loss": 0.8148,
+      "step": 2048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787263991733534e-05,
+      "loss": 0.7016,
+      "step": 2049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783958867522873e-05,
+      "loss": 0.8589,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678065236982932e-05,
+      "loss": 0.7573,
+      "step": 2051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777344499322307e-05,
+      "loss": 0.742,
+      "step": 2052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774035256671544e-05,
+      "loss": 0.7975,
+      "step": 2053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770724642547028e-05,
+      "loss": 0.754,
+      "step": 2054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767412657619024e-05,
+      "loss": 0.7847,
+      "step": 2055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764099302558073e-05,
+      "loss": 0.8286,
+      "step": 2056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760784578035002e-05,
+      "loss": 0.7695,
+      "step": 2057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757468484720912e-05,
+      "loss": 0.6967,
+      "step": 2058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675415102328718e-05,
+      "loss": 0.7799,
+      "step": 2059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750832194405466e-05,
+      "loss": 0.7632,
+      "step": 2060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674751199874769e-05,
+      "loss": 0.7105,
+      "step": 2061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674419043698607e-05,
+      "loss": 0.8853,
+      "step": 2062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740867509793085e-05,
+      "loss": 0.7714,
+      "step": 2063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737543217841497e-05,
+      "loss": 0.7185,
+      "step": 2064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734217561804346e-05,
+      "loss": 0.7716,
+      "step": 2065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673089054235494e-05,
+      "loss": 0.7361,
+      "step": 2066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727562160166878e-05,
+      "loss": 0.8026,
+      "step": 2067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672423241591402e-05,
+      "loss": 0.7301,
+      "step": 2068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67209013102705e-05,
+      "loss": 0.8166,
+      "step": 2069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671756884391075e-05,
+      "loss": 0.8333,
+      "step": 2070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671423501750945e-05,
+      "loss": 0.7807,
+      "step": 2071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710899831741573e-05,
+      "loss": 0.7031,
+      "step": 2072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707563287282362e-05,
+      "loss": 0.8182,
+      "step": 2073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704225384807336e-05,
+      "loss": 0.7285,
+      "step": 2074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700886124992284e-05,
+      "loss": 0.7648,
+      "step": 2075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669754550851328e-05,
+      "loss": 0.711,
+      "step": 2076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669420353604666e-05,
+      "loss": 0.7867,
+      "step": 2077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669086020826904e-05,
+      "loss": 0.8615,
+      "step": 2078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687515525857322e-05,
+      "loss": 0.797,
+      "step": 2079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668416948948866e-05,
+      "loss": 0.6994,
+      "step": 2080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66808220998405e-05,
+      "loss": 0.8032,
+      "step": 2081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677473357590558e-05,
+      "loss": 0.2445,
+      "step": 2082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674123263416814e-05,
+      "loss": 0.7675,
+      "step": 2083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670771817997536e-05,
+      "loss": 0.8168,
+      "step": 2084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666741902201125e-05,
+      "loss": 0.8468,
+      "step": 2085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666406487613678e-05,
+      "loss": 0.7703,
+      "step": 2086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666070938105319e-05,
+      "loss": 0.7253,
+      "step": 2087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657352537439846e-05,
+      "loss": 0.7456,
+      "step": 2088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653994345976378e-05,
+      "loss": 0.7892,
+      "step": 2089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650634807342677e-05,
+      "loss": 0.828,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664727392221892e-05,
+      "loss": 0.8048,
+      "step": 2091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643911691285555e-05,
+      "loss": 0.7366,
+      "step": 2092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66405481152233e-05,
+      "loss": 0.7613,
+      "step": 2093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637183194713146e-05,
+      "loss": 0.7214,
+      "step": 2094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663381693043636e-05,
+      "loss": 0.2369,
+      "step": 2095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630449323074463e-05,
+      "loss": 0.8238,
+      "step": 2096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627080373309277e-05,
+      "loss": 0.7376,
+      "step": 2097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623710081822877e-05,
+      "loss": 0.7957,
+      "step": 2098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620338449297613e-05,
+      "loss": 0.7845,
+      "step": 2099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616965476416105e-05,
+      "loss": 0.8071,
+      "step": 2100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661359116386125e-05,
+      "loss": 0.7496,
+      "step": 2101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661021551231621e-05,
+      "loss": 0.708,
+      "step": 2102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606838522464415e-05,
+      "loss": 0.7892,
+      "step": 2103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660346019498958e-05,
+      "loss": 0.6799,
+      "step": 2104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600080530575684e-05,
+      "loss": 0.7804,
+      "step": 2105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596699529906967e-05,
+      "loss": 0.8181,
+      "step": 2106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593317193667953e-05,
+      "loss": 0.745,
+      "step": 2107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589933522543436e-05,
+      "loss": 0.7281,
+      "step": 2108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586548517218463e-05,
+      "loss": 0.2405,
+      "step": 2109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583162178378373e-05,
+      "loss": 0.7919,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657977450670876e-05,
+      "loss": 0.7792,
+      "step": 2111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576385502895497e-05,
+      "loss": 0.7795,
+      "step": 2112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657299516762472e-05,
+      "loss": 0.7949,
+      "step": 2113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569603501582838e-05,
+      "loss": 0.8366,
+      "step": 2114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566210505456535e-05,
+      "loss": 0.755,
+      "step": 2115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656281617993275e-05,
+      "loss": 0.7406,
+      "step": 2116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65594205256987e-05,
+      "loss": 0.7377,
+      "step": 2117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655602354344188e-05,
+      "loss": 0.7978,
+      "step": 2118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655262523385003e-05,
+      "loss": 0.6907,
+      "step": 2119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6549225597611186e-05,
+      "loss": 0.7514,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545824635413633e-05,
+      "loss": 0.856,
+      "step": 2121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542422347945935e-05,
+      "loss": 0.24,
+      "step": 2122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539018735896917e-05,
+      "loss": 0.802,
+      "step": 2123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535613799955673e-05,
+      "loss": 0.7045,
+      "step": 2124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532207540811573e-05,
+      "loss": 0.7705,
+      "step": 2125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652879995915425e-05,
+      "loss": 0.7252,
+      "step": 2126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65253910556736e-05,
+      "loss": 0.7829,
+      "step": 2127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652198083105979e-05,
+      "loss": 0.8095,
+      "step": 2128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651856928600326e-05,
+      "loss": 0.7685,
+      "step": 2129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515156421194713e-05,
+      "loss": 0.8717,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511742237325115e-05,
+      "loss": 0.8251,
+      "step": 2131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508326735085705e-05,
+      "loss": 0.7876,
+      "step": 2132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504909915167986e-05,
+      "loss": 0.7516,
+      "step": 2133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650149177826373e-05,
+      "loss": 0.8032,
+      "step": 2134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498072325064974e-05,
+      "loss": 0.8462,
+      "step": 2135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494651556264022e-05,
+      "loss": 0.7639,
+      "step": 2136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491229472553442e-05,
+      "loss": 0.7188,
+      "step": 2137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487806074626074e-05,
+      "loss": 0.7113,
+      "step": 2138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484381363175014e-05,
+      "loss": 0.7487,
+      "step": 2139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480955338893637e-05,
+      "loss": 0.7738,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477528002475575e-05,
+      "loss": 0.743,
+      "step": 2141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647409935461473e-05,
+      "loss": 0.7366,
+      "step": 2142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470669396005264e-05,
+      "loss": 0.7674,
+      "step": 2143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467238127341606e-05,
+      "loss": 0.7366,
+      "step": 2144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463805549318458e-05,
+      "loss": 0.703,
+      "step": 2145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460371662630774e-05,
+      "loss": 0.7937,
+      "step": 2146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456936467973787e-05,
+      "loss": 0.735,
+      "step": 2147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645349996604299e-05,
+      "loss": 0.7799,
+      "step": 2148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450062157534125e-05,
+      "loss": 0.8114,
+      "step": 2149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446623043143222e-05,
+      "loss": 0.7918,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443182623566563e-05,
+      "loss": 0.7952,
+      "step": 2151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64397408995007e-05,
+      "loss": 0.8024,
+      "step": 2152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436297871642438e-05,
+      "loss": 0.7849,
+      "step": 2153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432853540688863e-05,
+      "loss": 0.7598,
+      "step": 2154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642940790733731e-05,
+      "loss": 0.8279,
+      "step": 2155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425960972285385e-05,
+      "loss": 0.759,
+      "step": 2156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422512736230958e-05,
+      "loss": 0.7357,
+      "step": 2157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641906319987215e-05,
+      "loss": 0.7973,
+      "step": 2158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415612363907372e-05,
+      "loss": 0.7723,
+      "step": 2159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412160229035265e-05,
+      "loss": 0.7751,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640870679595476e-05,
+      "loss": 0.7243,
+      "step": 2161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405252065365037e-05,
+      "loss": 0.8005,
+      "step": 2162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401796037965547e-05,
+      "loss": 0.7955,
+      "step": 2163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398338714455986e-05,
+      "loss": 0.721,
+      "step": 2164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639488009553634e-05,
+      "loss": 0.2472,
+      "step": 2165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391420181906832e-05,
+      "loss": 0.7391,
+      "step": 2166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638795897426796e-05,
+      "loss": 0.743,
+      "step": 2167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384496473320487e-05,
+      "loss": 0.7529,
+      "step": 2168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381032679765426e-05,
+      "loss": 0.7518,
+      "step": 2169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377567594304056e-05,
+      "loss": 0.767,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374101217637925e-05,
+      "loss": 0.7682,
+      "step": 2171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370633550468836e-05,
+      "loss": 0.8045,
+      "step": 2172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367164593498852e-05,
+      "loss": 0.7772,
+      "step": 2173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63636943474303e-05,
+      "loss": 0.8224,
+      "step": 2174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636022281296577e-05,
+      "loss": 0.6837,
+      "step": 2175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356749990808106e-05,
+      "loss": 0.8127,
+      "step": 2176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635327588166042e-05,
+      "loss": 0.8061,
+      "step": 2177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634980048622608e-05,
+      "loss": 0.806,
+      "step": 2178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346323805208716e-05,
+      "loss": 0.7493,
+      "step": 2179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6342845839312217e-05,
+      "loss": 0.7503,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339366589240738e-05,
+      "loss": 0.7365,
+      "step": 2181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335886055698687e-05,
+      "loss": 0.77,
+      "step": 2182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633240423939073e-05,
+      "loss": 0.7066,
+      "step": 2183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63289211410218e-05,
+      "loss": 0.7203,
+      "step": 2184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632543676129709e-05,
+      "loss": 0.7628,
+      "step": 2185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632195110092204e-05,
+      "loss": 0.7957,
+      "step": 2186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318464160602364e-05,
+      "loss": 0.6812,
+      "step": 2187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314975941044037e-05,
+      "loss": 0.7708,
+      "step": 2188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631148644295327e-05,
+      "loss": 0.7377,
+      "step": 2189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307995667036552e-05,
+      "loss": 0.2325,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630450361400064e-05,
+      "loss": 0.7482,
+      "step": 2191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630101028455252e-05,
+      "loss": 0.8142,
+      "step": 2192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297515679399457e-05,
+      "loss": 0.8115,
+      "step": 2193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294019799248975e-05,
+      "loss": 0.7745,
+      "step": 2194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629052264480885e-05,
+      "loss": 0.7822,
+      "step": 2195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287024216787113e-05,
+      "loss": 0.7943,
+      "step": 2196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283524515892057e-05,
+      "loss": 0.7298,
+      "step": 2197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628002354283224e-05,
+      "loss": 0.7374,
+      "step": 2198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276521298316466e-05,
+      "loss": 0.7283,
+      "step": 2199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62730177830538e-05,
+      "loss": 0.7353,
+      "step": 2200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269512997753567e-05,
+      "loss": 0.7754,
+      "step": 2201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626600694312534e-05,
+      "loss": 0.7584,
+      "step": 2202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262499619878968e-05,
+      "loss": 0.7299,
+      "step": 2203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258991028724537e-05,
+      "loss": 0.7574,
+      "step": 2204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62554811703724e-05,
+      "loss": 0.7364,
+      "step": 2205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625197004553316e-05,
+      "loss": 0.7926,
+      "step": 2206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624845765491769e-05,
+      "loss": 0.8084,
+      "step": 2207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244943999237102e-05,
+      "loss": 0.8212,
+      "step": 2208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241429079202774e-05,
+      "loss": 0.7646,
+      "step": 2209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623791289552634e-05,
+      "loss": 0.2469,
+      "step": 2210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234395448919688e-05,
+      "loss": 0.7863,
+      "step": 2211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230876740094958e-05,
+      "loss": 0.8184,
+      "step": 2212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622735676976455e-05,
+      "loss": 0.7844,
+      "step": 2213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622383553864112e-05,
+      "loss": 0.8369,
+      "step": 2214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220313047437582e-05,
+      "loss": 0.7285,
+      "step": 2215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621678929686709e-05,
+      "loss": 0.8214,
+      "step": 2216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213264287643076e-05,
+      "loss": 0.8,
+      "step": 2217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620973802047921e-05,
+      "loss": 0.7958,
+      "step": 2218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206210496089414e-05,
+      "loss": 0.7567,
+      "step": 2219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620268171518788e-05,
+      "loss": 0.7233,
+      "step": 2220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619915167848905e-05,
+      "loss": 0.7469,
+      "step": 2221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195620386707606e-05,
+      "loss": 0.8401,
+      "step": 2222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61920878405585e-05,
+      "loss": 0.7569,
+      "step": 2223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188554040756935e-05,
+      "loss": 0.7965,
+      "step": 2224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185018988018364e-05,
+      "loss": 0.8368,
+      "step": 2225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618148268305849e-05,
+      "loss": 0.7814,
+      "step": 2226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617794512659328e-05,
+      "loss": 0.806,
+      "step": 2227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617440631933895e-05,
+      "loss": 0.7908,
+      "step": 2228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170866262011965e-05,
+      "loss": 0.7731,
+      "step": 2229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616732495532905e-05,
+      "loss": 0.757,
+      "step": 2230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163782400007173e-05,
+      "loss": 0.7281,
+      "step": 2231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160238596763567e-05,
+      "loss": 0.7691,
+      "step": 2232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156693546315707e-05,
+      "loss": 0.7566,
+      "step": 2233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153147249381328e-05,
+      "loss": 0.7677,
+      "step": 2234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614959970667842e-05,
+      "loss": 0.7841,
+      "step": 2235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614605091892521e-05,
+      "loss": 0.733,
+      "step": 2236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142500886840192e-05,
+      "loss": 0.7482,
+      "step": 2237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138949611142104e-05,
+      "loss": 0.7776,
+      "step": 2238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135397092549944e-05,
+      "loss": 0.7397,
+      "step": 2239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613184333178295e-05,
+      "loss": 0.6954,
+      "step": 2240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128288329560625e-05,
+      "loss": 0.804,
+      "step": 2241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124732086602708e-05,
+      "loss": 0.7906,
+      "step": 2242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121174603629203e-05,
+      "loss": 0.7466,
+      "step": 2243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117615881360354e-05,
+      "loss": 0.6311,
+      "step": 2244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611405592051667e-05,
+      "loss": 0.7913,
+      "step": 2245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110494721818894e-05,
+      "loss": 0.7166,
+      "step": 2246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610693228598803e-05,
+      "loss": 0.2418,
+      "step": 2247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103368613745333e-05,
+      "loss": 0.7554,
+      "step": 2248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099803705812304e-05,
+      "loss": 0.7564,
+      "step": 2249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609623756291069e-05,
+      "loss": 0.7981,
+      "step": 2250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60926701857625e-05,
+      "loss": 0.7092,
+      "step": 2251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089101575089988e-05,
+      "loss": 0.7863,
+      "step": 2252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608553173161565e-05,
+      "loss": 0.7243,
+      "step": 2253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608196065606224e-05,
+      "loss": 0.8249,
+      "step": 2254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078388349152763e-05,
+      "loss": 0.7407,
+      "step": 2255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074814811610465e-05,
+      "loss": 0.8171,
+      "step": 2256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607124004415885e-05,
+      "loss": 0.7715,
+      "step": 2257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067664047521666e-05,
+      "loss": 0.7889,
+      "step": 2258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064086822422906e-05,
+      "loss": 0.7273,
+      "step": 2259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606050836958682e-05,
+      "loss": 0.8113,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6056928689737898e-05,
+      "loss": 0.7658,
+      "step": 2261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053347783600892e-05,
+      "loss": 0.7518,
+      "step": 2262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604976565190079e-05,
+      "loss": 0.8046,
+      "step": 2263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046182295362828e-05,
+      "loss": 0.7411,
+      "step": 2264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60425977147125e-05,
+      "loss": 0.7436,
+      "step": 2265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039011910675537e-05,
+      "loss": 0.275,
+      "step": 2266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035424883977924e-05,
+      "loss": 0.7227,
+      "step": 2267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603183663534589e-05,
+      "loss": 0.8287,
+      "step": 2268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602824716550592e-05,
+      "loss": 0.7965,
+      "step": 2269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024656475184727e-05,
+      "loss": 0.72,
+      "step": 2270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021064565109296e-05,
+      "loss": 0.7919,
+      "step": 2271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017471436006836e-05,
+      "loss": 0.7638,
+      "step": 2272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601387708860482e-05,
+      "loss": 0.6939,
+      "step": 2273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601028152363096e-05,
+      "loss": 0.8156,
+      "step": 2274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006684741813217e-05,
+      "loss": 0.7598,
+      "step": 2275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003086743879792e-05,
+      "loss": 0.8101,
+      "step": 2276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599948753055914e-05,
+      "loss": 0.8249,
+      "step": 2277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599588710257996e-05,
+      "loss": 0.7632,
+      "step": 2278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599228546067119e-05,
+      "loss": 0.7966,
+      "step": 2279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598868260556203e-05,
+      "loss": 0.7554,
+      "step": 2280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985078537981908e-05,
+      "loss": 0.7524,
+      "step": 2281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981473258660507e-05,
+      "loss": 0.7283,
+      "step": 2282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977866768327752e-05,
+      "loss": 0.7551,
+      "step": 2283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974259067713815e-05,
+      "loss": 0.8034,
+      "step": 2284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970650157549113e-05,
+      "loss": 0.8341,
+      "step": 2285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596704003856431e-05,
+      "loss": 0.7102,
+      "step": 2286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596342871149031e-05,
+      "loss": 0.7817,
+      "step": 2287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5959816177058258e-05,
+      "loss": 0.8091,
+      "step": 2288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956202435999558e-05,
+      "loss": 0.7775,
+      "step": 2289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595258748904584e-05,
+      "loss": 0.7381,
+      "step": 2290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948971336928998e-05,
+      "loss": 0.229,
+      "step": 2291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945353980381154e-05,
+      "loss": 0.7373,
+      "step": 2292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941735420134677e-05,
+      "loss": 0.789,
+      "step": 2293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593811565692219e-05,
+      "loss": 0.7265,
+      "step": 2294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934494691476544e-05,
+      "loss": 0.7764,
+      "step": 2295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593087252453084e-05,
+      "loss": 0.6963,
+      "step": 2296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592724915681843e-05,
+      "loss": 0.7796,
+      "step": 2297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59236245890729e-05,
+      "loss": 0.7889,
+      "step": 2298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591999882202808e-05,
+      "loss": 0.817,
+      "step": 2299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591637185641804e-05,
+      "loss": 0.7867,
+      "step": 2300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591274369297711e-05,
+      "loss": 0.7644,
+      "step": 2301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590911433243984e-05,
+      "loss": 0.8322,
+      "step": 2302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905483775541028e-05,
+      "loss": 0.7489,
+      "step": 2303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901852023015725e-05,
+      "loss": 0.7386,
+      "step": 2304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898219075599213e-05,
+      "loss": 0.7963,
+      "step": 2305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894584934027024e-05,
+      "loss": 0.7818,
+      "step": 2306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890949599034926e-05,
+      "loss": 0.6961,
+      "step": 2307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588731307135893e-05,
+      "loss": 0.7684,
+      "step": 2308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883675351735287e-05,
+      "loss": 0.8075,
+      "step": 2309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588003644090049e-05,
+      "loss": 0.8075,
+      "step": 2310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587639633959128e-05,
+      "loss": 0.7923,
+      "step": 2311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587275504854463e-05,
+      "loss": 0.7957,
+      "step": 2312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869112568497756e-05,
+      "loss": 0.7646,
+      "step": 2313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865468900188115e-05,
+      "loss": 0.7891,
+      "step": 2314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586182404435341e-05,
+      "loss": 0.7694,
+      "step": 2315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585817800173158e-05,
+      "loss": 0.709,
+      "step": 2316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854530773060795e-05,
+      "loss": 0.7202,
+      "step": 2317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850882359079487e-05,
+      "loss": 0.8331,
+      "step": 2318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847232760526306e-05,
+      "loss": 0.768,
+      "step": 2319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843581978140156e-05,
+      "loss": 0.7123,
+      "step": 2320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839930012660177e-05,
+      "loss": 0.7605,
+      "step": 2321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836276864825742e-05,
+      "loss": 0.7164,
+      "step": 2322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583262253537647e-05,
+      "loss": 0.7097,
+      "step": 2323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582896702505222e-05,
+      "loss": 0.7384,
+      "step": 2324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825310334593088e-05,
+      "loss": 0.8447,
+      "step": 2325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821652464739402e-05,
+      "loss": 0.7559,
+      "step": 2326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581799341623175e-05,
+      "loss": 0.7849,
+      "step": 2327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814333189810934e-05,
+      "loss": 0.78,
+      "step": 2328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810671786218003e-05,
+      "loss": 0.7733,
+      "step": 2329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807009206194254e-05,
+      "loss": 0.7654,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580334545048121e-05,
+      "loss": 0.7212,
+      "step": 2331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799680519820636e-05,
+      "loss": 0.8,
+      "step": 2332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796014414954533e-05,
+      "loss": 0.7608,
+      "step": 2333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5792347136625154e-05,
+      "loss": 0.7713,
+      "step": 2334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788678685574962e-05,
+      "loss": 0.7761,
+      "step": 2335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785009062546678e-05,
+      "loss": 0.7632,
+      "step": 2336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578133826828326e-05,
+      "loss": 0.7264,
+      "step": 2337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777666303527897e-05,
+      "loss": 0.7442,
+      "step": 2338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5773993169024014e-05,
+      "loss": 0.7652,
+      "step": 2339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5770318865515274e-05,
+      "loss": 0.7497,
+      "step": 2340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766643393745583e-05,
+      "loss": 0.7326,
+      "step": 2341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576296675445907e-05,
+      "loss": 0.7929,
+      "step": 2342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759288948400118e-05,
+      "loss": 0.7678,
+      "step": 2343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575560997631333e-05,
+      "loss": 0.7591,
+      "step": 2344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575192983894356e-05,
+      "loss": 0.7924,
+      "step": 2345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574824853703588e-05,
+      "loss": 0.7437,
+      "step": 2346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574456607133562e-05,
+      "loss": 0.7759,
+      "step": 2347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574088244258832e-05,
+      "loss": 0.7302,
+      "step": 2348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737197651539777e-05,
+      "loss": 0.7296,
+      "step": 2349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5733511698936014e-05,
+      "loss": 0.769,
+      "step": 2350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729824585523287e-05,
+      "loss": 0.7899,
+      "step": 2351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726136312048097e-05,
+      "loss": 0.7889,
+      "step": 2352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722446879257167e-05,
+      "loss": 0.7857,
+      "step": 2353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5718756287897465e-05,
+      "loss": 0.8114,
+      "step": 2354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571506453871619e-05,
+      "loss": 0.8179,
+      "step": 2355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571137163246077e-05,
+      "loss": 0.7866,
+      "step": 2356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707677569878874e-05,
+      "loss": 0.7775,
+      "step": 2357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5703982351718405e-05,
+      "loss": 0.844,
+      "step": 2358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700285978727497e-05,
+      "loss": 0.7102,
+      "step": 2359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569658845165452e-05,
+      "loss": 0.2285,
+      "step": 2360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569288977124808e-05,
+      "loss": 0.798,
+      "step": 2361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689189938257005e-05,
+      "loss": 0.7691,
+      "step": 2362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685488953430373e-05,
+      "loss": 0.7651,
+      "step": 2363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681786817517482e-05,
+      "loss": 0.7592,
+      "step": 2364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567808353126787e-05,
+      "loss": 0.7575,
+      "step": 2365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5674379095431302e-05,
+      "loss": 0.7893,
+      "step": 2366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5670673510757782e-05,
+      "loss": 0.7246,
+      "step": 2367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566696677799755e-05,
+      "loss": 0.8167,
+      "step": 2368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566325889790106e-05,
+      "loss": 0.7999,
+      "step": 2369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659549871219023e-05,
+      "loss": 0.7841,
+      "step": 2370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565583969870236e-05,
+      "loss": 0.7398,
+      "step": 2371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565212838110224e-05,
+      "loss": 0.7658,
+      "step": 2372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648415919170056e-05,
+      "loss": 0.7176,
+      "step": 2373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644702313657437e-05,
+      "loss": 0.7369,
+      "step": 2374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640987565316236e-05,
+      "loss": 0.8319,
+      "step": 2375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637271674898547e-05,
+      "loss": 0.7861,
+      "step": 2376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633554643156687e-05,
+      "loss": 0.2229,
+      "step": 2377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629836470843212e-05,
+      "loss": 0.7361,
+      "step": 2378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56261171587109e-05,
+      "loss": 0.7912,
+      "step": 2379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622396707512767e-05,
+      "loss": 0.7686,
+      "step": 2380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561867511800206e-05,
+      "loss": 0.7992,
+      "step": 2381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614952390932242e-05,
+      "loss": 0.7893,
+      "step": 2382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611228527057033e-05,
+      "loss": 0.7685,
+      "step": 2383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607503527130356e-05,
+      "loss": 0.8532,
+      "step": 2384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603777391906378e-05,
+      "loss": 0.2395,
+      "step": 2385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600050122139498e-05,
+      "loss": 0.7736,
+      "step": 2386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559632171858434e-05,
+      "loss": 0.7516,
+      "step": 2387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592592181995757e-05,
+      "loss": 0.7242,
+      "step": 2388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558886151312883e-05,
+      "loss": 0.2493,
+      "step": 2389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585129712738872e-05,
+      "loss": 0.714,
+      "step": 2390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558139678158143e-05,
+      "loss": 0.7821,
+      "step": 2391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557766272041227e-05,
+      "loss": 0.7313,
+      "step": 2392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557392752998739e-05,
+      "loss": 0.833,
+      "step": 2393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557019121106302e-05,
+      "loss": 0.7534,
+      "step": 2394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556645376439562e-05,
+      "loss": 0.8495,
+      "step": 2395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556271519074187e-05,
+      "loss": 0.7385,
+      "step": 2396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558975490858686e-05,
+      "loss": 0.7352,
+      "step": 2397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555523466550321e-05,
+      "loss": 0.7785,
+      "step": 2398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551492715432803e-05,
+      "loss": 0.7841,
+      "step": 2399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547749641405068e-05,
+      "loss": 0.7001,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544005444177836e-05,
+      "loss": 0.8156,
+      "step": 2401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540260124509145e-05,
+      "loss": 0.7785,
+      "step": 2402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553651368315728e-05,
+      "loss": 0.7454,
+      "step": 2403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532766120880748e-05,
+      "loss": 0.7477,
+      "step": 2404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529017438438283e-05,
+      "loss": 0.6949,
+      "step": 2405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552526763658884e-05,
+      "loss": 0.2355,
+      "step": 2406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521516716091612e-05,
+      "loss": 0.7418,
+      "step": 2407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517764677706003e-05,
+      "loss": 0.8562,
+      "step": 2408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514011522191662e-05,
+      "loss": 0.7841,
+      "step": 2409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510257250308446e-05,
+      "loss": 0.8264,
+      "step": 2410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506501862816448e-05,
+      "loss": 0.7315,
+      "step": 2411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550274536047599e-05,
+      "loss": 0.7637,
+      "step": 2412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498987744047608e-05,
+      "loss": 0.7598,
+      "step": 2413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495229014292075e-05,
+      "loss": 0.7204,
+      "step": 2414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491469171970386e-05,
+      "loss": 0.8165,
+      "step": 2415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487708217843755e-05,
+      "loss": 0.801,
+      "step": 2416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548394615267363e-05,
+      "loss": 0.699,
+      "step": 2417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548018297722168e-05,
+      "loss": 0.7723,
+      "step": 2418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5476418692249798e-05,
+      "loss": 0.8006,
+      "step": 2419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472653298520103e-05,
+      "loss": 0.8117,
+      "step": 2420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546888679679494e-05,
+      "loss": 0.7869,
+      "step": 2421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465119187836872e-05,
+      "loss": 0.772,
+      "step": 2422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461350472408693e-05,
+      "loss": 0.7525,
+      "step": 2423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457580651273416e-05,
+      "loss": 0.8195,
+      "step": 2424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453809725194284e-05,
+      "loss": 0.7535,
+      "step": 2425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450037694934762e-05,
+      "loss": 0.7436,
+      "step": 2426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446264561258532e-05,
+      "loss": 0.6948,
+      "step": 2427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442490324929505e-05,
+      "loss": 0.694,
+      "step": 2428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543871498671181e-05,
+      "loss": 0.8004,
+      "step": 2429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434938547369815e-05,
+      "loss": 0.6685,
+      "step": 2430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431161007668093e-05,
+      "loss": 0.7301,
+      "step": 2431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427382368371442e-05,
+      "loss": 0.8307,
+      "step": 2432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423602630244895e-05,
+      "loss": 0.7209,
+      "step": 2433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541982179405369e-05,
+      "loss": 0.7191,
+      "step": 2434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416039860563302e-05,
+      "loss": 0.7305,
+      "step": 2435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412256830539428e-05,
+      "loss": 0.7641,
+      "step": 2436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408472704747967e-05,
+      "loss": 0.7816,
+      "step": 2437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540468748395507e-05,
+      "loss": 0.7723,
+      "step": 2438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400901168927084e-05,
+      "loss": 0.8306,
+      "step": 2439
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397113760430592e-05,
+      "loss": 0.7442,
+      "step": 2440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393325259232396e-05,
+      "loss": 0.8295,
+      "step": 2441
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389535666099515e-05,
+      "loss": 0.7941,
+      "step": 2442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385744981799184e-05,
+      "loss": 0.7736,
+      "step": 2443
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381953207098882e-05,
+      "loss": 0.7587,
+      "step": 2444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537816034276628e-05,
+      "loss": 0.7968,
+      "step": 2445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537436638956929e-05,
+      "loss": 0.8466,
+      "step": 2446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370571348276037e-05,
+      "loss": 0.7587,
+      "step": 2447
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536677521965486e-05,
+      "loss": 0.7428,
+      "step": 2448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536297800447433e-05,
+      "loss": 0.7949,
+      "step": 2449
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359179703503234e-05,
+      "loss": 0.7601,
+      "step": 2450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355380317510572e-05,
+      "loss": 0.8114,
+      "step": 2451
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351579847265573e-05,
+      "loss": 0.6769,
+      "step": 2452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534777829353768e-05,
+      "loss": 0.8053,
+      "step": 2453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343975657096565e-05,
+      "loss": 0.7132,
+      "step": 2454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340171938712097e-05,
+      "loss": 0.8166,
+      "step": 2455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336367139154385e-05,
+      "loss": 0.78,
+      "step": 2456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332561259193752e-05,
+      "loss": 0.8124,
+      "step": 2457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328754299600736e-05,
+      "loss": 0.7687,
+      "step": 2458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5324946261146098e-05,
+      "loss": 0.8044,
+      "step": 2459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321137144600812e-05,
+      "loss": 0.7157,
+      "step": 2460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317326950736076e-05,
+      "loss": 0.8193,
+      "step": 2461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313515680323297e-05,
+      "loss": 0.795,
+      "step": 2462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5309703334134116e-05,
+      "loss": 0.2361,
+      "step": 2463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530588991294037e-05,
+      "loss": 0.7376,
+      "step": 2464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302075417514136e-05,
+      "loss": 0.7432,
+      "step": 2465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298259848627694e-05,
+      "loss": 0.7284,
+      "step": 2466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294443207053546e-05,
+      "loss": 0.7566,
+      "step": 2467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290625493564415e-05,
+      "loss": 0.7328,
+      "step": 2468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286806708933228e-05,
+      "loss": 0.7616,
+      "step": 2469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282986853933143e-05,
+      "loss": 0.8136,
+      "step": 2470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279165929337528e-05,
+      "loss": 0.713,
+      "step": 2471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275343935919972e-05,
+      "loss": 0.7376,
+      "step": 2472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527152087445427e-05,
+      "loss": 0.7576,
+      "step": 2473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526769674571445e-05,
+      "loss": 0.7277,
+      "step": 2474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5263871550474737e-05,
+      "loss": 0.7999,
+      "step": 2475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260045289509586e-05,
+      "loss": 0.7196,
+      "step": 2476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256217963593664e-05,
+      "loss": 0.7406,
+      "step": 2477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252389573501853e-05,
+      "loss": 0.6928,
+      "step": 2478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524856012000925e-05,
+      "loss": 0.6817,
+      "step": 2479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244729603891168e-05,
+      "loss": 0.7311,
+      "step": 2480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5240898025923131e-05,
+      "loss": 0.7776,
+      "step": 2481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237065386880887e-05,
+      "loss": 0.7811,
+      "step": 2482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5233231687540392e-05,
+      "loss": 0.677,
+      "step": 2483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229396928677815e-05,
+      "loss": 0.7018,
+      "step": 2484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225561111069547e-05,
+      "loss": 0.8023,
+      "step": 2485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221724235492184e-05,
+      "loss": 0.7914,
+      "step": 2486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217886302722549e-05,
+      "loss": 0.7414,
+      "step": 2487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521404731353766e-05,
+      "loss": 0.7779,
+      "step": 2488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210207268714773e-05,
+      "loss": 0.8181,
+      "step": 2489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206366169031336e-05,
+      "loss": 0.7741,
+      "step": 2490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202524015265026e-05,
+      "loss": 0.8833,
+      "step": 2491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519868080819372e-05,
+      "loss": 0.7165,
+      "step": 2492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194836548595519e-05,
+      "loss": 0.7738,
+      "step": 2493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190991237248734e-05,
+      "loss": 0.7533,
+      "step": 2494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187144874931888e-05,
+      "loss": 0.8092,
+      "step": 2495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183297462423712e-05,
+      "loss": 0.7058,
+      "step": 2496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179449000503161e-05,
+      "loss": 0.8029,
+      "step": 2497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175599489949395e-05,
+      "loss": 0.7875,
+      "step": 2498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171748931541786e-05,
+      "loss": 0.672,
+      "step": 2499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167897326059915e-05,
+      "loss": 0.8186,
+      "step": 2500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516404467428359e-05,
+      "loss": 0.699,
+      "step": 2501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516019097699281e-05,
+      "loss": 0.7413,
+      "step": 2502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156336234967801e-05,
+      "loss": 0.7495,
+      "step": 2503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152480448988995e-05,
+      "loss": 0.7092,
+      "step": 2504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148623619837037e-05,
+      "loss": 0.7338,
+      "step": 2505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144765748292782e-05,
+      "loss": 0.7778,
+      "step": 2506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140906835137295e-05,
+      "loss": 0.7641,
+      "step": 2507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137046881151851e-05,
+      "loss": 0.8169,
+      "step": 2508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133185887117941e-05,
+      "loss": 0.7624,
+      "step": 2509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129323853817263e-05,
+      "loss": 0.8016,
+      "step": 2510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125460782031722e-05,
+      "loss": 0.7193,
+      "step": 2511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121596672543444e-05,
+      "loss": 0.7834,
+      "step": 2512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511773152613475e-05,
+      "loss": 0.7223,
+      "step": 2513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113865343588185e-05,
+      "loss": 0.254,
+      "step": 2514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109998125686497e-05,
+      "loss": 0.8423,
+      "step": 2515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510612987321264e-05,
+      "loss": 0.842,
+      "step": 2516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102260586949787e-05,
+      "loss": 0.8336,
+      "step": 2517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098390267681314e-05,
+      "loss": 0.7449,
+      "step": 2518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094518916190804e-05,
+      "loss": 0.7759,
+      "step": 2519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090646533262057e-05,
+      "loss": 0.7552,
+      "step": 2520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086773119679076e-05,
+      "loss": 0.7893,
+      "step": 2521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508289867622607e-05,
+      "loss": 0.7835,
+      "step": 2522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5079023203687463e-05,
+      "loss": 0.7825,
+      "step": 2523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075146702847887e-05,
+      "loss": 0.8015,
+      "step": 2524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071269174492176e-05,
+      "loss": 0.7909,
+      "step": 2525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067390619405377e-05,
+      "loss": 0.7558,
+      "step": 2526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063511038372748e-05,
+      "loss": 0.727,
+      "step": 2527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059630432179741e-05,
+      "loss": 0.7252,
+      "step": 2528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055748801612034e-05,
+      "loss": 0.7206,
+      "step": 2529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051866147455498e-05,
+      "loss": 0.7766,
+      "step": 2530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047982470496222e-05,
+      "loss": 0.7609,
+      "step": 2531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504409777152049e-05,
+      "loss": 0.8271,
+      "step": 2532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040212051314804e-05,
+      "loss": 0.7663,
+      "step": 2533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503632531066587e-05,
+      "loss": 0.7546,
+      "step": 2534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503243755036059e-05,
+      "loss": 0.8447,
+      "step": 2535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028548771186093e-05,
+      "loss": 0.6922,
+      "step": 2536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024658973929696e-05,
+      "loss": 0.7485,
+      "step": 2537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020768159378932e-05,
+      "loss": 0.8072,
+      "step": 2538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501687632832153e-05,
+      "loss": 0.6539,
+      "step": 2539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012983481545442e-05,
+      "loss": 0.7239,
+      "step": 2540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500908961983881e-05,
+      "loss": 0.8123,
+      "step": 2541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005194743989987e-05,
+      "loss": 0.8004,
+      "step": 2542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001298854787526e-05,
+      "loss": 0.7602,
+      "step": 2543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997401953020199e-05,
+      "loss": 0.8356,
+      "step": 2544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993504039476969e-05,
+      "loss": 0.8325,
+      "step": 2545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989605114947011e-05,
+      "loss": 0.753,
+      "step": 2546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49857051802197e-05,
+      "loss": 0.8339,
+      "step": 2547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981804236084622e-05,
+      "loss": 0.6972,
+      "step": 2548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977902283331562e-05,
+      "loss": 0.7701,
+      "step": 2549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973999322750508e-05,
+      "loss": 0.6968,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970095355131662e-05,
+      "loss": 0.7547,
+      "step": 2551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966190381265415e-05,
+      "loss": 0.23,
+      "step": 2552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962284401942375e-05,
+      "loss": 0.6811,
+      "step": 2553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958377417953345e-05,
+      "loss": 0.7819,
+      "step": 2554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954469430089335e-05,
+      "loss": 0.8034,
+      "step": 2555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495056043914156e-05,
+      "loss": 0.767,
+      "step": 2556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946650445901432e-05,
+      "loss": 0.7769,
+      "step": 2557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942739451160571e-05,
+      "loss": 0.7192,
+      "step": 2558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49388274557108e-05,
+      "loss": 0.7211,
+      "step": 2559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934914460344143e-05,
+      "loss": 0.7985,
+      "step": 2560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931000465852825e-05,
+      "loss": 0.7114,
+      "step": 2561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927085473029275e-05,
+      "loss": 0.7496,
+      "step": 2562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923169482666126e-05,
+      "loss": 0.7796,
+      "step": 2563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919252495556207e-05,
+      "loss": 0.7002,
+      "step": 2564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915334512492559e-05,
+      "loss": 0.7746,
+      "step": 2565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491141553426841e-05,
+      "loss": 0.7609,
+      "step": 2566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907495561677206e-05,
+      "loss": 0.7494,
+      "step": 2567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490357459551258e-05,
+      "loss": 0.772,
+      "step": 2568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899652636568373e-05,
+      "loss": 0.7349,
+      "step": 2569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895729685638632e-05,
+      "loss": 0.7429,
+      "step": 2570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891805743517594e-05,
+      "loss": 0.8,
+      "step": 2571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887880810999702e-05,
+      "loss": 0.7363,
+      "step": 2572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48839548888796e-05,
+      "loss": 0.6872,
+      "step": 2573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880027977952134e-05,
+      "loss": 0.7474,
+      "step": 2574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876100079012343e-05,
+      "loss": 0.8202,
+      "step": 2575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872171192855476e-05,
+      "loss": 0.742,
+      "step": 2576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868241320276972e-05,
+      "loss": 0.7033,
+      "step": 2577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486431046207248e-05,
+      "loss": 0.7357,
+      "step": 2578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860378619037838e-05,
+      "loss": 0.7688,
+      "step": 2579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856445791969094e-05,
+      "loss": 0.7893,
+      "step": 2580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852511981662484e-05,
+      "loss": 0.8204,
+      "step": 2581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484857718891445e-05,
+      "loss": 0.7433,
+      "step": 2582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844641414521631e-05,
+      "loss": 0.7602,
+      "step": 2583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840704659280869e-05,
+      "loss": 0.2556,
+      "step": 2584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836766923989201e-05,
+      "loss": 0.7861,
+      "step": 2585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832828209443856e-05,
+      "loss": 0.7366,
+      "step": 2586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828888516442276e-05,
+      "loss": 0.8483,
+      "step": 2587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824947845782087e-05,
+      "loss": 0.7217,
+      "step": 2588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821006198261123e-05,
+      "loss": 0.7073,
+      "step": 2589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817063574677404e-05,
+      "loss": 0.8155,
+      "step": 2590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813119975829163e-05,
+      "loss": 0.212,
+      "step": 2591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809175402514817e-05,
+      "loss": 0.7924,
+      "step": 2592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805229855532992e-05,
+      "loss": 0.8035,
+      "step": 2593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48012833356825e-05,
+      "loss": 0.7992,
+      "step": 2594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797335843762355e-05,
+      "loss": 0.7244,
+      "step": 2595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793387380571768e-05,
+      "loss": 0.8056,
+      "step": 2596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478943794691015e-05,
+      "loss": 0.7681,
+      "step": 2597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785487543577097e-05,
+      "loss": 0.8255,
+      "step": 2598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478153617137242e-05,
+      "loss": 0.7529,
+      "step": 2599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777583831096108e-05,
+      "loss": 0.7431,
+      "step": 2600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773630523548352e-05,
+      "loss": 0.7902,
+      "step": 2601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769676249529545e-05,
+      "loss": 0.7528,
+      "step": 2602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765721009840265e-05,
+      "loss": 0.7552,
+      "step": 2603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761764805281302e-05,
+      "loss": 0.7021,
+      "step": 2604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757807636653618e-05,
+      "loss": 0.7626,
+      "step": 2605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753849504758394e-05,
+      "loss": 0.7982,
+      "step": 2606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749890410396987e-05,
+      "loss": 0.7845,
+      "step": 2607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474593035437096e-05,
+      "loss": 0.733,
+      "step": 2608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741969337482062e-05,
+      "loss": 0.7013,
+      "step": 2609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738007360532253e-05,
+      "loss": 0.756,
+      "step": 2610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734044424323666e-05,
+      "loss": 0.7886,
+      "step": 2611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730080529658645e-05,
+      "loss": 0.6724,
+      "step": 2612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726115677339719e-05,
+      "loss": 0.7477,
+      "step": 2613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722149868169613e-05,
+      "loss": 0.76,
+      "step": 2614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718183102951248e-05,
+      "loss": 0.7179,
+      "step": 2615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714215382487736e-05,
+      "loss": 0.7137,
+      "step": 2616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710246707582384e-05,
+      "loss": 0.826,
+      "step": 2617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706277079038685e-05,
+      "loss": 0.8161,
+      "step": 2618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702306497660343e-05,
+      "loss": 0.6803,
+      "step": 2619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698334964251236e-05,
+      "loss": 0.7553,
+      "step": 2620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694362479615443e-05,
+      "loss": 0.7644,
+      "step": 2621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690389044557234e-05,
+      "loss": 0.7767,
+      "step": 2622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686414659881076e-05,
+      "loss": 0.7718,
+      "step": 2623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468243932639162e-05,
+      "loss": 0.7592,
+      "step": 2624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678463044893716e-05,
+      "loss": 0.8008,
+      "step": 2625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46744858161924e-05,
+      "loss": 0.7841,
+      "step": 2626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670507641092914e-05,
+      "loss": 0.7195,
+      "step": 2627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666528520400667e-05,
+      "loss": 0.7587,
+      "step": 2628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662548454921281e-05,
+      "loss": 0.7935,
+      "step": 2629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658567445460558e-05,
+      "loss": 0.7147,
+      "step": 2630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46545854928245e-05,
+      "loss": 0.7646,
+      "step": 2631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650602597819288e-05,
+      "loss": 0.7803,
+      "step": 2632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646618761251306e-05,
+      "loss": 0.7953,
+      "step": 2633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642633983927119e-05,
+      "loss": 0.7676,
+      "step": 2634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4638648266653489e-05,
+      "loss": 0.2302,
+      "step": 2635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634661610237364e-05,
+      "loss": 0.8226,
+      "step": 2636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630674015485886e-05,
+      "loss": 0.7767,
+      "step": 2637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626685483206385e-05,
+      "loss": 0.7472,
+      "step": 2638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622696014206376e-05,
+      "loss": 0.7738,
+      "step": 2639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461870560929358e-05,
+      "loss": 0.7721,
+      "step": 2640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461471426927588e-05,
+      "loss": 0.7622,
+      "step": 2641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610721994961376e-05,
+      "loss": 0.7935,
+      "step": 2642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606728787158342e-05,
+      "loss": 0.7353,
+      "step": 2643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602734646675243e-05,
+      "loss": 0.814,
+      "step": 2644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598739574320734e-05,
+      "loss": 0.7932,
+      "step": 2645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594743570903663e-05,
+      "loss": 0.713,
+      "step": 2646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590746637233057e-05,
+      "loss": 0.7595,
+      "step": 2647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458674877411814e-05,
+      "loss": 0.2382,
+      "step": 2648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458274998236832e-05,
+      "loss": 0.6906,
+      "step": 2649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578750262793193e-05,
+      "loss": 0.6975,
+      "step": 2650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574749616202548e-05,
+      "loss": 0.2455,
+      "step": 2651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4570748043406351e-05,
+      "loss": 0.7615,
+      "step": 2652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456674554521477e-05,
+      "loss": 0.7315,
+      "step": 2653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562742122438144e-05,
+      "loss": 0.7488,
+      "step": 2654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455873777588701e-05,
+      "loss": 0.8215,
+      "step": 2655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554732506372089e-05,
+      "loss": 0.7951,
+      "step": 2656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550726314704296e-05,
+      "loss": 0.2467,
+      "step": 2657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546719201694717e-05,
+      "loss": 0.791,
+      "step": 2658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454271116815464e-05,
+      "loss": 0.767,
+      "step": 2659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538702214895527e-05,
+      "loss": 0.7269,
+      "step": 2660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534692342729039e-05,
+      "loss": 0.7968,
+      "step": 2661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530681552467012e-05,
+      "loss": 0.6681,
+      "step": 2662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526669844921473e-05,
+      "loss": 0.741,
+      "step": 2663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522657220904634e-05,
+      "loss": 0.7815,
+      "step": 2664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518643681228892e-05,
+      "loss": 0.7854,
+      "step": 2665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514629226706831e-05,
+      "loss": 0.6818,
+      "step": 2666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510613858151219e-05,
+      "loss": 0.7151,
+      "step": 2667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506597576375008e-05,
+      "loss": 0.8235,
+      "step": 2668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502580382191333e-05,
+      "loss": 0.7254,
+      "step": 2669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498562276413524e-05,
+      "loss": 0.7526,
+      "step": 2670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494543259855079e-05,
+      "loss": 0.7472,
+      "step": 2671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490523333329699e-05,
+      "loss": 0.2242,
+      "step": 2672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486502497651253e-05,
+      "loss": 0.7193,
+      "step": 2673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482480753633808e-05,
+      "loss": 0.7753,
+      "step": 2674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478458102091598e-05,
+      "loss": 0.7518,
+      "step": 2675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474434543839058e-05,
+      "loss": 0.7433,
+      "step": 2676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470410079690795e-05,
+      "loss": 0.7639,
+      "step": 2677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466384710461607e-05,
+      "loss": 0.7514,
+      "step": 2678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446235843696647e-05,
+      "loss": 0.7573,
+      "step": 2679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458331260020546e-05,
+      "loss": 0.6855,
+      "step": 2680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454303180439178e-05,
+      "loss": 0.753,
+      "step": 2681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445027419903789e-05,
+      "loss": 0.7087,
+      "step": 2682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446244316632396e-05,
+      "loss": 0.7832,
+      "step": 2683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442213534038582e-05,
+      "loss": 0.7867,
+      "step": 2684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438181852072529e-05,
+      "loss": 0.7526,
+      "step": 2685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434149271550484e-05,
+      "loss": 0.7318,
+      "step": 2686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430115793288895e-05,
+      "loss": 0.7133,
+      "step": 2687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426081418104372e-05,
+      "loss": 0.8107,
+      "step": 2688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422046146813721e-05,
+      "loss": 0.8445,
+      "step": 2689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418009980233924e-05,
+      "loss": 0.7222,
+      "step": 2690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4413972919182146e-05,
+      "loss": 0.75,
+      "step": 2691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409934964475728e-05,
+      "loss": 0.7227,
+      "step": 2692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405896116932203e-05,
+      "loss": 0.6721,
+      "step": 2693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401856377369268e-05,
+      "loss": 0.8073,
+      "step": 2694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397815746604815e-05,
+      "loss": 0.8149,
+      "step": 2695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393774225456914e-05,
+      "loss": 0.736,
+      "step": 2696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389731814743811e-05,
+      "loss": 0.2457,
+      "step": 2697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438568851528393e-05,
+      "loss": 0.7392,
+      "step": 2698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381644327895886e-05,
+      "loss": 0.8106,
+      "step": 2699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377599253398465e-05,
+      "loss": 0.753,
+      "step": 2700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373553292610627e-05,
+      "loss": 0.7399,
+      "step": 2701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369506446351528e-05,
+      "loss": 0.6891,
+      "step": 2702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436545871544049e-05,
+      "loss": 0.745,
+      "step": 2703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436141010069702e-05,
+      "loss": 0.8012,
+      "step": 2704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43573606029408e-05,
+      "loss": 0.7649,
+      "step": 2705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353310222991692e-05,
+      "loss": 0.8264,
+      "step": 2706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349258961669743e-05,
+      "loss": 0.7289,
+      "step": 2707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4345206819795165e-05,
+      "loss": 0.8352,
+      "step": 2708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341153798188361e-05,
+      "loss": 0.6986,
+      "step": 2709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337099897669906e-05,
+      "loss": 0.7997,
+      "step": 2710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333045119060555e-05,
+      "loss": 0.7438,
+      "step": 2711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328989463181242e-05,
+      "loss": 0.7758,
+      "step": 2712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324932930853073e-05,
+      "loss": 0.7837,
+      "step": 2713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320875522897333e-05,
+      "loss": 0.7349,
+      "step": 2714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316817240135494e-05,
+      "loss": 0.8004,
+      "step": 2715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431275808338919e-05,
+      "loss": 0.7491,
+      "step": 2716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308698053480246e-05,
+      "loss": 0.7234,
+      "step": 2717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304637151230651e-05,
+      "loss": 0.7057,
+      "step": 2718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300575377462582e-05,
+      "loss": 0.7525,
+      "step": 2719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296512732998384e-05,
+      "loss": 0.7106,
+      "step": 2720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292449218660584e-05,
+      "loss": 0.7009,
+      "step": 2721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288384835271878e-05,
+      "loss": 0.7071,
+      "step": 2722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284319583655149e-05,
+      "loss": 0.8486,
+      "step": 2723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280253464633443e-05,
+      "loss": 0.7489,
+      "step": 2724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276186479029992e-05,
+      "loss": 0.7904,
+      "step": 2725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272118627668198e-05,
+      "loss": 0.8043,
+      "step": 2726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268049911371646e-05,
+      "loss": 0.6405,
+      "step": 2727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263980330964078e-05,
+      "loss": 0.7585,
+      "step": 2728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259909887269431e-05,
+      "loss": 0.7011,
+      "step": 2729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255838581111804e-05,
+      "loss": 0.2333,
+      "step": 2730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251766413315481e-05,
+      "loss": 0.6225,
+      "step": 2731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424769338470491e-05,
+      "loss": 0.2338,
+      "step": 2732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424361949610472e-05,
+      "loss": 0.7688,
+      "step": 2733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239544748339711e-05,
+      "loss": 0.7859,
+      "step": 2734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423546914223486e-05,
+      "loss": 0.7886,
+      "step": 2735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231392678615312e-05,
+      "loss": 0.7007,
+      "step": 2736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422731535830639e-05,
+      "loss": 0.77,
+      "step": 2737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4223237182133596e-05,
+      "loss": 0.7645,
+      "step": 2738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421915815092259e-05,
+      "loss": 0.7611,
+      "step": 2739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215078265499223e-05,
+      "loss": 0.8218,
+      "step": 2740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210997526689501e-05,
+      "loss": 0.6796,
+      "step": 2741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206915935319619e-05,
+      "loss": 0.7755,
+      "step": 2742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202833492215936e-05,
+      "loss": 0.7386,
+      "step": 2743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198750198204983e-05,
+      "loss": 0.7916,
+      "step": 2744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194666054113464e-05,
+      "loss": 0.758,
+      "step": 2745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190581060768262e-05,
+      "loss": 0.7306,
+      "step": 2746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186495218996425e-05,
+      "loss": 0.7236,
+      "step": 2747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182408529625167e-05,
+      "loss": 0.7994,
+      "step": 2748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417832099348189e-05,
+      "loss": 0.7325,
+      "step": 2749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174232611394153e-05,
+      "loss": 0.7893,
+      "step": 2750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170143384189694e-05,
+      "loss": 0.8079,
+      "step": 2751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166053312696413e-05,
+      "loss": 0.7068,
+      "step": 2752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41619623977424e-05,
+      "loss": 0.8026,
+      "step": 2753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157870640155889e-05,
+      "loss": 0.7001,
+      "step": 2754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4153778040765309e-05,
+      "loss": 0.7284,
+      "step": 2755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149684600399242e-05,
+      "loss": 0.7234,
+      "step": 2756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145590319886455e-05,
+      "loss": 0.7988,
+      "step": 2757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414149520005587e-05,
+      "loss": 0.6969,
+      "step": 2758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137399241736593e-05,
+      "loss": 0.7702,
+      "step": 2759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133302445757893e-05,
+      "loss": 0.7469,
+      "step": 2760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129204812949201e-05,
+      "loss": 0.7589,
+      "step": 2761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125106344140133e-05,
+      "loss": 0.8244,
+      "step": 2762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121007040160466e-05,
+      "loss": 0.7239,
+      "step": 2763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116906901840141e-05,
+      "loss": 0.6529,
+      "step": 2764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411280593000928e-05,
+      "loss": 0.77,
+      "step": 2765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108704125498164e-05,
+      "loss": 0.7537,
+      "step": 2766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104601489137246e-05,
+      "loss": 0.7739,
+      "step": 2767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410049802175715e-05,
+      "loss": 0.7314,
+      "step": 2768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096393724188659e-05,
+      "loss": 0.7091,
+      "step": 2769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409228859726274e-05,
+      "loss": 0.7178,
+      "step": 2770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408818264181051e-05,
+      "loss": 0.7656,
+      "step": 2771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084075858663272e-05,
+      "loss": 0.787,
+      "step": 2772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079968248652476e-05,
+      "loss": 0.7731,
+      "step": 2773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075859812609758e-05,
+      "loss": 0.7746,
+      "step": 2774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071750551366908e-05,
+      "loss": 0.7627,
+      "step": 2775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067640465755893e-05,
+      "loss": 0.7158,
+      "step": 2776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063529556608839e-05,
+      "loss": 0.7351,
+      "step": 2777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059417824758045e-05,
+      "loss": 0.7476,
+      "step": 2778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055305271035974e-05,
+      "loss": 0.7189,
+      "step": 2779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051191896275254e-05,
+      "loss": 0.7063,
+      "step": 2780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047077701308677e-05,
+      "loss": 0.7163,
+      "step": 2781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042962686969213e-05,
+      "loss": 0.7334,
+      "step": 2782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403884685408998e-05,
+      "loss": 0.7521,
+      "step": 2783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034730203504277e-05,
+      "loss": 0.7571,
+      "step": 2784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030612736045561e-05,
+      "loss": 0.6947,
+      "step": 2785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026494452547456e-05,
+      "loss": 0.8252,
+      "step": 2786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022375353843756e-05,
+      "loss": 0.8032,
+      "step": 2787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018255440768405e-05,
+      "loss": 0.7399,
+      "step": 2788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401413471415553e-05,
+      "loss": 0.8008,
+      "step": 2789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401001317483941e-05,
+      "loss": 0.6483,
+      "step": 2790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005890823654498e-05,
+      "loss": 0.7951,
+      "step": 2791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001767661435404e-05,
+      "loss": 0.7192,
+      "step": 2792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399764368901691e-05,
+      "loss": 0.761,
+      "step": 2793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993518907233949e-05,
+      "loss": 0.7393,
+      "step": 2794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398939331692163e-05,
+      "loss": 0.7312,
+      "step": 2795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985266918915221e-05,
+      "loss": 0.7368,
+      "step": 2796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981139714050156e-05,
+      "loss": 0.817,
+      "step": 2797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977011703162026e-05,
+      "loss": 0.654,
+      "step": 2798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972882887086598e-05,
+      "loss": 0.7971,
+      "step": 2799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968753266659784e-05,
+      "loss": 0.7303,
+      "step": 2800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964622842717674e-05,
+      "loss": 0.832,
+      "step": 2801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960491616096514e-05,
+      "loss": 0.7204,
+      "step": 2802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956359587632712e-05,
+      "loss": 0.77,
+      "step": 2803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952226758162846e-05,
+      "loss": 0.7499,
+      "step": 2804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948093128523642e-05,
+      "loss": 0.7604,
+      "step": 2805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943958699552003e-05,
+      "loss": 0.7723,
+      "step": 2806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939823472084983e-05,
+      "loss": 0.8048,
+      "step": 2807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935687446959802e-05,
+      "loss": 0.6781,
+      "step": 2808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931550625013844e-05,
+      "loss": 0.7663,
+      "step": 2809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927413007084648e-05,
+      "loss": 0.6972,
+      "step": 2810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923274594009921e-05,
+      "loss": 0.8358,
+      "step": 2811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919135386627524e-05,
+      "loss": 0.2167,
+      "step": 2812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914995385775489e-05,
+      "loss": 0.7827,
+      "step": 2813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910854592291993e-05,
+      "loss": 0.7727,
+      "step": 2814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906713007015392e-05,
+      "loss": 0.2669,
+      "step": 2815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902570630784185e-05,
+      "loss": 0.6818,
+      "step": 2816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898427464437045e-05,
+      "loss": 0.2512,
+      "step": 2817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894283508812793e-05,
+      "loss": 0.7863,
+      "step": 2818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890138764750422e-05,
+      "loss": 0.7577,
+      "step": 2819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885993233089077e-05,
+      "loss": 0.8167,
+      "step": 2820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388184691466806e-05,
+      "loss": 0.8494,
+      "step": 2821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877699810326839e-05,
+      "loss": 0.8045,
+      "step": 2822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387355192090504e-05,
+      "loss": 0.7907,
+      "step": 2823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386940324724244e-05,
+      "loss": 0.7696,
+      "step": 2824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865253790178992e-05,
+      "loss": 0.6856,
+      "step": 2825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861103550554788e-05,
+      "loss": 0.7211,
+      "step": 2826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385695252921009e-05,
+      "loss": 0.8037,
+      "step": 2827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852800726985311e-05,
+      "loss": 0.7659,
+      "step": 2828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848648144721035e-05,
+      "loss": 0.7051,
+      "step": 2829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844494783257986e-05,
+      "loss": 0.7565,
+      "step": 2830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840340643437062e-05,
+      "loss": 0.6803,
+      "step": 2831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836185726099308e-05,
+      "loss": 0.7507,
+      "step": 2832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832030032085933e-05,
+      "loss": 0.8082,
+      "step": 2833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3827873562238298e-05,
+      "loss": 0.2265,
+      "step": 2834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823716317397926e-05,
+      "loss": 0.725,
+      "step": 2835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819558298406493e-05,
+      "loss": 0.7493,
+      "step": 2836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815399506105829e-05,
+      "loss": 0.7557,
+      "step": 2837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811239941337932e-05,
+      "loss": 0.7876,
+      "step": 2838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807079604944944e-05,
+      "loss": 0.7891,
+      "step": 2839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802918497769171e-05,
+      "loss": 0.7879,
+      "step": 2840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379875662065307e-05,
+      "loss": 0.7497,
+      "step": 2841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794593974439262e-05,
+      "loss": 0.7935,
+      "step": 2842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790430559970506e-05,
+      "loss": 0.8176,
+      "step": 2843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786266378089742e-05,
+      "loss": 0.665,
+      "step": 2844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378210142964004e-05,
+      "loss": 0.7564,
+      "step": 2845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777935715464647e-05,
+      "loss": 0.705,
+      "step": 2846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377376923640695e-05,
+      "loss": 0.7867,
+      "step": 2847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769601993310498e-05,
+      "loss": 0.7472,
+      "step": 2848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376543398701899e-05,
+      "loss": 0.7778,
+      "step": 2849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376126521837628e-05,
+      "loss": 0.7121,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375709568822639e-05,
+      "loss": 0.7622,
+      "step": 2851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375292539741347e-05,
+      "loss": 0.8319,
+      "step": 2852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374875434678185e-05,
+      "loss": 0.7812,
+      "step": 2853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3744582537175994e-05,
+      "loss": 0.7631,
+      "step": 2854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740409969440536e-05,
+      "loss": 0.7191,
+      "step": 2855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736236644420248e-05,
+      "loss": 0.8355,
+      "step": 2856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732062562960068e-05,
+      "loss": 0.7352,
+      "step": 2857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372788772590508e-05,
+      "loss": 0.7577,
+      "step": 2858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723712134100526e-05,
+      "loss": 0.8209,
+      "step": 2859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719535788391796e-05,
+      "loss": 0.7997,
+      "step": 2860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715358689624431e-05,
+      "loss": 0.7511,
+      "step": 2861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711180838644133e-05,
+      "loss": 0.7557,
+      "step": 2862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370700223629675e-05,
+      "loss": 0.7809,
+      "step": 2863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702822883428283e-05,
+      "loss": 0.8047,
+      "step": 2864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698642780884886e-05,
+      "loss": 0.7624,
+      "step": 2865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694461929512863e-05,
+      "loss": 0.7637,
+      "step": 2866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690280330158671e-05,
+      "loss": 0.8265,
+      "step": 2867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368609798366892e-05,
+      "loss": 0.758,
+      "step": 2868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681914890890364e-05,
+      "loss": 0.7572,
+      "step": 2869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677731052669923e-05,
+      "loss": 0.7532,
+      "step": 2870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673546469854649e-05,
+      "loss": 0.8276,
+      "step": 2871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669361143291759e-05,
+      "loss": 0.7592,
+      "step": 2872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665175073828616e-05,
+      "loss": 0.7467,
+      "step": 2873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660988262312733e-05,
+      "loss": 0.81,
+      "step": 2874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656800709591771e-05,
+      "loss": 0.7478,
+      "step": 2875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3652612416513547e-05,
+      "loss": 0.7362,
+      "step": 2876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648423383926019e-05,
+      "loss": 0.7674,
+      "step": 2877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644233612677309e-05,
+      "loss": 0.7798,
+      "step": 2878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364004310361567e-05,
+      "loss": 0.7973,
+      "step": 2879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635851857589521e-05,
+      "loss": 0.761,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631659875447421e-05,
+      "loss": 0.7593,
+      "step": 2881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627467158038078e-05,
+      "loss": 0.6835,
+      "step": 2882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623273706210355e-05,
+      "loss": 0.7208,
+      "step": 2883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619079520813258e-05,
+      "loss": 0.7209,
+      "step": 2884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614884602695947e-05,
+      "loss": 0.7491,
+      "step": 2885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610688952707718e-05,
+      "loss": 0.7268,
+      "step": 2886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606492571698031e-05,
+      "loss": 0.2284,
+      "step": 2887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602295460516484e-05,
+      "loss": 0.7542,
+      "step": 2888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359809762001283e-05,
+      "loss": 0.7555,
+      "step": 2889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593899051036957e-05,
+      "loss": 0.7963,
+      "step": 2890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589699754438916e-05,
+      "loss": 0.7309,
+      "step": 2891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585499731068897e-05,
+      "loss": 0.8472,
+      "step": 2892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358129898177724e-05,
+      "loss": 0.7949,
+      "step": 2893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577097507414422e-05,
+      "loss": 0.678,
+      "step": 2894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572895308831087e-05,
+      "loss": 0.8143,
+      "step": 2895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3568692386878002e-05,
+      "loss": 0.7867,
+      "step": 2896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564488742406102e-05,
+      "loss": 0.7251,
+      "step": 2897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356028437626645e-05,
+      "loss": 0.7159,
+      "step": 2898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556079289310273e-05,
+      "loss": 0.242,
+      "step": 2899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551873482388929e-05,
+      "loss": 0.7662,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547666956353922e-05,
+      "loss": 0.6831,
+      "step": 2901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543459712056918e-05,
+      "loss": 0.8511,
+      "step": 2902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539251750349708e-05,
+      "loss": 0.2323,
+      "step": 2903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535043072084245e-05,
+      "loss": 0.787,
+      "step": 2904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530833678112612e-05,
+      "loss": 0.7624,
+      "step": 2905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526623569287054e-05,
+      "loss": 0.7054,
+      "step": 2906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352241274645994e-05,
+      "loss": 0.7888,
+      "step": 2907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518201210483803e-05,
+      "loss": 0.6913,
+      "step": 2908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513988962211309e-05,
+      "loss": 0.7681,
+      "step": 2909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509776002495274e-05,
+      "loss": 0.7681,
+      "step": 2910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505562332188649e-05,
+      "loss": 0.7667,
+      "step": 2911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501347952144543e-05,
+      "loss": 0.2499,
+      "step": 2912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497132863216199e-05,
+      "loss": 0.6668,
+      "step": 2913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492917066257002e-05,
+      "loss": 0.8345,
+      "step": 2914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488700562120485e-05,
+      "loss": 0.7957,
+      "step": 2915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484483351660325e-05,
+      "loss": 0.7272,
+      "step": 2916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348026543573034e-05,
+      "loss": 0.7124,
+      "step": 2917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347604681518449e-05,
+      "loss": 0.7352,
+      "step": 2918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471827490876876e-05,
+      "loss": 0.758,
+      "step": 2919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467607463661751e-05,
+      "loss": 0.663,
+      "step": 2920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463386734393497e-05,
+      "loss": 0.7927,
+      "step": 2921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459165303926644e-05,
+      "loss": 0.2565,
+      "step": 2922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345494317311587e-05,
+      "loss": 0.7275,
+      "step": 2923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450720342815983e-05,
+      "loss": 0.7588,
+      "step": 2924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446496813881944e-05,
+      "loss": 0.8194,
+      "step": 2925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344227258716885e-05,
+      "loss": 0.8127,
+      "step": 2926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438047663531934e-05,
+      "loss": 0.7799,
+      "step": 2927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433822043826581e-05,
+      "loss": 0.7654,
+      "step": 2928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342959572890831e-05,
+      "loss": 0.7258,
+      "step": 2929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342536871963278e-05,
+      "loss": 0.8727,
+      "step": 2930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421141016855796e-05,
+      "loss": 0.7291,
+      "step": 2931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416912621433297e-05,
+      "loss": 0.7593,
+      "step": 2932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341268353422137e-05,
+      "loss": 0.7341,
+      "step": 2933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408453756076239e-05,
+      "loss": 0.8103,
+      "step": 2934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340422328785426e-05,
+      "loss": 0.7815,
+      "step": 2935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399992130411939e-05,
+      "loss": 0.7597,
+      "step": 2936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395760284605915e-05,
+      "loss": 0.7349,
+      "step": 2937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391527751292972e-05,
+      "loss": 0.7043,
+      "step": 2938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338729453133003e-05,
+      "loss": 0.7525,
+      "step": 2939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338306062557415e-05,
+      "loss": 0.7499,
+      "step": 2940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378826034882524e-05,
+      "loss": 0.6962,
+      "step": 2941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374590760112493e-05,
+      "loss": 0.782,
+      "step": 2942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370354802121532e-05,
+      "loss": 0.7509,
+      "step": 2943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366118161767257e-05,
+      "loss": 0.7276,
+      "step": 2944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361880839907415e-05,
+      "loss": 0.2302,
+      "step": 2945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33576428373999e-05,
+      "loss": 0.7384,
+      "step": 2946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353404155102738e-05,
+      "loss": 0.6529,
+      "step": 2947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349164793874094e-05,
+      "loss": 0.7109,
+      "step": 2948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344924754572271e-05,
+      "loss": 0.7576,
+      "step": 2949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3340684038055707e-05,
+      "loss": 0.6606,
+      "step": 2950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336442645182983e-05,
+      "loss": 0.7525,
+      "step": 2951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333220057681281e-05,
+      "loss": 0.713,
+      "step": 2952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327957833804042e-05,
+      "loss": 0.7476,
+      "step": 2953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323714417015663e-05,
+      "loss": 0.7628,
+      "step": 2954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319470327306797e-05,
+      "loss": 0.7691,
+      "step": 2955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315225565536705e-05,
+      "loss": 0.7303,
+      "step": 2956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310980132564782e-05,
+      "loss": 0.2296,
+      "step": 2957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330673402925056e-05,
+      "loss": 0.7934,
+      "step": 2958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330248725645371e-05,
+      "loss": 0.7839,
+      "step": 2959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298239815034032e-05,
+      "loss": 0.2435,
+      "step": 2960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329399170585147e-05,
+      "loss": 0.7927,
+      "step": 2961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289742929766085e-05,
+      "loss": 0.7646,
+      "step": 2962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285493487638102e-05,
+      "loss": 0.8066,
+      "step": 2963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281243380327851e-05,
+      "loss": 0.6902,
+      "step": 2964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276992608695818e-05,
+      "loss": 0.7331,
+      "step": 2965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3272741173602614e-05,
+      "loss": 0.7041,
+      "step": 2966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268489075908985e-05,
+      "loss": 0.714,
+      "step": 2967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3264236316475817e-05,
+      "loss": 0.7528,
+      "step": 2968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259982896164117e-05,
+      "loss": 0.8045,
+      "step": 2969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255728815835042e-05,
+      "loss": 0.7075,
+      "step": 2970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325147407634987e-05,
+      "loss": 0.7264,
+      "step": 2971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324721867857002e-05,
+      "loss": 0.7953,
+      "step": 2972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242962623357041e-05,
+      "loss": 0.7563,
+      "step": 2973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238705911572614e-05,
+      "loss": 0.8118,
+      "step": 2974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234448544078554e-05,
+      "loss": 0.7867,
+      "step": 2975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323019052173681e-05,
+      "loss": 0.7619,
+      "step": 2976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225931845409463e-05,
+      "loss": 0.2372,
+      "step": 2977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322167251595873e-05,
+      "loss": 0.7102,
+      "step": 2978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217412534246952e-05,
+      "loss": 0.7417,
+      "step": 2979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213151901136604e-05,
+      "loss": 0.7645,
+      "step": 2980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208890617490303e-05,
+      "loss": 0.8326,
+      "step": 2981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204628684170784e-05,
+      "loss": 0.8051,
+      "step": 2982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200366102040923e-05,
+      "loss": 0.7567,
+      "step": 2983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196102871963718e-05,
+      "loss": 0.7848,
+      "step": 2984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191838994802316e-05,
+      "loss": 0.695,
+      "step": 2985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318757447141997e-05,
+      "loss": 0.7056,
+      "step": 2986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3183309302680088e-05,
+      "loss": 0.7417,
+      "step": 2987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179043489446189e-05,
+      "loss": 0.7183,
+      "step": 2988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174777032581937e-05,
+      "loss": 0.7873,
+      "step": 2989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170509932951116e-05,
+      "loss": 0.744,
+      "step": 2990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166242191417651e-05,
+      "loss": 0.6677,
+      "step": 2991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161973808845583e-05,
+      "loss": 0.8055,
+      "step": 2992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157704786099095e-05,
+      "loss": 0.6867,
+      "step": 2993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153435124042496e-05,
+      "loss": 0.6786,
+      "step": 2994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149164823540218e-05,
+      "loss": 0.8134,
+      "step": 2995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144893885456833e-05,
+      "loss": 0.7911,
+      "step": 2996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140622310657034e-05,
+      "loss": 0.7365,
+      "step": 2997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136350100005646e-05,
+      "loss": 0.7998,
+      "step": 2998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132077254367626e-05,
+      "loss": 0.8032,
+      "step": 2999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312780377460805e-05,
+      "loss": 0.7452,
+      "step": 3000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123529661592129e-05,
+      "loss": 0.7195,
+      "step": 3001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3119254916185207e-05,
+      "loss": 0.7229,
+      "step": 3002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114979539252746e-05,
+      "loss": 0.7698,
+      "step": 3003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110703531660343e-05,
+      "loss": 0.7466,
+      "step": 3004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3106426894273716e-05,
+      "loss": 0.7688,
+      "step": 3005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102149627958723e-05,
+      "loss": 0.7442,
+      "step": 3006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097871733581328e-05,
+      "loss": 0.7404,
+      "step": 3007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309359321200765e-05,
+      "loss": 0.2402,
+      "step": 3008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089314064103906e-05,
+      "loss": 0.784,
+      "step": 3009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085034290736465e-05,
+      "loss": 0.7256,
+      "step": 3010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080753892771803e-05,
+      "loss": 0.7662,
+      "step": 3011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076472871076538e-05,
+      "loss": 0.7755,
+      "step": 3012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072191226517405e-05,
+      "loss": 0.8017,
+      "step": 3013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067908959961265e-05,
+      "loss": 0.7962,
+      "step": 3014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063626072275112e-05,
+      "loss": 0.7755,
+      "step": 3015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059342564326052e-05,
+      "loss": 0.7129,
+      "step": 3016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055058436981336e-05,
+      "loss": 0.704,
+      "step": 3017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050773691108325e-05,
+      "loss": 0.6783,
+      "step": 3018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046488327574514e-05,
+      "loss": 0.7451,
+      "step": 3019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042202347247512e-05,
+      "loss": 0.8136,
+      "step": 3020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037915750995068e-05,
+      "loss": 0.7953,
+      "step": 3021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033628539685042e-05,
+      "loss": 0.7936,
+      "step": 3022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302934071418543e-05,
+      "loss": 0.7496,
+      "step": 3023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302505227536434e-05,
+      "loss": 0.7962,
+      "step": 3024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020763224090018e-05,
+      "loss": 0.7382,
+      "step": 3025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016473561230823e-05,
+      "loss": 0.7519,
+      "step": 3026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301218328765524e-05,
+      "loss": 0.7681,
+      "step": 3027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007892404231886e-05,
+      "loss": 0.7375,
+      "step": 3028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003600911829486e-05,
+      "loss": 0.7443,
+      "step": 3029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999308811316906e-05,
+      "loss": 0.732,
+      "step": 3030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299501610356312e-05,
+      "loss": 0.8467,
+      "step": 3031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990722789437235e-05,
+      "loss": 0.7784,
+      "step": 3032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986428869808475e-05,
+      "loss": 0.7479,
+      "step": 3033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982134345546192e-05,
+      "loss": 0.7123,
+      "step": 3034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297783921751985e-05,
+      "loss": 0.7998,
+      "step": 3035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973543486599047e-05,
+      "loss": 0.7308,
+      "step": 3036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969247153653497e-05,
+      "loss": 0.7192,
+      "step": 3037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296495021955304e-05,
+      "loss": 0.7413,
+      "step": 3038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960652685167634e-05,
+      "loss": 0.7645,
+      "step": 3039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956354551367358e-05,
+      "loss": 0.7108,
+      "step": 3040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295205581902241e-05,
+      "loss": 0.6834,
+      "step": 3041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947756489003123e-05,
+      "loss": 0.7979,
+      "step": 3042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943456562179928e-05,
+      "loss": 0.744,
+      "step": 3043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939156039423401e-05,
+      "loss": 0.7074,
+      "step": 3044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934854921604223e-05,
+      "loss": 0.7508,
+      "step": 3045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930553209593201e-05,
+      "loss": 0.7103,
+      "step": 3046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926250904261261e-05,
+      "loss": 0.7578,
+      "step": 3047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921948006479448e-05,
+      "loss": 0.712,
+      "step": 3048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291764451711893e-05,
+      "loss": 0.7658,
+      "step": 3049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291334043705099e-05,
+      "loss": 0.7563,
+      "step": 3050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909035767147038e-05,
+      "loss": 0.7298,
+      "step": 3051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2904730508278595e-05,
+      "loss": 0.7697,
+      "step": 3052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900424661317313e-05,
+      "loss": 0.7795,
+      "step": 3053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896118227134945e-05,
+      "loss": 0.7679,
+      "step": 3054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289181120660338e-05,
+      "loss": 0.762,
+      "step": 3055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887503600594615e-05,
+      "loss": 0.8115,
+      "step": 3056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883195409980776e-05,
+      "loss": 0.8631,
+      "step": 3057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2878886635634094e-05,
+      "loss": 0.2413,
+      "step": 3058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2874577278426927e-05,
+      "loss": 0.8151,
+      "step": 3059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870267339231753e-05,
+      "loss": 0.8237,
+      "step": 3060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865956818921157e-05,
+      "loss": 0.7642,
+      "step": 3061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861645718367855e-05,
+      "loss": 0.7639,
+      "step": 3062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857334038444668e-05,
+      "loss": 0.7257,
+      "step": 3063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853021780024548e-05,
+      "loss": 0.2361,
+      "step": 3064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284870894398055e-05,
+      "loss": 0.7734,
+      "step": 3065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284439553118586e-05,
+      "loss": 0.6458,
+      "step": 3066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2840081542513763e-05,
+      "loss": 0.8531,
+      "step": 3067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835766978837676e-05,
+      "loss": 0.7632,
+      "step": 3068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831451841031126e-05,
+      "loss": 0.7319,
+      "step": 3069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827136129967761e-05,
+      "loss": 0.7369,
+      "step": 3070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822819846521337e-05,
+      "loss": 0.7394,
+      "step": 3071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818502991565735e-05,
+      "loss": 0.6826,
+      "step": 3072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2814185565974941e-05,
+      "loss": 0.84,
+      "step": 3073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280986757062307e-05,
+      "loss": 0.7933,
+      "step": 3074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805549006384337e-05,
+      "loss": 0.7145,
+      "step": 3075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801229874133085e-05,
+      "loss": 0.8171,
+      "step": 3076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796910174743764e-05,
+      "loss": 0.7038,
+      "step": 3077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792589909090948e-05,
+      "loss": 0.7614,
+      "step": 3078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788269078049318e-05,
+      "loss": 0.769,
+      "step": 3079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783947682493665e-05,
+      "loss": 0.7495,
+      "step": 3080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2779625723298905e-05,
+      "loss": 0.7179,
+      "step": 3081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775303201340061e-05,
+      "loss": 0.7581,
+      "step": 3082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277098011749228e-05,
+      "loss": 0.7875,
+      "step": 3083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766656472630802e-05,
+      "loss": 0.8494,
+      "step": 3084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762332267631007e-05,
+      "loss": 0.6858,
+      "step": 3085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275800750336837e-05,
+      "loss": 0.6751,
+      "step": 3086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753682180718483e-05,
+      "loss": 0.7341,
+      "step": 3087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749356300557053e-05,
+      "loss": 0.7333,
+      "step": 3088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745029863759898e-05,
+      "loss": 0.7052,
+      "step": 3089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740702871202953e-05,
+      "loss": 0.7863,
+      "step": 3090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736375323762263e-05,
+      "loss": 0.6937,
+      "step": 3091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732047222313981e-05,
+      "loss": 0.7852,
+      "step": 3092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727718567734378e-05,
+      "loss": 0.769,
+      "step": 3093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723389360899837e-05,
+      "loss": 0.7513,
+      "step": 3094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719059602686847e-05,
+      "loss": 0.7882,
+      "step": 3095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714729293972017e-05,
+      "loss": 0.6619,
+      "step": 3096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271039843563206e-05,
+      "loss": 0.2285,
+      "step": 3097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706067028543804e-05,
+      "loss": 0.7484,
+      "step": 3098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701735073584186e-05,
+      "loss": 0.8336,
+      "step": 3099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697402571630261e-05,
+      "loss": 0.7549,
+      "step": 3100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269306952355918e-05,
+      "loss": 0.7473,
+      "step": 3101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688735930248219e-05,
+      "loss": 0.715,
+      "step": 3102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684401792574756e-05,
+      "loss": 0.7749,
+      "step": 3103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268006711141629e-05,
+      "loss": 0.7302,
+      "step": 3104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675731887650411e-05,
+      "loss": 0.7868,
+      "step": 3105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267139612215484e-05,
+      "loss": 0.7522,
+      "step": 3106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266705981580739e-05,
+      "loss": 0.7417,
+      "step": 3107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662722969485995e-05,
+      "loss": 0.7431,
+      "step": 3108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658385584068694e-05,
+      "loss": 0.6709,
+      "step": 3109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654047660433635e-05,
+      "loss": 0.7666,
+      "step": 3110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649709199459074e-05,
+      "loss": 0.7689,
+      "step": 3111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645370202023384e-05,
+      "loss": 0.7952,
+      "step": 3112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641030669005036e-05,
+      "loss": 0.7575,
+      "step": 3113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263669060128261e-05,
+      "loss": 0.7204,
+      "step": 3114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632349999734804e-05,
+      "loss": 0.7621,
+      "step": 3115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2628008865240414e-05,
+      "loss": 0.713,
+      "step": 3116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262366719867835e-05,
+      "loss": 0.7999,
+      "step": 3117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619325000927623e-05,
+      "loss": 0.757,
+      "step": 3118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614982272867362e-05,
+      "loss": 0.7034,
+      "step": 3119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610639015376793e-05,
+      "loss": 0.7394,
+      "step": 3120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606295229335258e-05,
+      "loss": 0.7479,
+      "step": 3121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601950915622195e-05,
+      "loss": 0.2471,
+      "step": 3122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597606075117161e-05,
+      "loss": 0.7844,
+      "step": 3123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593260708699808e-05,
+      "loss": 0.7678,
+      "step": 3124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588914817249911e-05,
+      "loss": 0.807,
+      "step": 3125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584568401647332e-05,
+      "loss": 0.8556,
+      "step": 3126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258022146277205e-05,
+      "loss": 0.7538,
+      "step": 3127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575874001504149e-05,
+      "loss": 0.754,
+      "step": 3128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571526018723817e-05,
+      "loss": 0.2394,
+      "step": 3129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567177515311348e-05,
+      "loss": 0.7328,
+      "step": 3130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2562828492147141e-05,
+      "loss": 0.7176,
+      "step": 3131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558478950111705e-05,
+      "loss": 0.6969,
+      "step": 3132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554128890085644e-05,
+      "loss": 0.2129,
+      "step": 3133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549778312949678e-05,
+      "loss": 0.8054,
+      "step": 3134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254542721958462e-05,
+      "loss": 0.7978,
+      "step": 3135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541075610871402e-05,
+      "loss": 0.7489,
+      "step": 3136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536723487691047e-05,
+      "loss": 0.7326,
+      "step": 3137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532370850924691e-05,
+      "loss": 0.7839,
+      "step": 3138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528017701453568e-05,
+      "loss": 0.7777,
+      "step": 3139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523664040159016e-05,
+      "loss": 0.7226,
+      "step": 3140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519309867922482e-05,
+      "loss": 0.805,
+      "step": 3141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2514955185625515e-05,
+      "loss": 0.7211,
+      "step": 3142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510599994149764e-05,
+      "loss": 0.7365,
+      "step": 3143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506244294376981e-05,
+      "loss": 0.7084,
+      "step": 3144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501888087189023e-05,
+      "loss": 0.7293,
+      "step": 3145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497531373467853e-05,
+      "loss": 0.7659,
+      "step": 3146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249317415409553e-05,
+      "loss": 0.6804,
+      "step": 3147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488816429954217e-05,
+      "loss": 0.7831,
+      "step": 3148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484458201926184e-05,
+      "loss": 0.7375,
+      "step": 3149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480099470893796e-05,
+      "loss": 0.7043,
+      "step": 3150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475740237739527e-05,
+      "loss": 0.8009,
+      "step": 3151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471380503345945e-05,
+      "loss": 0.7035,
+      "step": 3152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467020268595729e-05,
+      "loss": 0.753,
+      "step": 3153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462659534371646e-05,
+      "loss": 0.8041,
+      "step": 3154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458298301556577e-05,
+      "loss": 0.7306,
+      "step": 3155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453936571033499e-05,
+      "loss": 0.7108,
+      "step": 3156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244957434368549e-05,
+      "loss": 0.7659,
+      "step": 3157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445211620395724e-05,
+      "loss": 0.725,
+      "step": 3158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440848402047483e-05,
+      "loss": 0.7296,
+      "step": 3159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436484689524147e-05,
+      "loss": 0.2487,
+      "step": 3160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432120483709192e-05,
+      "loss": 0.7944,
+      "step": 3161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24277557854862e-05,
+      "loss": 0.6868,
+      "step": 3162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423390595738844e-05,
+      "loss": 0.8093,
+      "step": 3163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419024915350908e-05,
+      "loss": 0.7971,
+      "step": 3164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414658745206265e-05,
+      "loss": 0.7349,
+      "step": 3165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410292086188898e-05,
+      "loss": 0.2614,
+      "step": 3166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405924939182871e-05,
+      "loss": 0.7569,
+      "step": 3167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240155730507237e-05,
+      "loss": 0.696,
+      "step": 3168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397189184741659e-05,
+      "loss": 0.8042,
+      "step": 3169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392820579075115e-05,
+      "loss": 0.7398,
+      "step": 3170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388451488957205e-05,
+      "loss": 0.7431,
+      "step": 3171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384081915272502e-05,
+      "loss": 0.7306,
+      "step": 3172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237971185890566e-05,
+      "loss": 0.6784,
+      "step": 3173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2375341320741452e-05,
+      "loss": 0.7662,
+      "step": 3174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370970301664735e-05,
+      "loss": 0.7757,
+      "step": 3175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236659880256047e-05,
+      "loss": 0.7602,
+      "step": 3176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362226824313707e-05,
+      "loss": 0.7375,
+      "step": 3177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357854367809605e-05,
+      "loss": 0.8103,
+      "step": 3178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353481433933411e-05,
+      "loss": 0.7311,
+      "step": 3179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349108023570466e-05,
+      "loss": 0.775,
+      "step": 3180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234473413760622e-05,
+      "loss": 0.7264,
+      "step": 3181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340359776926205e-05,
+      "loss": 0.7352,
+      "step": 3182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335984942416062e-05,
+      "loss": 0.7856,
+      "step": 3183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331609634961515e-05,
+      "loss": 0.7482,
+      "step": 3184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327233855448397e-05,
+      "loss": 0.7925,
+      "step": 3185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322857604762626e-05,
+      "loss": 0.6714,
+      "step": 3186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318480883790217e-05,
+      "loss": 0.76,
+      "step": 3187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314103693417288e-05,
+      "loss": 0.7872,
+      "step": 3188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309726034530045e-05,
+      "loss": 0.6952,
+      "step": 3189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305347908014787e-05,
+      "loss": 0.7851,
+      "step": 3190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300969314757917e-05,
+      "loss": 0.7797,
+      "step": 3191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296590255645924e-05,
+      "loss": 0.7746,
+      "step": 3192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229221073156539e-05,
+      "loss": 0.7295,
+      "step": 3193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287830743403003e-05,
+      "loss": 0.6838,
+      "step": 3194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283450292045527e-05,
+      "loss": 0.7242,
+      "step": 3195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227906937837984e-05,
+      "loss": 0.7319,
+      "step": 3196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274688003292897e-05,
+      "loss": 0.8279,
+      "step": 3197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270306167671757e-05,
+      "loss": 0.7281,
+      "step": 3198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265923872403565e-05,
+      "loss": 0.7945,
+      "step": 3199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261541118375563e-05,
+      "loss": 0.7334,
+      "step": 3200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257157906475086e-05,
+      "loss": 0.7513,
+      "step": 3201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225277423758956e-05,
+      "loss": 0.2653,
+      "step": 3202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22483901126065e-05,
+      "loss": 0.6941,
+      "step": 3203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244005532413528e-05,
+      "loss": 0.6619,
+      "step": 3204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239620497898338e-05,
+      "loss": 0.74,
+      "step": 3205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235235009948734e-05,
+      "loss": 0.6826,
+      "step": 3206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230849069452592e-05,
+      "loss": 0.6777,
+      "step": 3207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226462677297904e-05,
+      "loss": 0.719,
+      "step": 3208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222075834372731e-05,
+      "loss": 0.7399,
+      "step": 3209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217688541565242e-05,
+      "loss": 0.234,
+      "step": 3210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213300799763683e-05,
+      "loss": 0.7261,
+      "step": 3211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208912609856404e-05,
+      "loss": 0.7133,
+      "step": 3212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204523972731836e-05,
+      "loss": 0.7414,
+      "step": 3213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200134889278505e-05,
+      "loss": 0.7659,
+      "step": 3214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219574536038503e-05,
+      "loss": 0.7432,
+      "step": 3215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191355386940109e-05,
+      "loss": 0.7131,
+      "step": 3216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186964969832545e-05,
+      "loss": 0.7061,
+      "step": 3217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182574109951217e-05,
+      "loss": 0.2244,
+      "step": 3218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217818280818511e-05,
+      "loss": 0.738,
+      "step": 3219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173791065423277e-05,
+      "loss": 0.7993,
+      "step": 3220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169398882554882e-05,
+      "loss": 0.7981,
+      "step": 3221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216500626046916e-05,
+      "loss": 0.815,
+      "step": 3222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2160613200055447e-05,
+      "loss": 0.6979,
+      "step": 3223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156219702203162e-05,
+      "loss": 0.7904,
+      "step": 3224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151825767801817e-05,
+      "loss": 0.7718,
+      "step": 3225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147431397741007e-05,
+      "loss": 0.7227,
+      "step": 3226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214303659291042e-05,
+      "loss": 0.7962,
+      "step": 3227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138641354199827e-05,
+      "loss": 0.7974,
+      "step": 3228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213424568249909e-05,
+      "loss": 0.7657,
+      "step": 3229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2129849578698162e-05,
+      "loss": 0.7539,
+      "step": 3230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125453043687077e-05,
+      "loss": 0.7135,
+      "step": 3231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121056078355963e-05,
+      "loss": 0.7524,
+      "step": 3232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116658683595023e-05,
+      "loss": 0.7498,
+      "step": 3233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112260860294564e-05,
+      "loss": 0.7832,
+      "step": 3234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107862609344963e-05,
+      "loss": 0.6769,
+      "step": 3235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21034639316367e-05,
+      "loss": 0.6887,
+      "step": 3236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099064828060323e-05,
+      "loss": 0.7134,
+      "step": 3237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094665299506487e-05,
+      "loss": 0.7983,
+      "step": 3238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090265346865915e-05,
+      "loss": 0.7495,
+      "step": 3239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085864971029427e-05,
+      "loss": 0.7596,
+      "step": 3240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208146417288792e-05,
+      "loss": 0.7003,
+      "step": 3241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077062953332386e-05,
+      "loss": 0.7406,
+      "step": 3242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072661313253896e-05,
+      "loss": 0.7203,
+      "step": 3243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068259253543606e-05,
+      "loss": 0.7244,
+      "step": 3244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063856775092761e-05,
+      "loss": 0.7673,
+      "step": 3245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059453878792687e-05,
+      "loss": 0.7603,
+      "step": 3246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055050565534799e-05,
+      "loss": 0.7247,
+      "step": 3247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050646836210587e-05,
+      "loss": 0.7366,
+      "step": 3248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046242691711638e-05,
+      "loss": 0.6593,
+      "step": 3249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041838132929615e-05,
+      "loss": 0.7906,
+      "step": 3250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037433160756267e-05,
+      "loss": 0.757,
+      "step": 3251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033027776083427e-05,
+      "loss": 0.7374,
+      "step": 3252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028621979803008e-05,
+      "loss": 0.8209,
+      "step": 3253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024215772807008e-05,
+      "loss": 0.2506,
+      "step": 3254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019809155987517e-05,
+      "loss": 0.7518,
+      "step": 3255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015402130236692e-05,
+      "loss": 0.7587,
+      "step": 3256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010994696446782e-05,
+      "loss": 0.7095,
+      "step": 3257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006586855510122e-05,
+      "loss": 0.7505,
+      "step": 3258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002178608319129e-05,
+      "loss": 0.6953,
+      "step": 3259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997769955766288e-05,
+      "loss": 0.7294,
+      "step": 3260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199336089874418e-05,
+      "loss": 0.7166,
+      "step": 3261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988951438145469e-05,
+      "loss": 0.7525,
+      "step": 3262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198454157486289e-05,
+      "loss": 0.7947,
+      "step": 3263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980131309789269e-05,
+      "loss": 0.8336,
+      "step": 3264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197572064381751e-05,
+      "loss": 0.7282,
+      "step": 3265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971309577840598e-05,
+      "loss": 0.7577,
+      "step": 3266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966898112751597e-05,
+      "loss": 0.8102,
+      "step": 3267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962486249443657e-05,
+      "loss": 0.7668,
+      "step": 3268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958073988809999e-05,
+      "loss": 0.7833,
+      "step": 3269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953661331743942e-05,
+      "loss": 0.712,
+      "step": 3270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949248279138863e-05,
+      "loss": 0.7503,
+      "step": 3271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944834831888241e-05,
+      "loss": 0.778,
+      "step": 3272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940420990885615e-05,
+      "loss": 0.7054,
+      "step": 3273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936006757024616e-05,
+      "loss": 0.7156,
+      "step": 3274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1931592131198952e-05,
+      "loss": 0.8244,
+      "step": 3275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192717711430241e-05,
+      "loss": 0.6964,
+      "step": 3276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922761707228855e-05,
+      "loss": 0.7797,
+      "step": 3277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918345910872237e-05,
+      "loss": 0.8421,
+      "step": 3278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913929726126574e-05,
+      "loss": 0.7613,
+      "step": 3279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909513153885968e-05,
+      "loss": 0.8416,
+      "step": 3280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905096195044603e-05,
+      "loss": 0.2121,
+      "step": 3281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900678850496734e-05,
+      "loss": 0.696,
+      "step": 3282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896261121136706e-05,
+      "loss": 0.7917,
+      "step": 3283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891843007858927e-05,
+      "loss": 0.7834,
+      "step": 3284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887424511557896e-05,
+      "loss": 0.7367,
+      "step": 3285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1883005633128175e-05,
+      "loss": 0.6858,
+      "step": 3286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878586373464421e-05,
+      "loss": 0.8016,
+      "step": 3287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187416673346135e-05,
+      "loss": 0.7589,
+      "step": 3288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869746714013772e-05,
+      "loss": 0.7206,
+      "step": 3289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186532631601656e-05,
+      "loss": 0.7509,
+      "step": 3290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860905540364676e-05,
+      "loss": 0.8152,
+      "step": 3291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856484387953147e-05,
+      "loss": 0.7567,
+      "step": 3292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852062859677078e-05,
+      "loss": 0.6888,
+      "step": 3293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847640956431661e-05,
+      "loss": 0.7313,
+      "step": 3294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843218679112151e-05,
+      "loss": 0.744,
+      "step": 3295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838796028613886e-05,
+      "loss": 0.7102,
+      "step": 3296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834373005832275e-05,
+      "loss": 0.7778,
+      "step": 3297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829949611662812e-05,
+      "loss": 0.7205,
+      "step": 3298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825525847001052e-05,
+      "loss": 0.7423,
+      "step": 3299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821101712742633e-05,
+      "loss": 0.771,
+      "step": 3300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816677209783268e-05,
+      "loss": 0.7386,
+      "step": 3301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1812252339018744e-05,
+      "loss": 0.7185,
+      "step": 3302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180782710134492e-05,
+      "loss": 0.751,
+      "step": 3303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803401497657736e-05,
+      "loss": 0.8049,
+      "step": 3304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1798975528853196e-05,
+      "loss": 0.7971,
+      "step": 3305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794549195827388e-05,
+      "loss": 0.7791,
+      "step": 3306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790122499476463e-05,
+      "loss": 0.2349,
+      "step": 3307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785695440696658e-05,
+      "loss": 0.7726,
+      "step": 3308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178126802038427e-05,
+      "loss": 0.7618,
+      "step": 3309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776840239435687e-05,
+      "loss": 0.7985,
+      "step": 3310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772412098747347e-05,
+      "loss": 0.7944,
+      "step": 3311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767983599215783e-05,
+      "loss": 0.8017,
+      "step": 3312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763554741737586e-05,
+      "loss": 0.7734,
+      "step": 3313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759125527209421e-05,
+      "loss": 0.7675,
+      "step": 3314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754695956528035e-05,
+      "loss": 0.6918,
+      "step": 3315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750266030590236e-05,
+      "loss": 0.694,
+      "step": 3316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745835750292914e-05,
+      "loss": 0.7696,
+      "step": 3317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741405116533015e-05,
+      "loss": 0.7531,
+      "step": 3318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736974130207582e-05,
+      "loss": 0.7488,
+      "step": 3319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732542792213696e-05,
+      "loss": 0.7627,
+      "step": 3320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1728111103448543e-05,
+      "loss": 0.7786,
+      "step": 3321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723679064809357e-05,
+      "loss": 0.7828,
+      "step": 3322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171924667719345e-05,
+      "loss": 0.7657,
+      "step": 3323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714813941498204e-05,
+      "loss": 0.7248,
+      "step": 3324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171038085862108e-05,
+      "loss": 0.7346,
+      "step": 3325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705947429459595e-05,
+      "loss": 0.8202,
+      "step": 3326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170151365491134e-05,
+      "loss": 0.7339,
+      "step": 3327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697079535873987e-05,
+      "loss": 0.7184,
+      "step": 3328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692645073245262e-05,
+      "loss": 0.7542,
+      "step": 3329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688210267922972e-05,
+      "loss": 0.6662,
+      "step": 3330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168377512080499e-05,
+      "loss": 0.7518,
+      "step": 3331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679339632789257e-05,
+      "loss": 0.741,
+      "step": 3332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674903804773778e-05,
+      "loss": 0.7578,
+      "step": 3333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670467637656643e-05,
+      "loss": 0.6823,
+      "step": 3334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666031132335989e-05,
+      "loss": 0.6806,
+      "step": 3335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166159428971004e-05,
+      "loss": 0.6816,
+      "step": 3336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657157110677077e-05,
+      "loss": 0.7416,
+      "step": 3337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652719596135457e-05,
+      "loss": 0.7316,
+      "step": 3338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648281746983599e-05,
+      "loss": 0.7463,
+      "step": 3339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164384356411999e-05,
+      "loss": 0.7245,
+      "step": 3340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639405048443184e-05,
+      "loss": 0.8027,
+      "step": 3341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634966200851812e-05,
+      "loss": 0.7624,
+      "step": 3342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630527022244558e-05,
+      "loss": 0.739,
+      "step": 3343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626087513520184e-05,
+      "loss": 0.699,
+      "step": 3344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621647675577511e-05,
+      "loss": 0.6916,
+      "step": 3345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617207509315433e-05,
+      "loss": 0.7555,
+      "step": 3346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161276701563291e-05,
+      "loss": 0.6835,
+      "step": 3347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608326195428958e-05,
+      "loss": 0.7067,
+      "step": 3348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603885049602674e-05,
+      "loss": 0.7342,
+      "step": 3349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1599443579053214e-05,
+      "loss": 0.7582,
+      "step": 3350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595001784679799e-05,
+      "loss": 0.7533,
+      "step": 3351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1590559667381713e-05,
+      "loss": 0.7099,
+      "step": 3352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586117228058314e-05,
+      "loss": 0.7864,
+      "step": 3353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581674467609013e-05,
+      "loss": 0.7491,
+      "step": 3354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577231386933298e-05,
+      "loss": 0.6983,
+      "step": 3355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572787986930718e-05,
+      "loss": 0.7854,
+      "step": 3356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568344268500884e-05,
+      "loss": 0.803,
+      "step": 3357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156390023254347e-05,
+      "loss": 0.6868,
+      "step": 3358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559455879958221e-05,
+      "loss": 0.7894,
+      "step": 3359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155501121164494e-05,
+      "loss": 0.6955,
+      "step": 3360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550566228503497e-05,
+      "loss": 0.7109,
+      "step": 3361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546120931433827e-05,
+      "loss": 0.7322,
+      "step": 3362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541675321335921e-05,
+      "loss": 0.6966,
+      "step": 3363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537229399109846e-05,
+      "loss": 0.741,
+      "step": 3364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532783165655724e-05,
+      "loss": 0.6623,
+      "step": 3365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528336621873734e-05,
+      "loss": 0.7491,
+      "step": 3366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523889768664129e-05,
+      "loss": 0.7502,
+      "step": 3367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519442606927225e-05,
+      "loss": 0.7404,
+      "step": 3368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514995137563389e-05,
+      "loss": 0.7833,
+      "step": 3369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510547361473065e-05,
+      "loss": 0.7737,
+      "step": 3370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506099279556743e-05,
+      "loss": 0.7506,
+      "step": 3371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501650892714993e-05,
+      "loss": 0.7198,
+      "step": 3372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497202201848425e-05,
+      "loss": 0.7178,
+      "step": 3373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492753207857734e-05,
+      "loss": 0.7317,
+      "step": 3374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488303911643656e-05,
+      "loss": 0.7385,
+      "step": 3375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483854314107006e-05,
+      "loss": 0.676,
+      "step": 3376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479404416148642e-05,
+      "loss": 0.7799,
+      "step": 3377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14749542186695e-05,
+      "loss": 0.7617,
+      "step": 3378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470503722570568e-05,
+      "loss": 0.8187,
+      "step": 3379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466052928752886e-05,
+      "loss": 0.7756,
+      "step": 3380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461601838117575e-05,
+      "loss": 0.7781,
+      "step": 3381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457150451565794e-05,
+      "loss": 0.774,
+      "step": 3382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452698769998785e-05,
+      "loss": 0.7418,
+      "step": 3383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448246794317824e-05,
+      "loss": 0.7429,
+      "step": 3384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443794525424272e-05,
+      "loss": 0.7216,
+      "step": 3385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439341964219526e-05,
+      "loss": 0.8596,
+      "step": 3386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434889111605059e-05,
+      "loss": 0.731,
+      "step": 3387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430435968482394e-05,
+      "loss": 0.7733,
+      "step": 3388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425982535753121e-05,
+      "loss": 0.7995,
+      "step": 3389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142152881431888e-05,
+      "loss": 0.7763,
+      "step": 3390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417074805081376e-05,
+      "loss": 0.7639,
+      "step": 3391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412620508942367e-05,
+      "loss": 0.6985,
+      "step": 3392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408165926803669e-05,
+      "loss": 0.6058,
+      "step": 3393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1403711059567164e-05,
+      "loss": 0.7891,
+      "step": 3394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399255908134783e-05,
+      "loss": 0.7402,
+      "step": 3395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394800473408518e-05,
+      "loss": 0.7819,
+      "step": 3396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390344756290416e-05,
+      "loss": 0.7605,
+      "step": 3397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385888757682592e-05,
+      "loss": 0.7316,
+      "step": 3398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381432478487198e-05,
+      "loss": 0.7176,
+      "step": 3399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376975919606459e-05,
+      "loss": 0.7476,
+      "step": 3400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137251908194265e-05,
+      "loss": 0.7434,
+      "step": 3401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368061966398108e-05,
+      "loss": 0.8218,
+      "step": 3402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363604573875215e-05,
+      "loss": 0.732,
+      "step": 3403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359146905276424e-05,
+      "loss": 0.7668,
+      "step": 3404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354688961504233e-05,
+      "loss": 0.7453,
+      "step": 3405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13502307434612e-05,
+      "loss": 0.7396,
+      "step": 3406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345772252049933e-05,
+      "loss": 0.7868,
+      "step": 3407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341313488173107e-05,
+      "loss": 0.7178,
+      "step": 3408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133685445273344e-05,
+      "loss": 0.7105,
+      "step": 3409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332395146633712e-05,
+      "loss": 0.7228,
+      "step": 3410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1327935570776756e-05,
+      "loss": 0.734,
+      "step": 3411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323475726065457e-05,
+      "loss": 0.6033,
+      "step": 3412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319015613402762e-05,
+      "loss": 0.7825,
+      "step": 3413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314555233691659e-05,
+      "loss": 0.7205,
+      "step": 3414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310094587835208e-05,
+      "loss": 0.8091,
+      "step": 3415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305633676736505e-05,
+      "loss": 0.7018,
+      "step": 3416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301172501298713e-05,
+      "loss": 0.7232,
+      "step": 3417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296711062425039e-05,
+      "loss": 0.7248,
+      "step": 3418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1292249361018751e-05,
+      "loss": 0.7259,
+      "step": 3419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287787397983165e-05,
+      "loss": 0.2492,
+      "step": 3420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128332517422165e-05,
+      "loss": 0.7282,
+      "step": 3421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278862690637635e-05,
+      "loss": 0.2539,
+      "step": 3422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127439994813459e-05,
+      "loss": 0.6913,
+      "step": 3423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269936947616047e-05,
+      "loss": 0.7587,
+      "step": 3424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265473689985589e-05,
+      "loss": 0.7578,
+      "step": 3425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261010176146844e-05,
+      "loss": 0.7364,
+      "step": 3426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256546407003497e-05,
+      "loss": 0.6706,
+      "step": 3427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252082383459289e-05,
+      "loss": 0.8283,
+      "step": 3428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247618106418004e-05,
+      "loss": 0.7371,
+      "step": 3429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243153576783481e-05,
+      "loss": 0.8107,
+      "step": 3430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238688795459612e-05,
+      "loss": 0.6619,
+      "step": 3431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123422376335034e-05,
+      "loss": 0.7378,
+      "step": 3432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229758481359655e-05,
+      "loss": 0.797,
+      "step": 3433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225292950391598e-05,
+      "loss": 0.6679,
+      "step": 3434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122082717135026e-05,
+      "loss": 0.7222,
+      "step": 3435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1216361145139793e-05,
+      "loss": 0.7578,
+      "step": 3436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211894872664383e-05,
+      "loss": 0.7825,
+      "step": 3437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207428354828282e-05,
+      "loss": 0.7396,
+      "step": 3438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120296159253577e-05,
+      "loss": 0.6711,
+      "step": 3439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11984945866912e-05,
+      "loss": 0.7432,
+      "step": 3440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194027338198957e-05,
+      "loss": 0.751,
+      "step": 3441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189559847963484e-05,
+      "loss": 0.8046,
+      "step": 3442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185092116889273e-05,
+      "loss": 0.7636,
+      "step": 3443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180624145880861e-05,
+      "loss": 0.7607,
+      "step": 3444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176155935842836e-05,
+      "loss": 0.7323,
+      "step": 3445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117168748767983e-05,
+      "loss": 0.7471,
+      "step": 3446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116721880229653e-05,
+      "loss": 0.7062,
+      "step": 3447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162749880597668e-05,
+      "loss": 0.6962,
+      "step": 3448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158280723488021e-05,
+      "loss": 0.748,
+      "step": 3449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153811331872418e-05,
+      "loss": 0.7247,
+      "step": 3450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149341706655736e-05,
+      "loss": 0.706,
+      "step": 3451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144871848742891e-05,
+      "loss": 0.7489,
+      "step": 3452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140401759038859e-05,
+      "loss": 0.6625,
+      "step": 3453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135931438448649e-05,
+      "loss": 0.6884,
+      "step": 3454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113146088787733e-05,
+      "loss": 0.6168,
+      "step": 3455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1126990108230007e-05,
+      "loss": 0.7793,
+      "step": 3456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122519100411842e-05,
+      "loss": 0.7653,
+      "step": 3457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111804786532803e-05,
+      "loss": 0.7782,
+      "step": 3458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111357640388382e-05,
+      "loss": 0.7247,
+      "step": 3459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1109104716984509e-05,
+      "loss": 0.7536,
+      "step": 3460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104632805535434e-05,
+      "loss": 0.8093,
+      "step": 3461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100160670441982e-05,
+      "loss": 0.7712,
+      "step": 3462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095688312609578e-05,
+      "loss": 0.7678,
+      "step": 3463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091215732943705e-05,
+      "loss": 0.6971,
+      "step": 3464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086742932349878e-05,
+      "loss": 0.7854,
+      "step": 3465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082269911733668e-05,
+      "loss": 0.7388,
+      "step": 3466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077796672000672e-05,
+      "loss": 0.7507,
+      "step": 3467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073323214056555e-05,
+      "loss": 0.7018,
+      "step": 3468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068849538807011e-05,
+      "loss": 0.6956,
+      "step": 3469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064375647157783e-05,
+      "loss": 0.7043,
+      "step": 3470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059901540014658e-05,
+      "loss": 0.7297,
+      "step": 3471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055427218283462e-05,
+      "loss": 0.6968,
+      "step": 3472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050952682870068e-05,
+      "loss": 0.7888,
+      "step": 3473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046477934680395e-05,
+      "loss": 0.7955,
+      "step": 3474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042002974620401e-05,
+      "loss": 0.681,
+      "step": 3475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103752780359609e-05,
+      "loss": 0.7423,
+      "step": 3476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10330524225135e-05,
+      "loss": 0.2592,
+      "step": 3477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028576832278728e-05,
+      "loss": 0.7427,
+      "step": 3478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024101033797898e-05,
+      "loss": 0.6708,
+      "step": 3479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019625027977182e-05,
+      "loss": 0.8128,
+      "step": 3480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015148815722793e-05,
+      "loss": 0.6562,
+      "step": 3481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101067239794099e-05,
+      "loss": 0.6628,
+      "step": 3482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1006195775538069e-05,
+      "loss": 0.6907,
+      "step": 3483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001718949420368e-05,
+      "loss": 0.7242,
+      "step": 3484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099724192049427e-05,
+      "loss": 0.7625,
+      "step": 3485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992764689666188e-05,
+      "loss": 0.7324,
+      "step": 3486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988287257842593e-05,
+      "loss": 0.7837,
+      "step": 3487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098380962592998e-05,
+      "loss": 0.6968,
+      "step": 3488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979331794834897e-05,
+      "loss": 0.7067,
+      "step": 3489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974853765463927e-05,
+      "loss": 0.7888,
+      "step": 3490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0970375538723692e-05,
+      "loss": 0.7261,
+      "step": 3491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965897115520858e-05,
+      "loss": 0.7938,
+      "step": 3492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961418496762122e-05,
+      "loss": 0.7229,
+      "step": 3493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956939683354235e-05,
+      "loss": 0.7011,
+      "step": 3494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0952460676203974e-05,
+      "loss": 0.7188,
+      "step": 3495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947981476218161e-05,
+      "loss": 0.7554,
+      "step": 3496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943502084303656e-05,
+      "loss": 0.2149,
+      "step": 3497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939022501367367e-05,
+      "loss": 0.7218,
+      "step": 3498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934542728316218e-05,
+      "loss": 0.7251,
+      "step": 3499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930062766057194e-05,
+      "loss": 0.2334,
+      "step": 3500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925582615497308e-05,
+      "loss": 0.7363,
+      "step": 3501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921102277543616e-05,
+      "loss": 0.6398,
+      "step": 3502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916621753103203e-05,
+      "loss": 0.6832,
+      "step": 3503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912141043083202e-05,
+      "loss": 0.7283,
+      "step": 3504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090766014839078e-05,
+      "loss": 0.7314,
+      "step": 3505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090317906993314e-05,
+      "loss": 0.7841,
+      "step": 3506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898697808617516e-05,
+      "loss": 0.7474,
+      "step": 3507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894216365351196e-05,
+      "loss": 0.8292,
+      "step": 3508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889734741041487e-05,
+      "loss": 0.7484,
+      "step": 3509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885252936595747e-05,
+      "loss": 0.6697,
+      "step": 3510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880770952921358e-05,
+      "loss": 0.7327,
+      "step": 3511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876288790925746e-05,
+      "loss": 0.7472,
+      "step": 3512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871806451516375e-05,
+      "loss": 0.7576,
+      "step": 3513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867323935600735e-05,
+      "loss": 0.7453,
+      "step": 3514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862841244086363e-05,
+      "loss": 0.7306,
+      "step": 3515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858358377880823e-05,
+      "loss": 0.7417,
+      "step": 3516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085387533789172e-05,
+      "loss": 0.7063,
+      "step": 3517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849392125026689e-05,
+      "loss": 0.7551,
+      "step": 3518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844908740193409e-05,
+      "loss": 0.7025,
+      "step": 3519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840425184299579e-05,
+      "loss": 0.7456,
+      "step": 3520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083594145825295e-05,
+      "loss": 0.7675,
+      "step": 3521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831457562961292e-05,
+      "loss": 0.6685,
+      "step": 3522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826973499332424e-05,
+      "loss": 0.7173,
+      "step": 3523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082248926827418e-05,
+      "loss": 0.8232,
+      "step": 3524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081800487069445e-05,
+      "loss": 0.7219,
+      "step": 3525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813520307501141e-05,
+      "loss": 0.2527,
+      "step": 3526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809035579602201e-05,
+      "loss": 0.759,
+      "step": 3527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080455068790561e-05,
+      "loss": 0.8074,
+      "step": 3528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800065633319382e-05,
+      "loss": 0.6853,
+      "step": 3529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079558041675156e-05,
+      "loss": 0.7071,
+      "step": 3530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791095039110222e-05,
+      "loss": 0.7039,
+      "step": 3531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786609501303484e-05,
+      "loss": 0.2638,
+      "step": 3532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782123804239487e-05,
+      "loss": 0.7719,
+      "step": 3533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777637948826405e-05,
+      "loss": 0.7979,
+      "step": 3534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773151935972448e-05,
+      "loss": 0.7207,
+      "step": 3535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768665766585858e-05,
+      "loss": 0.7456,
+      "step": 3536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07641794415749e-05,
+      "loss": 0.7519,
+      "step": 3537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759692961847887e-05,
+      "loss": 0.2394,
+      "step": 3538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755206328313143e-05,
+      "loss": 0.7227,
+      "step": 3539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075071954187904e-05,
+      "loss": 0.7392,
+      "step": 3540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746232603453973e-05,
+      "loss": 0.7405,
+      "step": 3541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741745513946369e-05,
+      "loss": 0.809,
+      "step": 3542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737258274264684e-05,
+      "loss": 0.8003,
+      "step": 3543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073277088531741e-05,
+      "loss": 0.7893,
+      "step": 3544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728283348013065e-05,
+      "loss": 0.7597,
+      "step": 3545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723795663260193e-05,
+      "loss": 0.7106,
+      "step": 3546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719307831967379e-05,
+      "loss": 0.7558,
+      "step": 3547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071481985504322e-05,
+      "loss": 0.7501,
+      "step": 3548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710331733396369e-05,
+      "loss": 0.8033,
+      "step": 3549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705843467935482e-05,
+      "loss": 0.7718,
+      "step": 3550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070135505956926e-05,
+      "loss": 0.7849,
+      "step": 3551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696866509206422e-05,
+      "loss": 0.725,
+      "step": 3552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692377817755728e-05,
+      "loss": 0.7824,
+      "step": 3553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687888986125956e-05,
+      "loss": 0.7745,
+      "step": 3554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068340001522592e-05,
+      "loss": 0.7326,
+      "step": 3555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678910905964452e-05,
+      "loss": 0.718,
+      "step": 3556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674421659250428e-05,
+      "loss": 0.7393,
+      "step": 3557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669932275992738e-05,
+      "loss": 0.6915,
+      "step": 3558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0665442757100301e-05,
+      "loss": 0.7745,
+      "step": 3559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660953103482074e-05,
+      "loss": 0.7435,
+      "step": 3560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656463316047028e-05,
+      "loss": 0.7023,
+      "step": 3561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651973395704169e-05,
+      "loss": 0.8014,
+      "step": 3562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0647483343362528e-05,
+      "loss": 0.7316,
+      "step": 3563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642993159931163e-05,
+      "loss": 0.7465,
+      "step": 3564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638502846319161e-05,
+      "loss": 0.6396,
+      "step": 3565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634012403435628e-05,
+      "loss": 0.799,
+      "step": 3566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629521832189699e-05,
+      "loss": 0.8119,
+      "step": 3567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625031133490545e-05,
+      "loss": 0.7658,
+      "step": 3568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0620540308247347e-05,
+      "loss": 0.6999,
+      "step": 3569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616049357369323e-05,
+      "loss": 0.7948,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061155828176571e-05,
+      "loss": 0.7234,
+      "step": 3571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607067082345777e-05,
+      "loss": 0.7351,
+      "step": 3572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060257576001881e-05,
+      "loss": 0.7248,
+      "step": 3573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598084315694127e-05,
+      "loss": 0.731,
+      "step": 3574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059359275028106e-05,
+      "loss": 0.7045,
+      "step": 3575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589101064688985e-05,
+      "loss": 0.8158,
+      "step": 3576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058460925982728e-05,
+      "loss": 0.7594,
+      "step": 3577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580117336605364e-05,
+      "loss": 0.6759,
+      "step": 3578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575625295932672e-05,
+      "loss": 0.6374,
+      "step": 3579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571133138718662e-05,
+      "loss": 0.7712,
+      "step": 3580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566640865872821e-05,
+      "loss": 0.7651,
+      "step": 3581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562148478304653e-05,
+      "loss": 0.7441,
+      "step": 3582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557655976923689e-05,
+      "loss": 0.7207,
+      "step": 3583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553163362639487e-05,
+      "loss": 0.7858,
+      "step": 3584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548670636361617e-05,
+      "loss": 0.7166,
+      "step": 3585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0544177798999683e-05,
+      "loss": 0.2763,
+      "step": 3586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539684851463304e-05,
+      "loss": 0.7247,
+      "step": 3587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535191794662122e-05,
+      "loss": 0.7052,
+      "step": 3588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530698629505808e-05,
+      "loss": 0.7497,
+      "step": 3589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526205356904047e-05,
+      "loss": 0.78,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521711977766551e-05,
+      "loss": 0.7467,
+      "step": 3591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517218493003048e-05,
+      "loss": 0.7509,
+      "step": 3592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051272490352329e-05,
+      "loss": 0.7374,
+      "step": 3593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508231210237056e-05,
+      "loss": 0.7318,
+      "step": 3594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503737414054133e-05,
+      "loss": 0.7246,
+      "step": 3595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499243515884348e-05,
+      "loss": 0.8033,
+      "step": 3596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494749516637526e-05,
+      "loss": 0.7922,
+      "step": 3597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490255417223531e-05,
+      "loss": 0.7533,
+      "step": 3598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485761218552238e-05,
+      "loss": 0.7671,
+      "step": 3599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481266921533541e-05,
+      "loss": 0.728,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047677252707736e-05,
+      "loss": 0.2243,
+      "step": 3601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472278036093635e-05,
+      "loss": 0.7004,
+      "step": 3602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467783449492315e-05,
+      "loss": 0.7518,
+      "step": 3603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463288768183385e-05,
+      "loss": 0.7647,
+      "step": 3604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045879399307683e-05,
+      "loss": 0.7378,
+      "step": 3605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045429912508267e-05,
+      "loss": 0.6282,
+      "step": 3606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449804165110935e-05,
+      "loss": 0.6921,
+      "step": 3607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0445309114071678e-05,
+      "loss": 0.7519,
+      "step": 3608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440813972874968e-05,
+      "loss": 0.6701,
+      "step": 3609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436318742430895e-05,
+      "loss": 0.7102,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431823423649563e-05,
+      "loss": 0.7567,
+      "step": 3611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427328017441094e-05,
+      "loss": 0.7799,
+      "step": 3612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422832524715634e-05,
+      "loss": 0.8091,
+      "step": 3613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418336946383339e-05,
+      "loss": 0.7707,
+      "step": 3614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413841283354387e-05,
+      "loss": 0.7072,
+      "step": 3615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040934553653897e-05,
+      "loss": 0.7699,
+      "step": 3616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404849706847302e-05,
+      "loss": 0.7481,
+      "step": 3617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400353795189609e-05,
+      "loss": 0.6615,
+      "step": 3618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395857802476133e-05,
+      "loss": 0.7592,
+      "step": 3619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391361729617134e-05,
+      "loss": 0.7004,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038686557752289e-05,
+      "loss": 0.8451,
+      "step": 3621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382369347103696e-05,
+      "loss": 0.709,
+      "step": 3622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377873039269859e-05,
+      "loss": 0.7333,
+      "step": 3623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0373376654931701e-05,
+      "loss": 0.7015,
+      "step": 3624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368880194999561e-05,
+      "loss": 0.7899,
+      "step": 3625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0364383660383801e-05,
+      "loss": 0.7224,
+      "step": 3626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359887051994784e-05,
+      "loss": 0.6837,
+      "step": 3627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355390370742899e-05,
+      "loss": 0.6689,
+      "step": 3628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350893617538542e-05,
+      "loss": 0.7072,
+      "step": 3629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346396793292131e-05,
+      "loss": 0.7021,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341899898914094e-05,
+      "loss": 0.697,
+      "step": 3631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337402935314872e-05,
+      "loss": 0.6899,
+      "step": 3632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332905903404924e-05,
+      "loss": 0.7576,
+      "step": 3633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032840880409472e-05,
+      "loss": 0.7487,
+      "step": 3634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0323911638294743e-05,
+      "loss": 0.7628,
+      "step": 3635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319414406915493e-05,
+      "loss": 0.7108,
+      "step": 3636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0314917110867482e-05,
+      "loss": 0.7086,
+      "step": 3637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310419751061233e-05,
+      "loss": 0.7126,
+      "step": 3638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305922328407283e-05,
+      "loss": 0.7493,
+      "step": 3639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301424843816182e-05,
+      "loss": 0.6975,
+      "step": 3640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029692729819849e-05,
+      "loss": 0.8885,
+      "step": 3641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0292429692464785e-05,
+      "loss": 0.7562,
+      "step": 3642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287932027525653e-05,
+      "loss": 0.793,
+      "step": 3643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283434304291696e-05,
+      "loss": 0.7299,
+      "step": 3644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027893652367352e-05,
+      "loss": 0.7489,
+      "step": 3645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274438686581748e-05,
+      "loss": 0.2238,
+      "step": 3646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269940793927016e-05,
+      "loss": 0.7377,
+      "step": 3647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265442846619966e-05,
+      "loss": 0.7371,
+      "step": 3648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260944845571259e-05,
+      "loss": 0.7565,
+      "step": 3649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256446791691556e-05,
+      "loss": 0.7208,
+      "step": 3650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251948685891543e-05,
+      "loss": 0.741,
+      "step": 3651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247450529081901e-05,
+      "loss": 0.7398,
+      "step": 3652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242952322173332e-05,
+      "loss": 0.7431,
+      "step": 3653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023845406607654e-05,
+      "loss": 0.7494,
+      "step": 3654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233955761702248e-05,
+      "loss": 0.7432,
+      "step": 3655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229457409961183e-05,
+      "loss": 0.718,
+      "step": 3656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224959011764086e-05,
+      "loss": 0.7119,
+      "step": 3657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220460568021701e-05,
+      "loss": 0.7067,
+      "step": 3658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215962079644783e-05,
+      "loss": 0.758,
+      "step": 3659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211463547544102e-05,
+      "loss": 0.8601,
+      "step": 3660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020696497263043e-05,
+      "loss": 0.7604,
+      "step": 3661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202466355814553e-05,
+      "loss": 0.7094,
+      "step": 3662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197967698007256e-05,
+      "loss": 0.7642,
+      "step": 3663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193469000119352e-05,
+      "loss": 0.7506,
+      "step": 3664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188970263061633e-05,
+      "loss": 0.7004,
+      "step": 3665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184471487744925e-05,
+      "loss": 0.727,
+      "step": 3666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179972675080047e-05,
+      "loss": 0.7565,
+      "step": 3667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175473825977835e-05,
+      "loss": 0.7783,
+      "step": 3668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170974941349126e-05,
+      "loss": 0.7319,
+      "step": 3669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166476022104762e-05,
+      "loss": 0.6831,
+      "step": 3670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161977069155604e-05,
+      "loss": 0.7525,
+      "step": 3671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157478083412507e-05,
+      "loss": 0.7522,
+      "step": 3672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152979065786332e-05,
+      "loss": 0.6723,
+      "step": 3673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014848001718796e-05,
+      "loss": 0.6821,
+      "step": 3674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143980938528266e-05,
+      "loss": 0.7712,
+      "step": 3675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139481830718139e-05,
+      "loss": 0.7495,
+      "step": 3676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134982694668465e-05,
+      "loss": 0.7804,
+      "step": 3677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130483531290145e-05,
+      "loss": 0.8296,
+      "step": 3678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012598434149408e-05,
+      "loss": 0.7121,
+      "step": 3679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121485126191174e-05,
+      "loss": 0.646,
+      "step": 3680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116985886292346e-05,
+      "loss": 0.6726,
+      "step": 3681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112486622708508e-05,
+      "loss": 0.8177,
+      "step": 3682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107987336350588e-05,
+      "loss": 0.6969,
+      "step": 3683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0103488028129511e-05,
+      "loss": 0.7352,
+      "step": 3684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009898869895621e-05,
+      "loss": 0.7128,
+      "step": 3685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094489349741616e-05,
+      "loss": 0.7107,
+      "step": 3686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089989981396672e-05,
+      "loss": 0.7687,
+      "step": 3687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085490594832324e-05,
+      "loss": 0.7993,
+      "step": 3688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080991190959517e-05,
+      "loss": 0.7323,
+      "step": 3689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076491770689203e-05,
+      "loss": 0.7992,
+      "step": 3690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071992334932337e-05,
+      "loss": 0.7319,
+      "step": 3691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067492884599874e-05,
+      "loss": 0.2213,
+      "step": 3692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062993420602776e-05,
+      "loss": 0.6725,
+      "step": 3693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058493943852006e-05,
+      "loss": 0.7785,
+      "step": 3694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005399445525853e-05,
+      "loss": 0.7392,
+      "step": 3695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049494955733317e-05,
+      "loss": 0.7769,
+      "step": 3696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044995446187333e-05,
+      "loss": 0.8021,
+      "step": 3697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004049592753156e-05,
+      "loss": 0.7101,
+      "step": 3698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003599640067696e-05,
+      "loss": 0.7772,
+      "step": 3699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031496866534518e-05,
+      "loss": 0.6813,
+      "step": 3700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026997326015206e-05,
+      "loss": 0.675,
+      "step": 3701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002249778003001e-05,
+      "loss": 0.7615,
+      "step": 3702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00179982294899e-05,
+      "loss": 0.748,
+      "step": 3703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013498675305863e-05,
+      "loss": 0.7001,
+      "step": 3704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008999118388883e-05,
+      "loss": 0.7672,
+      "step": 3705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004499559649931e-05,
+      "loss": 0.7737,
+      "step": 3706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.7594,
+      "step": 3707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99550044035007e-06,
+      "loss": 0.8599,
+      "step": 3708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991000881611122e-06,
+      "loss": 0.7319,
+      "step": 3709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986501324694139e-06,
+      "loss": 0.7739,
+      "step": 3710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982001770510102e-06,
+      "loss": 0.718,
+      "step": 3711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977502219969994e-06,
+      "loss": 0.728,
+      "step": 3712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973002673984793e-06,
+      "loss": 0.6863,
+      "step": 3713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968503133465485e-06,
+      "loss": 0.7503,
+      "step": 3714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964003599323044e-06,
+      "loss": 0.746,
+      "step": 3715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959504072468444e-06,
+      "loss": 0.7172,
+      "step": 3716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95500455381267e-06,
+      "loss": 0.7077,
+      "step": 3717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950505044266685e-06,
+      "loss": 0.7286,
+      "step": 3718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946005544741474e-06,
+      "loss": 0.7182,
+      "step": 3719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941506056147994e-06,
+      "loss": 0.726,
+      "step": 3720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937006579397229e-06,
+      "loss": 0.7213,
+      "step": 3721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93250711540013e-06,
+      "loss": 0.7453,
+      "step": 3722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928007665067668e-06,
+      "loss": 0.7141,
+      "step": 3723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9235082293108e-06,
+      "loss": 0.6563,
+      "step": 3724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919008809040485e-06,
+      "loss": 0.5991,
+      "step": 3725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91450940516768e-06,
+      "loss": 0.7397,
+      "step": 3726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910010018603331e-06,
+      "loss": 0.6882,
+      "step": 3727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905510650258387e-06,
+      "loss": 0.6994,
+      "step": 3728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901011301043794e-06,
+      "loss": 0.7107,
+      "step": 3729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89651197187049e-06,
+      "loss": 0.2691,
+      "step": 3730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892012663649414e-06,
+      "loss": 0.7785,
+      "step": 3731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887513377291492e-06,
+      "loss": 0.7459,
+      "step": 3732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883014113707658e-06,
+      "loss": 0.7298,
+      "step": 3733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878514873808828e-06,
+      "loss": 0.8212,
+      "step": 3734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874015658505924e-06,
+      "loss": 0.6216,
+      "step": 3735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869516468709859e-06,
+      "loss": 0.7509,
+      "step": 3736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865017305331536e-06,
+      "loss": 0.6803,
+      "step": 3737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860518169281865e-06,
+      "loss": 0.7657,
+      "step": 3738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856019061471734e-06,
+      "loss": 0.7583,
+      "step": 3739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851519982812041e-06,
+      "loss": 0.7589,
+      "step": 3740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847020934213673e-06,
+      "loss": 0.7558,
+      "step": 3741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842521916587497e-06,
+      "loss": 0.7044,
+      "step": 3742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8380229308444e-06,
+      "loss": 0.677,
+      "step": 3743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83352397789524e-06,
+      "loss": 0.7618,
+      "step": 3744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829025058650879e-06,
+      "loss": 0.7432,
+      "step": 3745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824526174022167e-06,
+      "loss": 0.7295,
+      "step": 3746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820027324919954e-06,
+      "loss": 0.7133,
+      "step": 3747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81552851225508e-06,
+      "loss": 0.7067,
+      "step": 3748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811029736938372e-06,
+      "loss": 0.6676,
+      "step": 3749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806530999880655e-06,
+      "loss": 0.7248,
+      "step": 3750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802032301992743e-06,
+      "loss": 0.7308,
+      "step": 3751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79753364418545e-06,
+      "loss": 0.7105,
+      "step": 3752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79303502736957e-06,
+      "loss": 0.7697,
+      "step": 3753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7885364524559e-06,
+      "loss": 0.705,
+      "step": 3754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784037920355219e-06,
+      "loss": 0.8069,
+      "step": 3755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779539431978302e-06,
+      "loss": 0.7328,
+      "step": 3756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775040988235918e-06,
+      "loss": 0.7355,
+      "step": 3757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770542590038817e-06,
+      "loss": 0.8243,
+      "step": 3758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766044238297754e-06,
+      "loss": 0.7781,
+      "step": 3759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761545933923462e-06,
+      "loss": 0.7945,
+      "step": 3760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757047677826673e-06,
+      "loss": 0.7166,
+      "step": 3761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752549470918105e-06,
+      "loss": 0.6804,
+      "step": 3762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748051314108458e-06,
+      "loss": 0.7476,
+      "step": 3763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743553208308447e-06,
+      "loss": 0.7796,
+      "step": 3764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739055154428741e-06,
+      "loss": 0.7317,
+      "step": 3765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734557153380037e-06,
+      "loss": 0.629,
+      "step": 3766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73005920607299e-06,
+      "loss": 0.827,
+      "step": 3767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725561313418258e-06,
+      "loss": 0.8092,
+      "step": 3768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721063476326487e-06,
+      "loss": 0.711,
+      "step": 3769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71656569570831e-06,
+      "loss": 0.707,
+      "step": 3770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71206797247435e-06,
+      "loss": 0.7957,
+      "step": 3771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707570307535218e-06,
+      "loss": 0.7554,
+      "step": 3772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703072701801514e-06,
+      "loss": 0.7075,
+      "step": 3773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698575156183825e-06,
+      "loss": 0.7484,
+      "step": 3774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694077671592722e-06,
+      "loss": 0.721,
+      "step": 3775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689580248938772e-06,
+      "loss": 0.7351,
+      "step": 3776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68508288913252e-06,
+      "loss": 0.6861,
+      "step": 3777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680585593084508e-06,
+      "loss": 0.8156,
+      "step": 3778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.676088361705258e-06,
+      "loss": 0.6677,
+      "step": 3779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671591195905283e-06,
+      "loss": 0.6347,
+      "step": 3780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.667094096595083e-06,
+      "loss": 0.6823,
+      "step": 3781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66259706468513e-06,
+      "loss": 0.7441,
+      "step": 3782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658100101085911e-06,
+      "loss": 0.6852,
+      "step": 3783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65360320670787e-06,
+      "loss": 0.2403,
+      "step": 3784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649106382461461e-06,
+      "loss": 0.7165,
+      "step": 3785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644609629257102e-06,
+      "loss": 0.7108,
+      "step": 3786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64011294800522e-06,
+      "loss": 0.6061,
+      "step": 3787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.635616339616204e-06,
+      "loss": 0.6963,
+      "step": 3788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63111980500044e-06,
+      "loss": 0.8181,
+      "step": 3789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626623345068304e-06,
+      "loss": 0.7586,
+      "step": 3790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622126960730144e-06,
+      "loss": 0.7166,
+      "step": 3791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617630652896307e-06,
+      "loss": 0.2462,
+      "step": 3792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61313442247711e-06,
+      "loss": 0.7533,
+      "step": 3793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60863827038287e-06,
+      "loss": 0.6919,
+      "step": 3794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604142197523872e-06,
+      "loss": 0.7576,
+      "step": 3795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599646204810394e-06,
+      "loss": 0.7757,
+      "step": 3796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595150293152701e-06,
+      "loss": 0.7968,
+      "step": 3797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590654463461032e-06,
+      "loss": 0.7785,
+      "step": 3798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586158716645617e-06,
+      "loss": 0.7663,
+      "step": 3799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581663053616663e-06,
+      "loss": 0.7732,
+      "step": 3800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57716747528437e-06,
+      "loss": 0.7798,
+      "step": 3801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572671982558908e-06,
+      "loss": 0.7491,
+      "step": 3802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56817657635044e-06,
+      "loss": 0.8162,
+      "step": 3803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563681257569108e-06,
+      "loss": 0.6697,
+      "step": 3804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559186027125032e-06,
+      "loss": 0.6458,
+      "step": 3805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554690885928324e-06,
+      "loss": 0.7186,
+      "step": 3806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550195834889072e-06,
+      "loss": 0.7575,
+      "step": 3807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545700874917334e-06,
+      "loss": 0.6247,
+      "step": 3808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541206006923175e-06,
+      "loss": 0.8071,
+      "step": 3809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536711231816619e-06,
+      "loss": 0.727,
+      "step": 3810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532216550507689e-06,
+      "loss": 0.6982,
+      "step": 3811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527721963906366e-06,
+      "loss": 0.7669,
+      "step": 3812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523227472922643e-06,
+      "loss": 0.7858,
+      "step": 3813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518733078466464e-06,
+      "loss": 0.7385,
+      "step": 3814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514238781447767e-06,
+      "loss": 0.7096,
+      "step": 3815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509744582776474e-06,
+      "loss": 0.7679,
+      "step": 3816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505250483362475e-06,
+      "loss": 0.7181,
+      "step": 3817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500756484115655e-06,
+      "loss": 0.7742,
+      "step": 3818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496262585945865e-06,
+      "loss": 0.6856,
+      "step": 3819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491768789762948e-06,
+      "loss": 0.7185,
+      "step": 3820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487275096476712e-06,
+      "loss": 0.746,
+      "step": 3821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482781506996956e-06,
+      "loss": 0.694,
+      "step": 3822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478288022233454e-06,
+      "loss": 0.7457,
+      "step": 3823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473794643095954e-06,
+      "loss": 0.6913,
+      "step": 3824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469301370494194e-06,
+      "loss": 0.7804,
+      "step": 3825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46480820533788e-06,
+      "loss": 0.6779,
+      "step": 3826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4603151485367e-06,
+      "loss": 0.752,
+      "step": 3827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455822201000322e-06,
+      "loss": 0.7909,
+      "step": 3828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451329363638385e-06,
+      "loss": 0.6996,
+      "step": 3829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44683663736052e-06,
+      "loss": 0.7173,
+      "step": 3830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442344023076311e-06,
+      "loss": 0.7578,
+      "step": 3831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437851521695352e-06,
+      "loss": 0.7373,
+      "step": 3832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43335913412718e-06,
+      "loss": 0.667,
+      "step": 3833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42886686128134e-06,
+      "loss": 0.7631,
+      "step": 3834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424374704067331e-06,
+      "loss": 0.7509,
+      "step": 3835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419882663394639e-06,
+      "loss": 0.7861,
+      "step": 3836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415390740172723e-06,
+      "loss": 0.7736,
+      "step": 3837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410898935311019e-06,
+      "loss": 0.6984,
+      "step": 3838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406407249718941e-06,
+      "loss": 0.7433,
+      "step": 3839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40191568430588e-06,
+      "loss": 0.7376,
+      "step": 3840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397424239981193e-06,
+      "loss": 0.7684,
+      "step": 3841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392932917654226e-06,
+      "loss": 0.7405,
+      "step": 3842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388441718234292e-06,
+      "loss": 0.6815,
+      "step": 3843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38395064263068e-06,
+      "loss": 0.6874,
+      "step": 3844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379459691752655e-06,
+      "loss": 0.7553,
+      "step": 3845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374968866509459e-06,
+      "loss": 0.7383,
+      "step": 3846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370478167810304e-06,
+      "loss": 0.7314,
+      "step": 3847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365987596564376e-06,
+      "loss": 0.7684,
+      "step": 3848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361497153680844e-06,
+      "loss": 0.8216,
+      "step": 3849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357006840068836e-06,
+      "loss": 0.7487,
+      "step": 3850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.352516656637475e-06,
+      "loss": 0.7471,
+      "step": 3851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348026604295831e-06,
+      "loss": 0.7173,
+      "step": 3852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343536683952975e-06,
+      "loss": 0.7323,
+      "step": 3853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33904689651793e-06,
+      "loss": 0.7478,
+      "step": 3854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3345572428997e-06,
+      "loss": 0.7363,
+      "step": 3855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330067724007267e-06,
+      "loss": 0.7005,
+      "step": 3856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325578340749575e-06,
+      "loss": 0.7886,
+      "step": 3857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321089094035551e-06,
+      "loss": 0.6407,
+      "step": 3858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316599984774085e-06,
+      "loss": 0.754,
+      "step": 3859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312111013874046e-06,
+      "loss": 0.6929,
+      "step": 3860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307622182244276e-06,
+      "loss": 0.7655,
+      "step": 3861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30313349079358e-06,
+      "loss": 0.7357,
+      "step": 3862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298644940430745e-06,
+      "loss": 0.7753,
+      "step": 3863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29415653206452e-06,
+      "loss": 0.6589,
+      "step": 3864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289668266603633e-06,
+      "loss": 0.6839,
+      "step": 3865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285180144956778e-06,
+      "loss": 0.8078,
+      "step": 3866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280692168032626e-06,
+      "loss": 0.6522,
+      "step": 3867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27620433673981e-06,
+      "loss": 0.7354,
+      "step": 3868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271716651986938e-06,
+      "loss": 0.7847,
+      "step": 3869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267229114682594e-06,
+      "loss": 0.7689,
+      "step": 3870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262741725735318e-06,
+      "loss": 0.7839,
+      "step": 3871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258254486053634e-06,
+      "loss": 0.7283,
+      "step": 3872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253767396546029e-06,
+      "loss": 0.7295,
+      "step": 3873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249280458120963e-06,
+      "loss": 0.7231,
+      "step": 3874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244793671686862e-06,
+      "loss": 0.7962,
+      "step": 3875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240307038152117e-06,
+      "loss": 0.6802,
+      "step": 3876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.235820558425104e-06,
+      "loss": 0.6911,
+      "step": 3877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231334233414144e-06,
+      "loss": 0.7043,
+      "step": 3878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226848064027557e-06,
+      "loss": 0.6847,
+      "step": 3879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2223620511736e-06,
+      "loss": 0.6853,
+      "step": 3880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217876195760518e-06,
+      "loss": 0.7266,
+      "step": 3881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213390498696521e-06,
+      "loss": 0.7782,
+      "step": 3882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20890496088978e-06,
+      "loss": 0.2113,
+      "step": 3883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204419583248444e-06,
+      "loss": 0.669,
+      "step": 3884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19993436668062e-06,
+      "loss": 0.728,
+      "step": 3885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195449312094393e-06,
+      "loss": 0.7319,
+      "step": 3886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1909644203978e-06,
+      "loss": 0.7734,
+      "step": 3887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18647969249886e-06,
+      "loss": 0.8052,
+      "step": 3888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181995129305554e-06,
+      "loss": 0.7612,
+      "step": 3889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17751073172582e-06,
+      "loss": 0.7164,
+      "step": 3890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173026500667581e-06,
+      "loss": 0.7657,
+      "step": 3891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16854243703871e-06,
+      "loss": 0.74,
+      "step": 3892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164058541747053e-06,
+      "loss": 0.7801,
+      "step": 3893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159574815700426e-06,
+      "loss": 0.7729,
+      "step": 3894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155091259806594e-06,
+      "loss": 0.7353,
+      "step": 3895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150607874973316e-06,
+      "loss": 0.6925,
+      "step": 3896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146124662108282e-06,
+      "loss": 0.7158,
+      "step": 3897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14164162211918e-06,
+      "loss": 0.7519,
+      "step": 3898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137158755913638e-06,
+      "loss": 0.73,
+      "step": 3899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132676064399267e-06,
+      "loss": 0.7471,
+      "step": 3900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12819354848363e-06,
+      "loss": 0.6711,
+      "step": 3901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123711209074256e-06,
+      "loss": 0.7664,
+      "step": 3902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119229047078645e-06,
+      "loss": 0.7081,
+      "step": 3903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114747063404256e-06,
+      "loss": 0.7752,
+      "step": 3904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110265258958517e-06,
+      "loss": 0.7032,
+      "step": 3905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.105783634648808e-06,
+      "loss": 0.6923,
+      "step": 3906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101302191382487e-06,
+      "loss": 0.7076,
+      "step": 3907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096820930066866e-06,
+      "loss": 0.7369,
+      "step": 3908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092339851609222e-06,
+      "loss": 0.7446,
+      "step": 3909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0878589569168e-06,
+      "loss": 0.7078,
+      "step": 3910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083378246896797e-06,
+      "loss": 0.7353,
+      "step": 3911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078897722456387e-06,
+      "loss": 0.6711,
+      "step": 3912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074417384502692e-06,
+      "loss": 0.746,
+      "step": 3913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069937233942808e-06,
+      "loss": 0.7254,
+      "step": 3914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065457271683787e-06,
+      "loss": 0.7189,
+      "step": 3915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060977498632637e-06,
+      "loss": 0.8075,
+      "step": 3916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056497915696346e-06,
+      "loss": 0.6126,
+      "step": 3917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052018523781839e-06,
+      "loss": 0.7574,
+      "step": 3918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047539323796031e-06,
+      "loss": 0.7409,
+      "step": 3919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04306031664577e-06,
+      "loss": 0.6435,
+      "step": 3920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03858150323788e-06,
+      "loss": 0.7777,
+      "step": 3921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034102884479148e-06,
+      "loss": 0.6973,
+      "step": 3922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.029624461276311e-06,
+      "loss": 0.7606,
+      "step": 3923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025146234536076e-06,
+      "loss": 0.6686,
+      "step": 3924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020668205165104e-06,
+      "loss": 0.683,
+      "step": 3925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016190374070024e-06,
+      "loss": 0.7821,
+      "step": 3926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011712742157412e-06,
+      "loss": 0.6322,
+      "step": 3927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007235310333813e-06,
+      "loss": 0.6344,
+      "step": 3928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002758079505736e-06,
+      "loss": 0.6609,
+      "step": 3929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998281050579634e-06,
+      "loss": 0.7159,
+      "step": 3930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993804224461933e-06,
+      "loss": 0.6919,
+      "step": 3931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98932760205901e-06,
+      "loss": 0.7702,
+      "step": 3932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984851184277209e-06,
+      "loss": 0.6947,
+      "step": 3933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980374972022823e-06,
+      "loss": 0.7644,
+      "step": 3934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975898966202105e-06,
+      "loss": 0.7671,
+      "step": 3935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971423167721277e-06,
+      "loss": 0.2293,
+      "step": 3936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9669475774865e-06,
+      "loss": 0.7451,
+      "step": 3937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962472196403914e-06,
+      "loss": 0.7307,
+      "step": 3938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9579970253796e-06,
+      "loss": 0.6555,
+      "step": 3939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953522065319608e-06,
+      "loss": 0.7549,
+      "step": 3940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949047317129937e-06,
+      "loss": 0.6906,
+      "step": 3941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94457278171654e-06,
+      "loss": 0.7395,
+      "step": 3942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940098459985347e-06,
+      "loss": 0.7262,
+      "step": 3943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935624352842217e-06,
+      "loss": 0.7411,
+      "step": 3944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.931150461192992e-06,
+      "loss": 0.833,
+      "step": 3945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926676785943445e-06,
+      "loss": 0.7264,
+      "step": 3946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92220332799933e-06,
+      "loss": 0.7799,
+      "step": 3947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917730088266339e-06,
+      "loss": 0.2439,
+      "step": 3948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913257067650123e-06,
+      "loss": 0.7531,
+      "step": 3949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908784267056298e-06,
+      "loss": 0.6622,
+      "step": 3950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904311687390422e-06,
+      "loss": 0.7399,
+      "step": 3951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899839329558022e-06,
+      "loss": 0.6541,
+      "step": 3952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895367194464568e-06,
+      "loss": 0.7428,
+      "step": 3953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890895283015495e-06,
+      "loss": 0.68,
+      "step": 3954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886423596116184e-06,
+      "loss": 0.7425,
+      "step": 3955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881952134671973e-06,
+      "loss": 0.8414,
+      "step": 3956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877480899588161e-06,
+      "loss": 0.7595,
+      "step": 3957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.873009891769993e-06,
+      "loss": 0.7537,
+      "step": 3958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.868539112122671e-06,
+      "loss": 0.6969,
+      "step": 3959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864068561551356e-06,
+      "loss": 0.7273,
+      "step": 3960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859598240961143e-06,
+      "loss": 0.7491,
+      "step": 3961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855128151257114e-06,
+      "loss": 0.7113,
+      "step": 3962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850658293344266e-06,
+      "loss": 0.7015,
+      "step": 3963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846188668127585e-06,
+      "loss": 0.7998,
+      "step": 3964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84171927651198e-06,
+      "loss": 0.7352,
+      "step": 3965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837250119402336e-06,
+      "loss": 0.7817,
+      "step": 3966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832781197703474e-06,
+      "loss": 0.7161,
+      "step": 3967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828312512320173e-06,
+      "loss": 0.725,
+      "step": 3968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82384406415717e-06,
+      "loss": 0.6713,
+      "step": 3969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81937585411914e-06,
+      "loss": 0.7233,
+      "step": 3970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81490788311073e-06,
+      "loss": 0.7568,
+      "step": 3971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810440152036518e-06,
+      "loss": 0.7371,
+      "step": 3972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805972661801047e-06,
+      "loss": 0.7083,
+      "step": 3973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801505413308806e-06,
+      "loss": 0.2608,
+      "step": 3974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797038407464232e-06,
+      "loss": 0.7623,
+      "step": 3975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792571645171723e-06,
+      "loss": 0.7547,
+      "step": 3976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788105127335615e-06,
+      "loss": 0.7491,
+      "step": 3977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783638854860208e-06,
+      "loss": 0.6649,
+      "step": 3978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77917282864974e-06,
+      "loss": 0.6966,
+      "step": 3979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774707049608406e-06,
+      "loss": 0.6782,
+      "step": 3980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770241518640352e-06,
+      "loss": 0.7488,
+      "step": 3981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765776236649663e-06,
+      "loss": 0.7165,
+      "step": 3982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761311204540392e-06,
+      "loss": 0.7295,
+      "step": 3983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75684642321652e-06,
+      "loss": 0.252,
+      "step": 3984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752381893582001e-06,
+      "loss": 0.7861,
+      "step": 3985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747917616540711e-06,
+      "loss": 0.7296,
+      "step": 3986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743453592996504e-06,
+      "loss": 0.7576,
+      "step": 3987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738989823853161e-06,
+      "loss": 0.7187,
+      "step": 3988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734526310014415e-06,
+      "loss": 0.757,
+      "step": 3989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730063052383955e-06,
+      "loss": 0.7498,
+      "step": 3990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725600051865411e-06,
+      "loss": 0.6563,
+      "step": 3991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721137309362368e-06,
+      "loss": 0.6964,
+      "step": 3992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716674825778353e-06,
+      "loss": 0.2409,
+      "step": 3993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712212602016838e-06,
+      "loss": 0.7739,
+      "step": 3994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.707750638981254e-06,
+      "loss": 0.7573,
+      "step": 3995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703288937574963e-06,
+      "loss": 0.7429,
+      "step": 3996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69882749870129e-06,
+      "loss": 0.252,
+      "step": 3997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694366323263495e-06,
+      "loss": 0.7526,
+      "step": 3998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689905412164794e-06,
+      "loss": 0.7584,
+      "step": 3999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685444766308343e-06,
+      "loss": 0.7958,
+      "step": 4000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680984386597242e-06,
+      "loss": 0.693,
+      "step": 4001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676524273934544e-06,
+      "loss": 0.777,
+      "step": 4002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672064429223246e-06,
+      "loss": 0.2399,
+      "step": 4003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66760485336629e-06,
+      "loss": 0.7936,
+      "step": 4004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66314554726656e-06,
+      "loss": 0.7377,
+      "step": 4005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658686511826894e-06,
+      "loss": 0.6479,
+      "step": 4006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654227747950072e-06,
+      "loss": 0.2278,
+      "step": 4007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649769256538802e-06,
+      "loss": 0.7028,
+      "step": 4008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645311038495772e-06,
+      "loss": 0.6525,
+      "step": 4009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640853094723576e-06,
+      "loss": 0.7506,
+      "step": 4010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636395426124788e-06,
+      "loss": 0.2117,
+      "step": 4011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631938033601894e-06,
+      "loss": 0.6962,
+      "step": 4012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627480918057353e-06,
+      "loss": 0.7629,
+      "step": 4013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623024080393546e-06,
+      "loss": 0.7594,
+      "step": 4014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618567521512806e-06,
+      "loss": 0.7794,
+      "step": 4015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614111242317413e-06,
+      "loss": 0.7786,
+      "step": 4016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.609655243709582e-06,
+      "loss": 0.7614,
+      "step": 4017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605199526591485e-06,
+      "loss": 0.7712,
+      "step": 4018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600744091865218e-06,
+      "loss": 0.6958,
+      "step": 4019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596288940432838e-06,
+      "loss": 0.7095,
+      "step": 4020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591834073196333e-06,
+      "loss": 0.7002,
+      "step": 4021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587379491057635e-06,
+      "loss": 0.7637,
+      "step": 4022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582925194918626e-06,
+      "loss": 0.7188,
+      "step": 4023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57847118568112e-06,
+      "loss": 0.261,
+      "step": 4024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57401746424688e-06,
+      "loss": 0.7546,
+      "step": 4025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569564031517605e-06,
+      "loss": 0.7686,
+      "step": 4026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565110888394943e-06,
+      "loss": 0.7715,
+      "step": 4027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56065803578048e-06,
+      "loss": 0.7459,
+      "step": 4028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556205474575731e-06,
+      "loss": 0.6785,
+      "step": 4029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551753205682178e-06,
+      "loss": 0.7758,
+      "step": 4030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547301230001217e-06,
+      "loss": 0.6901,
+      "step": 4031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542849548434207e-06,
+      "loss": 0.758,
+      "step": 4032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53839816188243e-06,
+      "loss": 0.8024,
+      "step": 4033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533947071247116e-06,
+      "loss": 0.6836,
+      "step": 4034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529496277429439e-06,
+      "loss": 0.7419,
+      "step": 4035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525045781330503e-06,
+      "loss": 0.6577,
+      "step": 4036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52059558385136e-06,
+      "loss": 0.7077,
+      "step": 4037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516145685892997e-06,
+      "loss": 0.7692,
+      "step": 4038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511696088356345e-06,
+      "loss": 0.8325,
+      "step": 4039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507246792142273e-06,
+      "loss": 0.2346,
+      "step": 4040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502797798151577e-06,
+      "loss": 0.7634,
+      "step": 4041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498349107285014e-06,
+      "loss": 0.746,
+      "step": 4042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493900720443258e-06,
+      "loss": 0.5965,
+      "step": 4043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489452638526938e-06,
+      "loss": 0.6705,
+      "step": 4044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48500486243661e-06,
+      "loss": 0.7218,
+      "step": 4045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480557393072777e-06,
+      "loss": 0.7039,
+      "step": 4046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476110231335874e-06,
+      "loss": 0.7501,
+      "step": 4047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471663378126268e-06,
+      "loss": 0.6819,
+      "step": 4048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467216834344283e-06,
+      "loss": 0.7361,
+      "step": 4049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462770600890153e-06,
+      "loss": 0.7342,
+      "step": 4050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45832467866408e-06,
+      "loss": 0.6907,
+      "step": 4051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453879068566175e-06,
+      "loss": 0.7708,
+      "step": 4052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449433771496505e-06,
+      "loss": 0.2649,
+      "step": 4053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444988788355064e-06,
+      "loss": 0.7851,
+      "step": 4054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440544120041782e-06,
+      "loss": 0.7815,
+      "step": 4055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436099767456534e-06,
+      "loss": 0.7243,
+      "step": 4056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43165573149912e-06,
+      "loss": 0.6836,
+      "step": 4057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427212013069285e-06,
+      "loss": 0.6653,
+      "step": 4058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422768613066703e-06,
+      "loss": 0.7609,
+      "step": 4059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41832553239099e-06,
+      "loss": 0.802,
+      "step": 4060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413882771941693e-06,
+      "loss": 0.7285,
+      "step": 4061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40944033261829e-06,
+      "loss": 0.7261,
+      "step": 4062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404998215320206e-06,
+      "loss": 0.6907,
+      "step": 4063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400556420946788e-06,
+      "loss": 0.7039,
+      "step": 4064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396114950397327e-06,
+      "loss": 0.7576,
+      "step": 4065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391673804571044e-06,
+      "loss": 0.7356,
+      "step": 4066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.387232984367094e-06,
+      "loss": 0.7813,
+      "step": 4067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38279249068457e-06,
+      "loss": 0.7279,
+      "step": 4068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37835232442249e-06,
+      "loss": 0.7498,
+      "step": 4069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37391248647982e-06,
+      "loss": 0.7574,
+      "step": 4070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369472977755444e-06,
+      "loss": 0.7299,
+      "step": 4071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365033799148192e-06,
+      "loss": 0.67,
+      "step": 4072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360594951556821e-06,
+      "loss": 0.8091,
+      "step": 4073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356156435880013e-06,
+      "loss": 0.7132,
+      "step": 4074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351718253016408e-06,
+      "loss": 0.7868,
+      "step": 4075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347280403864544e-06,
+      "loss": 0.6923,
+      "step": 4076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342842889322926e-06,
+      "loss": 0.7586,
+      "step": 4077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33840571028996e-06,
+      "loss": 0.7612,
+      "step": 4078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333968867664015e-06,
+      "loss": 0.808,
+      "step": 4079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329532362343363e-06,
+      "loss": 0.7397,
+      "step": 4080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325096195226224e-06,
+      "loss": 0.7686,
+      "step": 4081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320660367210748e-06,
+      "loss": 0.7746,
+      "step": 4082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316224879195011e-06,
+      "loss": 0.6766,
+      "step": 4083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31178973207703e-06,
+      "loss": 0.6717,
+      "step": 4084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30735492675474e-06,
+      "loss": 0.7334,
+      "step": 4085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302920464126016e-06,
+      "loss": 0.2295,
+      "step": 4086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298486345088662e-06,
+      "loss": 0.7323,
+      "step": 4087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294052570540408e-06,
+      "loss": 0.6854,
+      "step": 4088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289619141378923e-06,
+      "loss": 0.7473,
+      "step": 4089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285186058501796e-06,
+      "loss": 0.7332,
+      "step": 4090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280753322806554e-06,
+      "loss": 0.7491,
+      "step": 4091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276320935190644e-06,
+      "loss": 0.6942,
+      "step": 4092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.271888896551459e-06,
+      "loss": 0.2251,
+      "step": 4093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267457207786307e-06,
+      "loss": 0.7495,
+      "step": 4094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263025869792421e-06,
+      "loss": 0.7897,
+      "step": 4095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258594883466986e-06,
+      "loss": 0.6748,
+      "step": 4096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254164249707088e-06,
+      "loss": 0.7104,
+      "step": 4097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249733969409766e-06,
+      "loss": 0.7944,
+      "step": 4098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245304043471965e-06,
+      "loss": 0.7012,
+      "step": 4099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24087447279058e-06,
+      "loss": 0.7496,
+      "step": 4100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23644525826242e-06,
+      "loss": 0.7768,
+      "step": 4101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232016400784219e-06,
+      "loss": 0.7462,
+      "step": 4102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227587901252655e-06,
+      "loss": 0.6613,
+      "step": 4103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223159760564316e-06,
+      "loss": 0.7266,
+      "step": 4104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218731979615731e-06,
+      "loss": 0.8089,
+      "step": 4105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214304559303347e-06,
+      "loss": 0.6663,
+      "step": 4106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20987750052354e-06,
+      "loss": 0.7213,
+      "step": 4107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205450804172617e-06,
+      "loss": 0.7902,
+      "step": 4108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201024471146805e-06,
+      "loss": 0.7649,
+      "step": 4109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196598502342268e-06,
+      "loss": 0.7015,
+      "step": 4110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19217289865508e-06,
+      "loss": 0.7042,
+      "step": 4111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.187747660981257e-06,
+      "loss": 0.7899,
+      "step": 4112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183322790216737e-06,
+      "loss": 0.6276,
+      "step": 4113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178898287257368e-06,
+      "loss": 0.6946,
+      "step": 4114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174474152998953e-06,
+      "loss": 0.7669,
+      "step": 4115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17005038833719e-06,
+      "loss": 0.6813,
+      "step": 4116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165626994167727e-06,
+      "loss": 0.698,
+      "step": 4117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161203971386114e-06,
+      "loss": 0.6866,
+      "step": 4118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156781320887852e-06,
+      "loss": 0.7104,
+      "step": 4119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152359043568344e-06,
+      "loss": 0.6512,
+      "step": 4120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147937140322925e-06,
+      "loss": 0.8286,
+      "step": 4121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14351561204686e-06,
+      "loss": 0.7685,
+      "step": 4122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139094459635327e-06,
+      "loss": 0.694,
+      "step": 4123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134673683983441e-06,
+      "loss": 0.2366,
+      "step": 4124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13025328598623e-06,
+      "loss": 0.7029,
+      "step": 4125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125833266538652e-06,
+      "loss": 0.729,
+      "step": 4126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121413626535584e-06,
+      "loss": 0.7567,
+      "step": 4127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116994366871827e-06,
+      "loss": 0.7465,
+      "step": 4128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112575488442109e-06,
+      "loss": 0.8085,
+      "step": 4129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108156992141075e-06,
+      "loss": 0.7138,
+      "step": 4130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103738878863296e-06,
+      "loss": 0.6852,
+      "step": 4131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099321149503265e-06,
+      "loss": 0.8002,
+      "step": 4132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.0949038049554e-06,
+      "loss": 0.6911,
+      "step": 4133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090486846114037e-06,
+      "loss": 0.7905,
+      "step": 4134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08607027387343e-06,
+      "loss": 0.7433,
+      "step": 4135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081654089127766e-06,
+      "loss": 0.7572,
+      "step": 4136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077238292771143e-06,
+      "loss": 0.6963,
+      "step": 4137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072822885697591e-06,
+      "loss": 0.7688,
+      "step": 4138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.06840786880105e-06,
+      "loss": 0.2331,
+      "step": 4139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063993242975386e-06,
+      "loss": 0.7076,
+      "step": 4140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05957900911439e-06,
+      "loss": 0.6536,
+      "step": 4141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055165168111762e-06,
+      "loss": 0.7185,
+      "step": 4142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050751720861139e-06,
+      "loss": 0.7743,
+      "step": 4143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04633866825606e-06,
+      "loss": 0.7379,
+      "step": 4144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041926011190003e-06,
+      "loss": 0.72,
+      "step": 4145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03751375055635e-06,
+      "loss": 0.6926,
+      "step": 4146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033101887248407e-06,
+      "loss": 0.7797,
+      "step": 4147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028690422159406e-06,
+      "loss": 0.747,
+      "step": 4148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024279356182492e-06,
+      "loss": 0.8099,
+      "step": 4149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019868690210733e-06,
+      "loss": 0.7947,
+      "step": 4150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015458425137112e-06,
+      "loss": 0.2314,
+      "step": 4151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011048561854535e-06,
+      "loss": 0.6839,
+      "step": 4152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006639101255823e-06,
+      "loss": 0.7609,
+      "step": 4153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002230044233715e-06,
+      "loss": 0.7732,
+      "step": 4154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997821391680875e-06,
+      "loss": 0.7238,
+      "step": 4155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993413144489876e-06,
+      "loss": 0.6926,
+      "step": 4156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98900530355322e-06,
+      "loss": 0.7506,
+      "step": 4157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98459786976331e-06,
+      "loss": 0.7027,
+      "step": 4158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980190844012488e-06,
+      "loss": 0.642,
+      "step": 4159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975784227192997e-06,
+      "loss": 0.7468,
+      "step": 4160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971378020196995e-06,
+      "loss": 0.7298,
+      "step": 4161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966972223916578e-06,
+      "loss": 0.7122,
+      "step": 4162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962566839243733e-06,
+      "loss": 0.7254,
+      "step": 4163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958161867070389e-06,
+      "loss": 0.7443,
+      "step": 4164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953757308288362e-06,
+      "loss": 0.73,
+      "step": 4165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949353163789416e-06,
+      "loss": 0.7376,
+      "step": 4166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944949434465208e-06,
+      "loss": 0.7415,
+      "step": 4167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940546121207317e-06,
+      "loss": 0.7412,
+      "step": 4168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936143224907244e-06,
+      "loss": 0.7109,
+      "step": 4169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931740746456397e-06,
+      "loss": 0.7618,
+      "step": 4170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92733868674611e-06,
+      "loss": 0.754,
+      "step": 4171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922937046667617e-06,
+      "loss": 0.6979,
+      "step": 4172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.918535827112082e-06,
+      "loss": 0.6845,
+      "step": 4173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914135028970578e-06,
+      "loss": 0.7377,
+      "step": 4174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909734653134086e-06,
+      "loss": 0.7566,
+      "step": 4175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905334700493517e-06,
+      "loss": 0.7374,
+      "step": 4176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900935171939677e-06,
+      "loss": 0.7043,
+      "step": 4177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896536068363304e-06,
+      "loss": 0.7616,
+      "step": 4178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892137390655037e-06,
+      "loss": 0.6938,
+      "step": 4179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88773913970544e-06,
+      "loss": 0.7654,
+      "step": 4180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883341316404982e-06,
+      "loss": 0.7063,
+      "step": 4181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87894392164404e-06,
+      "loss": 0.7452,
+      "step": 4182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874546956312925e-06,
+      "loss": 0.742,
+      "step": 4183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870150421301836e-06,
+      "loss": 0.7115,
+      "step": 4184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865754317500911e-06,
+      "loss": 0.7897,
+      "step": 4185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861358645800176e-06,
+      "loss": 0.2582,
+      "step": 4186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856963407089584e-06,
+      "loss": 0.7194,
+      "step": 4187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852568602258997e-06,
+      "loss": 0.6903,
+      "step": 4188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848174232198186e-06,
+      "loss": 0.7273,
+      "step": 4189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843780297796841e-06,
+      "loss": 0.2352,
+      "step": 4190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839386799944554e-06,
+      "loss": 0.2347,
+      "step": 4191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834993739530844e-06,
+      "loss": 0.7859,
+      "step": 4192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830601117445123e-06,
+      "loss": 0.7154,
+      "step": 4193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826208934576725e-06,
+      "loss": 0.6894,
+      "step": 4194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821817191814894e-06,
+      "loss": 0.7227,
+      "step": 4195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817425890048783e-06,
+      "loss": 0.7342,
+      "step": 4196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813035030167458e-06,
+      "loss": 0.6998,
+      "step": 4197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808644613059893e-06,
+      "loss": 0.2552,
+      "step": 4198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804254639614975e-06,
+      "loss": 0.7994,
+      "step": 4199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799865110721498e-06,
+      "loss": 0.7322,
+      "step": 4200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795476027268165e-06,
+      "loss": 0.7705,
+      "step": 4201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7910873901436e-06,
+      "loss": 0.7364,
+      "step": 4202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786699200236319e-06,
+      "loss": 0.2562,
+      "step": 4203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782311458434762e-06,
+      "loss": 0.7681,
+      "step": 4204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77792416562727e-06,
+      "loss": 0.7481,
+      "step": 4205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773537322702098e-06,
+      "loss": 0.713,
+      "step": 4206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769150930547411e-06,
+      "loss": 0.6969,
+      "step": 4207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76476499005127e-06,
+      "loss": 0.7149,
+      "step": 4208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760379502101665e-06,
+      "loss": 0.6756,
+      "step": 4209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755994467586473e-06,
+      "loss": 0.6633,
+      "step": 4210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.751609887393501e-06,
+      "loss": 0.6492,
+      "step": 4211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747225762410442e-06,
+      "loss": 0.8,
+      "step": 4212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742842093524919e-06,
+      "loss": 0.7524,
+      "step": 4213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73845888162444e-06,
+      "loss": 0.2269,
+      "step": 4214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734076127596439e-06,
+      "loss": 0.6126,
+      "step": 4215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729693832328246e-06,
+      "loss": 0.7048,
+      "step": 4216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725311996707105e-06,
+      "loss": 0.6705,
+      "step": 4217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720930621620163e-06,
+      "loss": 0.7323,
+      "step": 4218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716549707954475e-06,
+      "loss": 0.7113,
+      "step": 4219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712169256597002e-06,
+      "loss": 0.7195,
+      "step": 4220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707789268434613e-06,
+      "loss": 0.7257,
+      "step": 4221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70340974435408e-06,
+      "loss": 0.2338,
+      "step": 4222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699030685242086e-06,
+      "loss": 0.6432,
+      "step": 4223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694652091985213e-06,
+      "loss": 0.7589,
+      "step": 4224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690273965469957e-06,
+      "loss": 0.6906,
+      "step": 4225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685896306582716e-06,
+      "loss": 0.6579,
+      "step": 4226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681519116209784e-06,
+      "loss": 0.6547,
+      "step": 4227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67714239523738e-06,
+      "loss": 0.7286,
+      "step": 4228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672766144551604e-06,
+      "loss": 0.7009,
+      "step": 4229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668390365038486e-06,
+      "loss": 0.7321,
+      "step": 4230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664015057583938e-06,
+      "loss": 0.7211,
+      "step": 4231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659640223073797e-06,
+      "loss": 0.7645,
+      "step": 4232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655265862393783e-06,
+      "loss": 0.2197,
+      "step": 4233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650891976429536e-06,
+      "loss": 0.6679,
+      "step": 4234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646518566066594e-06,
+      "loss": 0.8403,
+      "step": 4235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642145632190396e-06,
+      "loss": 0.2373,
+      "step": 4236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637773175686295e-06,
+      "loss": 0.7885,
+      "step": 4237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633401197439533e-06,
+      "loss": 0.7596,
+      "step": 4238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6290296983352684e-06,
+      "loss": 0.7196,
+      "step": 4239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624658679258552e-06,
+      "loss": 0.786,
+      "step": 4240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620288141094343e-06,
+      "loss": 0.8312,
+      "step": 4241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615918084727505e-06,
+      "loss": 0.7432,
+      "step": 4242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6115485110427965e-06,
+      "loss": 0.6908,
+      "step": 4243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607179420924887e-06,
+      "loss": 0.6822,
+      "step": 4244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602810815258341e-06,
+      "loss": 0.7438,
+      "step": 4245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598442694927633e-06,
+      "loss": 0.7654,
+      "step": 4246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594075060817133e-06,
+      "loss": 0.7652,
+      "step": 4247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589707913811107e-06,
+      "loss": 0.2465,
+      "step": 4248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5853412547937385e-06,
+      "loss": 0.6637,
+      "step": 4249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580975084649093e-06,
+      "loss": 0.7428,
+      "step": 4250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576609404261159e-06,
+      "loss": 0.6994,
+      "step": 4251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572244214513802e-06,
+      "loss": 0.7251,
+      "step": 4252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567879516290812e-06,
+      "loss": 0.7362,
+      "step": 4253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5635153104758575e-06,
+      "loss": 0.6804,
+      "step": 4254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559151597952519e-06,
+      "loss": 0.7725,
+      "step": 4255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55478837960428e-06,
+      "loss": 0.674,
+      "step": 4256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550425656314514e-06,
+      "loss": 0.7266,
+      "step": 4257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546063428966505e-06,
+      "loss": 0.6981,
+      "step": 4258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541701698443427e-06,
+      "loss": 0.7272,
+      "step": 4259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537340465628357e-06,
+      "loss": 0.7189,
+      "step": 4260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532979731404278e-06,
+      "loss": 0.7033,
+      "step": 4261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5286194966540575e-06,
+      "loss": 0.6571,
+      "step": 4262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5242597622604775e-06,
+      "loss": 0.6754,
+      "step": 4263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5199005291062055e-06,
+      "loss": 0.7233,
+      "step": 4264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515541798073819e-06,
+      "loss": 0.7676,
+      "step": 4265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511183570045786e-06,
+      "loss": 0.7276,
+      "step": 4266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506825845904473e-06,
+      "loss": 0.6448,
+      "step": 4267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50246862653215e-06,
+      "loss": 0.7664,
+      "step": 4268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498111912810977e-06,
+      "loss": 0.6989,
+      "step": 4269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493755705623022e-06,
+      "loss": 0.7776,
+      "step": 4270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489400005850238e-06,
+      "loss": 0.6689,
+      "step": 4271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.485044814374487e-06,
+      "loss": 0.6862,
+      "step": 4272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4806901320775226e-06,
+      "loss": 0.7048,
+      "step": 4273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4763359598409856e-06,
+      "loss": 0.6235,
+      "step": 4274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4719822985464385e-06,
+      "loss": 0.7946,
+      "step": 4275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467629149075311e-06,
+      "loss": 0.6767,
+      "step": 4276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463276512308957e-06,
+      "loss": 0.6381,
+      "step": 4277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4589243891285975e-06,
+      "loss": 0.7428,
+      "step": 4278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454572780415381e-06,
+      "loss": 0.752,
+      "step": 4279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450221687050327e-06,
+      "loss": 0.7102,
+      "step": 4280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445871109914358e-06,
+      "loss": 0.6773,
+      "step": 4281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441521049888298e-06,
+      "loss": 0.6721,
+      "step": 4282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4371715078528595e-06,
+      "loss": 0.7034,
+      "step": 4283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432822484688656e-06,
+      "loss": 0.7166,
+      "step": 4284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428473981276185e-06,
+      "loss": 0.8128,
+      "step": 4285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424125998495853e-06,
+      "loss": 0.7153,
+      "step": 4286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419778537227953e-06,
+      "loss": 0.5579,
+      "step": 4287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4154315983526705e-06,
+      "loss": 0.7384,
+      "step": 4288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411085182750093e-06,
+      "loss": 0.7167,
+      "step": 4289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406739291300191e-06,
+      "loss": 0.7213,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402393924882842e-06,
+      "loss": 0.7292,
+      "step": 4291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398049084377807e-06,
+      "loss": 0.7459,
+      "step": 4292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3937047706647466e-06,
+      "loss": 0.7031,
+      "step": 4293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389360984623212e-06,
+      "loss": 0.7247,
+      "step": 4294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3850177271326395e-06,
+      "loss": 0.7195,
+      "step": 4295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3806749990723815e-06,
+      "loss": 0.713,
+      "step": 4296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.376332801321653e-06,
+      "loss": 0.7379,
+      "step": 4297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37199113475959e-06,
+      "loss": 0.7086,
+      "step": 4298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367650000265201e-06,
+      "loss": 0.7407,
+      "step": 4299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363309398717392e-06,
+      "loss": 0.751,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3589693309949695e-06,
+      "loss": 0.7785,
+      "step": 4301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354629797976618e-06,
+      "loss": 0.2471,
+      "step": 4302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3502908005409265e-06,
+      "loss": 0.6776,
+      "step": 4303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345952339566368e-06,
+      "loss": 0.7324,
+      "step": 4304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341614415931311e-06,
+      "loss": 0.7466,
+      "step": 4305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337277030514009e-06,
+      "loss": 0.7953,
+      "step": 4306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332940184192614e-06,
+      "loss": 0.7356,
+      "step": 4307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328603877845165e-06,
+      "loss": 0.6587,
+      "step": 4308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324268112349591e-06,
+      "loss": 0.695,
+      "step": 4309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319932888583714e-06,
+      "loss": 0.7296,
+      "step": 4310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3155982074252426e-06,
+      "loss": 0.7554,
+      "step": 4311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311264069751783e-06,
+      "loss": 0.7642,
+      "step": 4312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306930476440826e-06,
+      "loss": 0.7703,
+      "step": 4313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302597428369742e-06,
+      "loss": 0.7247,
+      "step": 4314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298264926415817e-06,
+      "loss": 0.7782,
+      "step": 4315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293932971456197e-06,
+      "loss": 0.6754,
+      "step": 4316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289601564367944e-06,
+      "loss": 0.7021,
+      "step": 4317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285270706027984e-06,
+      "loss": 0.6982,
+      "step": 4318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2809403973131565e-06,
+      "loss": 0.7499,
+      "step": 4319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276610639100168e-06,
+      "loss": 0.727,
+      "step": 4320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2722814322656255e-06,
+      "loss": 0.2541,
+      "step": 4321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2679527776860245e-06,
+      "loss": 0.757,
+      "step": 4322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263624676237741e-06,
+      "loss": 0.8111,
+      "step": 4323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259297128797052e-06,
+      "loss": 0.7913,
+      "step": 4324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.254970136240105e-06,
+      "loss": 0.7611,
+      "step": 4325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2506436994429515e-06,
+      "loss": 0.8089,
+      "step": 4326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246317819281522e-06,
+      "loss": 0.7635,
+      "step": 4327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2419924966316335e-06,
+      "loss": 0.7844,
+      "step": 4328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237667732368995e-06,
+      "loss": 0.7399,
+      "step": 4329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.233343527369196e-06,
+      "loss": 0.7173,
+      "step": 4330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229019882507724e-06,
+      "loss": 0.7611,
+      "step": 4331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224696798659938e-06,
+      "loss": 0.7867,
+      "step": 4332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2203742767010974e-06,
+      "loss": 0.6478,
+      "step": 4333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216052317506339e-06,
+      "loss": 0.6675,
+      "step": 4334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211730921950686e-06,
+      "loss": 0.7054,
+      "step": 4335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207410090909054e-06,
+      "loss": 0.7237,
+      "step": 4336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2030898252562355e-06,
+      "loss": 0.7861,
+      "step": 4337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198770125866918e-06,
+      "loss": 0.7107,
+      "step": 4338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194450993615669e-06,
+      "loss": 0.8102,
+      "step": 4339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1901324293769345e-06,
+      "loss": 0.6912,
+      "step": 4340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185814434025063e-06,
+      "loss": 0.2646,
+      "step": 4341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181497008434269e-06,
+      "loss": 0.7154,
+      "step": 4342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177180153478667e-06,
+      "loss": 0.7972,
+      "step": 4343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17286387003224e-06,
+      "loss": 0.8106,
+      "step": 4344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1685481589688766e-06,
+      "loss": 0.7545,
+      "step": 4345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164233021162328e-06,
+      "loss": 0.7084,
+      "step": 4346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159918457486241e-06,
+      "loss": 0.7216,
+      "step": 4347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1556044688141455e-06,
+      "loss": 0.7254,
+      "step": 4348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151291056019449e-06,
+      "loss": 0.7064,
+      "step": 4349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1469782199754535e-06,
+      "loss": 0.2367,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142665961555331e-06,
+      "loss": 0.7427,
+      "step": 4351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138354281632147e-06,
+      "loss": 0.6569,
+      "step": 4352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134043181078846e-06,
+      "loss": 0.6423,
+      "step": 4353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.12973266076825e-06,
+      "loss": 0.2134,
+      "step": 4354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1254227215730755e-06,
+      "loss": 0.6748,
+      "step": 4355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.121113364365908e-06,
+      "loss": 0.7203,
+      "step": 4356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116804590019227e-06,
+      "loss": 0.7715,
+      "step": 4357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112496399405384e-06,
+      "loss": 0.7493,
+      "step": 4358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108188793396622e-06,
+      "loss": 0.7953,
+      "step": 4359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10388177286506e-06,
+      "loss": 0.7204,
+      "step": 4360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099575338682689e-06,
+      "loss": 0.7548,
+      "step": 4361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.095269491721407e-06,
+      "loss": 0.7415,
+      "step": 4362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090964232852962e-06,
+      "loss": 0.6404,
+      "step": 4363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086659562949013e-06,
+      "loss": 0.7885,
+      "step": 4364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082355482881071e-06,
+      "loss": 0.7033,
+      "step": 4365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078051993520556e-06,
+      "loss": 0.7197,
+      "step": 4366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073749095738744e-06,
+      "loss": 0.6914,
+      "step": 4367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069446790406802e-06,
+      "loss": 0.7299,
+      "step": 4368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06514507839578e-06,
+      "loss": 0.7046,
+      "step": 4369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0608439605766e-06,
+      "loss": 0.6802,
+      "step": 4370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056543437820074e-06,
+      "loss": 0.7369,
+      "step": 4371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.052243510996884e-06,
+      "loss": 0.7446,
+      "step": 4372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0479441809775926e-06,
+      "loss": 0.2313,
+      "step": 4373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0436454486326475e-06,
+      "loss": 0.7746,
+      "step": 4374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039347314832369e-06,
+      "loss": 0.2294,
+      "step": 4375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035049780446962e-06,
+      "loss": 0.7281,
+      "step": 4376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0307528463465025e-06,
+      "loss": 0.7982,
+      "step": 4377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0264565134009554e-06,
+      "loss": 0.6968,
+      "step": 4378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022160782480154e-06,
+      "loss": 0.6819,
+      "step": 4379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017865654453813e-06,
+      "loss": 0.7021,
+      "step": 4380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013571130191526e-06,
+      "loss": 0.7097,
+      "step": 4381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009277210562766e-06,
+      "loss": 0.6336,
+      "step": 4382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004983896436881e-06,
+      "loss": 0.6712,
+      "step": 4383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000691188683095e-06,
+      "loss": 0.7725,
+      "step": 4384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996399088170516e-06,
+      "loss": 0.2324,
+      "step": 4385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992107595768119e-06,
+      "loss": 0.7395,
+      "step": 4386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987816712344762e-06,
+      "loss": 0.7062,
+      "step": 4387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983526438769182e-06,
+      "loss": 0.784,
+      "step": 4388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979236775909985e-06,
+      "loss": 0.6811,
+      "step": 4389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974947724635663e-06,
+      "loss": 0.6963,
+      "step": 4390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970659285814574e-06,
+      "loss": 0.7285,
+      "step": 4391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96637146031496e-06,
+      "loss": 0.7244,
+      "step": 4392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962084249004937e-06,
+      "loss": 0.6987,
+      "step": 4393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957797652752489e-06,
+      "loss": 0.7779,
+      "step": 4394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9535116724254905e-06,
+      "loss": 0.7109,
+      "step": 4395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949226308891677e-06,
+      "loss": 0.7267,
+      "step": 4396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9449415630186655e-06,
+      "loss": 0.7053,
+      "step": 4397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940657435673949e-06,
+      "loss": 0.6537,
+      "step": 4398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9363739277248935e-06,
+      "loss": 0.7061,
+      "step": 4399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932091040038739e-06,
+      "loss": 0.7213,
+      "step": 4400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927808773482598e-06,
+      "loss": 0.6608,
+      "step": 4401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923527128923464e-06,
+      "loss": 0.7398,
+      "step": 4402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919246107228197e-06,
+      "loss": 0.7485,
+      "step": 4403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914965709263539e-06,
+      "loss": 0.661,
+      "step": 4404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910685935896096e-06,
+      "loss": 0.7163,
+      "step": 4405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906406787992355e-06,
+      "loss": 0.7026,
+      "step": 4406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902128266418675e-06,
+      "loss": 0.7624,
+      "step": 4407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897850372041281e-06,
+      "loss": 0.713,
+      "step": 4408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893573105726286e-06,
+      "loss": 0.8004,
+      "step": 4409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889296468339658e-06,
+      "loss": 0.7305,
+      "step": 4410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885020460747257e-06,
+      "loss": 0.6404,
+      "step": 4411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.880745083814797e-06,
+      "loss": 0.7659,
+      "step": 4412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8764703384078725e-06,
+      "loss": 0.7509,
+      "step": 4413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872196225391955e-06,
+      "loss": 0.6979,
+      "step": 4414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8679227456323784e-06,
+      "loss": 0.7937,
+      "step": 4415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863649899994355e-06,
+      "loss": 0.6572,
+      "step": 4416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859377689342968e-06,
+      "loss": 0.7267,
+      "step": 4417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85510611454317e-06,
+      "loss": 0.733,
+      "step": 4418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850835176459785e-06,
+      "loss": 0.6853,
+      "step": 4419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846564875957508e-06,
+      "loss": 0.731,
+      "step": 4420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842295213900907e-06,
+      "loss": 0.7271,
+      "step": 4421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838026191154419e-06,
+      "loss": 0.6741,
+      "step": 4422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833757808582353e-06,
+      "loss": 0.8124,
+      "step": 4423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829490067048884e-06,
+      "loss": 0.6515,
+      "step": 4424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825222967418066e-06,
+      "loss": 0.2356,
+      "step": 4425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820956510553816e-06,
+      "loss": 0.7191,
+      "step": 4426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.816690697319914e-06,
+      "loss": 0.6738,
+      "step": 4427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812425528580033e-06,
+      "loss": 0.2246,
+      "step": 4428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808161005197687e-06,
+      "loss": 0.7228,
+      "step": 4429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803897128036284e-06,
+      "loss": 0.6913,
+      "step": 4430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799633897959079e-06,
+      "loss": 0.7313,
+      "step": 4431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.79537131582922e-06,
+      "loss": 0.713,
+      "step": 4432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791109382509702e-06,
+      "loss": 0.7881,
+      "step": 4433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786848098863399e-06,
+      "loss": 0.692,
+      "step": 4434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7825874657530545e-06,
+      "loss": 0.7484,
+      "step": 4435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7783274840412735e-06,
+      "loss": 0.7277,
+      "step": 4436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774068154590538e-06,
+      "loss": 0.7039,
+      "step": 4437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769809478263192e-06,
+      "loss": 0.7002,
+      "step": 4438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7655514559214506e-06,
+      "loss": 0.7183,
+      "step": 4439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761294088427391e-06,
+      "loss": 0.7257,
+      "step": 4440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757037376642963e-06,
+      "loss": 0.6623,
+      "step": 4441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752781321429983e-06,
+      "loss": 0.6949,
+      "step": 4442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748525923650131e-06,
+      "loss": 0.654,
+      "step": 4443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744271184164961e-06,
+      "loss": 0.6881,
+      "step": 4444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740017103835883e-06,
+      "loss": 0.7812,
+      "step": 4445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735763683524187e-06,
+      "loss": 0.7032,
+      "step": 4446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731510924091017e-06,
+      "loss": 0.7377,
+      "step": 4447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.727258826397388e-06,
+      "loss": 0.6609,
+      "step": 4448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723007391304185e-06,
+      "loss": 0.7006,
+      "step": 4449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718756619672151e-06,
+      "loss": 0.7444,
+      "step": 4450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714506512361902e-06,
+      "loss": 0.6566,
+      "step": 4451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710257070233919e-06,
+      "loss": 0.6396,
+      "step": 4452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706008294148534e-06,
+      "loss": 0.7299,
+      "step": 4453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701760184965971e-06,
+      "loss": 0.7303,
+      "step": 4454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697512743546291e-06,
+      "loss": 0.7463,
+      "step": 4455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693265970749443e-06,
+      "loss": 0.6993,
+      "step": 4456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68901986743522e-06,
+      "loss": 0.7319,
+      "step": 4457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684774434463299e-06,
+      "loss": 0.7052,
+      "step": 4458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680529672693208e-06,
+      "loss": 0.7341,
+      "step": 4459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676285582984341e-06,
+      "loss": 0.7433,
+      "step": 4460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6720421661959624e-06,
+      "loss": 0.7543,
+      "step": 4461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667799423187191e-06,
+      "loss": 0.6733,
+      "step": 4462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6635573548170185e-06,
+      "loss": 0.6651,
+      "step": 4463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659315961944292e-06,
+      "loss": 0.7288,
+      "step": 4464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655075245427731e-06,
+      "loss": 0.7196,
+      "step": 4465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65083520612591e-06,
+      "loss": 0.7509,
+      "step": 4466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6465958448972635e-06,
+      "loss": 0.681,
+      "step": 4467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.642357162600101e-06,
+      "loss": 0.7521,
+      "step": 4468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638119160092587e-06,
+      "loss": 0.6933,
+      "step": 4469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633881838232745e-06,
+      "loss": 0.7249,
+      "step": 4470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629645197878469e-06,
+      "loss": 0.7672,
+      "step": 4471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6254092398875096e-06,
+      "loss": 0.7457,
+      "step": 4472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621173965117482e-06,
+      "loss": 0.7192,
+      "step": 4473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616939374425855e-06,
+      "loss": 0.6908,
+      "step": 4474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6127054686699744e-06,
+      "loss": 0.7392,
+      "step": 4475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60847224870703e-06,
+      "loss": 0.7135,
+      "step": 4476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604239715394089e-06,
+      "loss": 0.6518,
+      "step": 4477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600007869588063e-06,
+      "loss": 0.7111,
+      "step": 4478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595776712145744e-06,
+      "loss": 0.7802,
+      "step": 4479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591546243923765e-06,
+      "loss": 0.7567,
+      "step": 4480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58731646577863e-06,
+      "loss": 0.2618,
+      "step": 4481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583087378566705e-06,
+      "loss": 0.7116,
+      "step": 4482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578858983144207e-06,
+      "loss": 0.7577,
+      "step": 4483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574631280367225e-06,
+      "loss": 0.2422,
+      "step": 4484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570404271091697e-06,
+      "loss": 0.762,
+      "step": 4485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566177956173423e-06,
+      "loss": 0.2437,
+      "step": 4486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56195233646807e-06,
+      "loss": 0.6653,
+      "step": 4487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557727412831154e-06,
+      "loss": 0.734,
+      "step": 4488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553503186118057e-06,
+      "loss": 0.7396,
+      "step": 4489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549279657184017e-06,
+      "loss": 0.7425,
+      "step": 4490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545056826884133e-06,
+      "loss": 0.7122,
+      "step": 4491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54083469607336e-06,
+      "loss": 0.6941,
+      "step": 4492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5366132656065065e-06,
+      "loss": 0.6766,
+      "step": 4493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5323925363382545e-06,
+      "loss": 0.712,
+      "step": 4494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528172509123123e-06,
+      "loss": 0.7531,
+      "step": 4495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523953184815515e-06,
+      "loss": 0.7256,
+      "step": 4496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519734564269661e-06,
+      "loss": 0.7416,
+      "step": 4497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515516648339678e-06,
+      "loss": 0.2568,
+      "step": 4498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.511299437879519e-06,
+      "loss": 0.7108,
+      "step": 4499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507082933743003e-06,
+      "loss": 0.7429,
+      "step": 4500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5028671367838056e-06,
+      "loss": 0.6723,
+      "step": 4501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49865204785546e-06,
+      "loss": 0.686,
+      "step": 4502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494437667811353e-06,
+      "loss": 0.742,
+      "step": 4503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49022399750473e-06,
+      "loss": 0.732,
+      "step": 4504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486011037788694e-06,
+      "loss": 0.7784,
+      "step": 4505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.481798789516201e-06,
+      "loss": 0.7702,
+      "step": 4506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477587253540063e-06,
+      "loss": 0.6972,
+      "step": 4507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473376430712951e-06,
+      "loss": 0.6759,
+      "step": 4508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469166321887388e-06,
+      "loss": 0.7518,
+      "step": 4509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464956927915758e-06,
+      "loss": 0.6929,
+      "step": 4510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4607482496502924e-06,
+      "loss": 0.7259,
+      "step": 4511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456540287943085e-06,
+      "loss": 0.6533,
+      "step": 4512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452333043646079e-06,
+      "loss": 0.77,
+      "step": 4513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4481265176110755e-06,
+      "loss": 0.7491,
+      "step": 4514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443920710689731e-06,
+      "loss": 0.7996,
+      "step": 4515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439715623733549e-06,
+      "loss": 0.7558,
+      "step": 4516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435511257593901e-06,
+      "loss": 0.7443,
+      "step": 4517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4313076131219974e-06,
+      "loss": 0.6946,
+      "step": 4518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427104691168917e-06,
+      "loss": 0.7003,
+      "step": 4519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422902492585581e-06,
+      "loss": 0.7689,
+      "step": 4520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418701018222764e-06,
+      "loss": 0.7588,
+      "step": 4521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414500268931107e-06,
+      "loss": 0.7318,
+      "step": 4522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410300245561084e-06,
+      "loss": 0.6679,
+      "step": 4523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406100948963047e-06,
+      "loss": 0.6743,
+      "step": 4524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401902379987177e-06,
+      "loss": 0.7622,
+      "step": 4525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397704539483519e-06,
+      "loss": 0.6872,
+      "step": 4526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3935074283019724e-06,
+      "loss": 0.7428,
+      "step": 4527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389311047292285e-06,
+      "loss": 0.7001,
+      "step": 4528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385115397304058e-06,
+      "loss": 0.6932,
+      "step": 4529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380920479186743e-06,
+      "loss": 0.7142,
+      "step": 4530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3767262937896455e-06,
+      "loss": 0.6727,
+      "step": 4531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372532841961924e-06,
+      "loss": 0.7775,
+      "step": 4532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368340124552581e-06,
+      "loss": 0.7259,
+      "step": 4533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364148142410481e-06,
+      "loss": 0.7377,
+      "step": 4534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359956896384331e-06,
+      "loss": 0.7465,
+      "step": 4535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355766387322694e-06,
+      "loss": 0.661,
+      "step": 4536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35157661607398e-06,
+      "loss": 0.7753,
+      "step": 4537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347387583486455e-06,
+      "loss": 0.6628,
+      "step": 4538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343199290408233e-06,
+      "loss": 0.6871,
+      "step": 4539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3390117376872685e-06,
+      "loss": 0.653,
+      "step": 4540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334824926171388e-06,
+      "loss": 0.7194,
+      "step": 4541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330638856708241e-06,
+      "loss": 0.7746,
+      "step": 4542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326453530145355e-06,
+      "loss": 0.6767,
+      "step": 4543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322268947330079e-06,
+      "loss": 0.7739,
+      "step": 4544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318085109109637e-06,
+      "loss": 0.7684,
+      "step": 4545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313902016331086e-06,
+      "loss": 0.736,
+      "step": 4546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309719669841332e-06,
+      "loss": 0.7284,
+      "step": 4547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305538070487141e-06,
+      "loss": 0.8046,
+      "step": 4548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301357219115116e-06,
+      "loss": 0.7441,
+      "step": 4549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29717711657172e-06,
+      "loss": 0.6939,
+      "step": 4550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292997763703251e-06,
+      "loss": 0.7674,
+      "step": 4551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288819161355869e-06,
+      "loss": 0.7453,
+      "step": 4552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284641310375572e-06,
+      "loss": 0.6974,
+      "step": 4553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280464211608208e-06,
+      "loss": 0.8131,
+      "step": 4554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276287865899478e-06,
+      "loss": 0.7342,
+      "step": 4555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272112274094922e-06,
+      "loss": 0.7057,
+      "step": 4556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267937437039935e-06,
+      "loss": 0.7294,
+      "step": 4557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263763355579753e-06,
+      "loss": 0.7114,
+      "step": 4558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2595900305594684e-06,
+      "loss": 0.6911,
+      "step": 4559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255417462824009e-06,
+      "loss": 0.6867,
+      "step": 4560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251245653218152e-06,
+      "loss": 0.7246,
+      "step": 4561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247074602586534e-06,
+      "loss": 0.2361,
+      "step": 4562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242904311773612e-06,
+      "loss": 0.6835,
+      "step": 4563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238734781623721e-06,
+      "loss": 0.7301,
+      "step": 4564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234566012981016e-06,
+      "loss": 0.7392,
+      "step": 4565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230398006689506e-06,
+      "loss": 0.7367,
+      "step": 4566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226230763593053e-06,
+      "loss": 0.7981,
+      "step": 4567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222064284535355e-06,
+      "loss": 0.6489,
+      "step": 4568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217898570359962e-06,
+      "loss": 0.7285,
+      "step": 4569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2137336219102625e-06,
+      "loss": 0.6804,
+      "step": 4570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.209569440029497e-06,
+      "loss": 0.6685,
+      "step": 4571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205406025560745e-06,
+      "loss": 0.7313,
+      "step": 4572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201243379346932e-06,
+      "loss": 0.7618,
+      "step": 4573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197081502230833e-06,
+      "loss": 0.8123,
+      "step": 4574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192920395055058e-06,
+      "loss": 0.7589,
+      "step": 4575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188760058662071e-06,
+      "loss": 0.6807,
+      "step": 4576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184600493894173e-06,
+      "loss": 0.6878,
+      "step": 4577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1804417015935114e-06,
+      "loss": 0.7019,
+      "step": 4578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176283682602078e-06,
+      "loss": 0.7937,
+      "step": 4579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172126437761703e-06,
+      "loss": 0.6859,
+      "step": 4580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167969967914069e-06,
+      "loss": 0.6828,
+      "step": 4581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163814273900692e-06,
+      "loss": 0.7947,
+      "step": 4582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159659356562941e-06,
+      "loss": 0.6716,
+      "step": 4583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1555052167420145e-06,
+      "loss": 0.7104,
+      "step": 4584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151351855278968e-06,
+      "loss": 0.7367,
+      "step": 4585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147199273014692e-06,
+      "loss": 0.768,
+      "step": 4586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143047470789913e-06,
+      "loss": 0.687,
+      "step": 4587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1388964494452165e-06,
+      "loss": 0.74,
+      "step": 4588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134746209821009e-06,
+      "loss": 0.7463,
+      "step": 4589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13059675275756e-06,
+      "loss": 0.7224,
+      "step": 4590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12644807909496e-06,
+      "loss": 0.6306,
+      "step": 4591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122300189673164e-06,
+      "loss": 0.7108,
+      "step": 4592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118153085331944e-06,
+      "loss": 0.7471,
+      "step": 4593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114006766910927e-06,
+      "loss": 0.7411,
+      "step": 4594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1098612352495815e-06,
+      "loss": 0.6269,
+      "step": 4595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105716491187208e-06,
+      "loss": 0.676,
+      "step": 4596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101572535562959e-06,
+      "loss": 0.7643,
+      "step": 4597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097429369215816e-06,
+      "loss": 0.7169,
+      "step": 4598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093286992984611e-06,
+      "loss": 0.6794,
+      "step": 4599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0891454077080085e-06,
+      "loss": 0.713,
+      "step": 4600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085004614224513e-06,
+      "loss": 0.7419,
+      "step": 4601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080864613372477e-06,
+      "loss": 0.8231,
+      "step": 4602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07672540599008e-06,
+      "loss": 0.7286,
+      "step": 4603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072586992915353e-06,
+      "loss": 0.7159,
+      "step": 4604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068449374986161e-06,
+      "loss": 0.7584,
+      "step": 4605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064312553040199e-06,
+      "loss": 0.7948,
+      "step": 4606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060176527915023e-06,
+      "loss": 0.6891,
+      "step": 4607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056041300447999e-06,
+      "loss": 0.7594,
+      "step": 4608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051906871476361e-06,
+      "loss": 0.7417,
+      "step": 4609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047773241837156e-06,
+      "loss": 0.649,
+      "step": 4610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04364041236729e-06,
+      "loss": 0.807,
+      "step": 4611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039508383903489e-06,
+      "loss": 0.7152,
+      "step": 4612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0353771572823295e-06,
+      "loss": 0.7478,
+      "step": 4613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031246733340219e-06,
+      "loss": 0.8186,
+      "step": 4614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027117112913405e-06,
+      "loss": 0.7084,
+      "step": 4615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022988296837975e-06,
+      "loss": 0.7883,
+      "step": 4616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018860285949846e-06,
+      "loss": 0.2369,
+      "step": 4617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0147330810847815e-06,
+      "loss": 0.7172,
+      "step": 4618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010606683078374e-06,
+      "loss": 0.6841,
+      "step": 4619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0064810927660545e-06,
+      "loss": 0.6951,
+      "step": 4620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002356310983095e-06,
+      "loss": 0.6801,
+      "step": 4621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9982323385645955e-06,
+      "loss": 0.8269,
+      "step": 4622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994109176345504e-06,
+      "loss": 0.7162,
+      "step": 4623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98998682516059e-06,
+      "loss": 0.6758,
+      "step": 4624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985865285844473e-06,
+      "loss": 0.7208,
+      "step": 4625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9817445592316e-06,
+      "loss": 0.6852,
+      "step": 4626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977624646156248e-06,
+      "loss": 0.6965,
+      "step": 4627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973505547452546e-06,
+      "loss": 0.7359,
+      "step": 4628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9693872639544385e-06,
+      "loss": 0.8075,
+      "step": 4629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965269796495725e-06,
+      "loss": 0.7375,
+      "step": 4630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9611531459100195e-06,
+      "loss": 0.6549,
+      "step": 4631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957037313030791e-06,
+      "loss": 0.7613,
+      "step": 4632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952922298691326e-06,
+      "loss": 0.8039,
+      "step": 4633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94880810372475e-06,
+      "loss": 0.709,
+      "step": 4634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94469472896403e-06,
+      "loss": 0.7158,
+      "step": 4635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940582175241958e-06,
+      "loss": 0.7266,
+      "step": 4636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936470443391164e-06,
+      "loss": 0.7359,
+      "step": 4637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932359534244112e-06,
+      "loss": 0.7159,
+      "step": 4638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928249448633095e-06,
+      "loss": 0.721,
+      "step": 4639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9241401873902475e-06,
+      "loss": 0.6498,
+      "step": 4640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9200317513475255e-06,
+      "loss": 0.7042,
+      "step": 4641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915924141336733e-06,
+      "loss": 0.7915,
+      "step": 4642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.911817358189489e-06,
+      "loss": 0.7463,
+      "step": 4643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9077114027372616e-06,
+      "loss": 0.7082,
+      "step": 4644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903606275811342e-06,
+      "loss": 0.6977,
+      "step": 4645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899501978242853e-06,
+      "loss": 0.7018,
+      "step": 4646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8953985108627575e-06,
+      "loss": 0.7355,
+      "step": 4647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891295874501839e-06,
+      "loss": 0.642,
+      "step": 4648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8871940699907225e-06,
+      "loss": 0.7697,
+      "step": 4649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88309309815986e-06,
+      "loss": 0.2168,
+      "step": 4650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878992959839538e-06,
+      "loss": 0.7205,
+      "step": 4651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874893655859872e-06,
+      "loss": 0.713,
+      "step": 4652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870795187050799e-06,
+      "loss": 0.699,
+      "step": 4653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866697554242114e-06,
+      "loss": 0.6931,
+      "step": 4654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862600758263406e-06,
+      "loss": 0.7025,
+      "step": 4655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858504799944132e-06,
+      "loss": 0.6725,
+      "step": 4656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854409680113546e-06,
+      "loss": 0.7229,
+      "step": 4657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85031539960076e-06,
+      "loss": 0.7626,
+      "step": 4658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846221959234696e-06,
+      "loss": 0.7072,
+      "step": 4659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842129359844113e-06,
+      "loss": 0.7447,
+      "step": 4660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838037602257604e-06,
+      "loss": 0.6837,
+      "step": 4661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833946687303584e-06,
+      "loss": 0.7294,
+      "step": 4662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829856615810312e-06,
+      "loss": 0.7508,
+      "step": 4663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825767388605847e-06,
+      "loss": 0.6587,
+      "step": 4664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821679006518113e-06,
+      "loss": 0.731,
+      "step": 4665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817591470374835e-06,
+      "loss": 0.2647,
+      "step": 4666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81350478100358e-06,
+      "loss": 0.7974,
+      "step": 4667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8094189392317414e-06,
+      "loss": 0.7545,
+      "step": 4668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805333945886535e-06,
+      "loss": 0.7043,
+      "step": 4669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801249801795021e-06,
+      "loss": 0.754,
+      "step": 4670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797166507784069e-06,
+      "loss": 0.6941,
+      "step": 4671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7930840646803854e-06,
+      "loss": 0.6886,
+      "step": 4672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789002473310504e-06,
+      "loss": 0.7072,
+      "step": 4673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784921734500782e-06,
+      "loss": 0.6156,
+      "step": 4674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780841849077412e-06,
+      "loss": 0.7377,
+      "step": 4675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776762817866408e-06,
+      "loss": 0.7567,
+      "step": 4676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772684641693612e-06,
+      "loss": 0.7409,
+      "step": 4677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7686073213846935e-06,
+      "loss": 0.7462,
+      "step": 4678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764530857765143e-06,
+      "loss": 0.7163,
+      "step": 4679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760455251660291e-06,
+      "loss": 0.7003,
+      "step": 4680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756380503895282e-06,
+      "loss": 0.7737,
+      "step": 4681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7523066152950935e-06,
+      "loss": 0.6937,
+      "step": 4682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7482335866845195e-06,
+      "loss": 0.7296,
+      "step": 4683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744161418888196e-06,
+      "loss": 0.7516,
+      "step": 4684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740090112730576e-06,
+      "loss": 0.7153,
+      "step": 4685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736019669035925e-06,
+      "loss": 0.6768,
+      "step": 4686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731950088628359e-06,
+      "loss": 0.237,
+      "step": 4687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7278813723318e-06,
+      "loss": 0.7819,
+      "step": 4688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723813520970012e-06,
+      "loss": 0.2542,
+      "step": 4689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719746535366558e-06,
+      "loss": 0.6878,
+      "step": 4690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7156804163448545e-06,
+      "loss": 0.751,
+      "step": 4691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711615164728124e-06,
+      "loss": 0.6384,
+      "step": 4692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707550781339422e-06,
+      "loss": 0.7184,
+      "step": 4693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70348726700162e-06,
+      "loss": 0.7072,
+      "step": 4694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6994246225374184e-06,
+      "loss": 0.8139,
+      "step": 4695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6953628487693495e-06,
+      "loss": 0.7217,
+      "step": 4696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6913019465197565e-06,
+      "loss": 0.7492,
+      "step": 4697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687241916610812e-06,
+      "loss": 0.7853,
+      "step": 4698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68318275986451e-06,
+      "loss": 0.6972,
+      "step": 4699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6791244771026665e-06,
+      "loss": 0.7511,
+      "step": 4700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67506706914693e-06,
+      "loss": 0.666,
+      "step": 4701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671010536818762e-06,
+      "loss": 0.7415,
+      "step": 4702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666954880939447e-06,
+      "loss": 0.6803,
+      "step": 4703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662900102330093e-06,
+      "loss": 0.7934,
+      "step": 4704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658846201811641e-06,
+      "loss": 0.7427,
+      "step": 4705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65479318020484e-06,
+      "loss": 0.693,
+      "step": 4706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650741038330261e-06,
+      "loss": 0.6936,
+      "step": 4707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6466897770083096e-06,
+      "loss": 0.7017,
+      "step": 4708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6426393970592e-06,
+      "loss": 0.6535,
+      "step": 4709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638589899302984e-06,
+      "loss": 0.7568,
+      "step": 4710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6345412845595096e-06,
+      "loss": 0.7061,
+      "step": 4711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630493553648473e-06,
+      "loss": 0.7025,
+      "step": 4712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6264467073893744e-06,
+      "loss": 0.7293,
+      "step": 4713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6224007466015405e-06,
+      "loss": 0.7638,
+      "step": 4714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6183556721041186e-06,
+      "loss": 0.6787,
+      "step": 4715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614311484716071e-06,
+      "loss": 0.7163,
+      "step": 4716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610268185256193e-06,
+      "loss": 0.7182,
+      "step": 4717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606225774543091e-06,
+      "loss": 0.6775,
+      "step": 4718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602184253395189e-06,
+      "loss": 0.7406,
+      "step": 4719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598143622630739e-06,
+      "loss": 0.6942,
+      "step": 4720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594103883067801e-06,
+      "loss": 0.7428,
+      "step": 4721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590065035524272e-06,
+      "loss": 0.6964,
+      "step": 4722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.586027080817857e-06,
+      "loss": 0.7514,
+      "step": 4723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581990019766079e-06,
+      "loss": 0.6949,
+      "step": 4724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577953853186282e-06,
+      "loss": 0.7325,
+      "step": 4725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573918581895628e-06,
+      "loss": 0.7223,
+      "step": 4726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569884206711112e-06,
+      "loss": 0.6515,
+      "step": 4727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565850728449516e-06,
+      "loss": 0.6943,
+      "step": 4728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5618181479274745e-06,
+      "loss": 0.699,
+      "step": 4729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557786465961417e-06,
+      "loss": 0.7674,
+      "step": 4730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55375568336761e-06,
+      "loss": 0.6792,
+      "step": 4731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549725800962116e-06,
+      "loss": 0.6517,
+      "step": 4732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545696819560826e-06,
+      "loss": 0.6568,
+      "step": 4733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541668739979458e-06,
+      "loss": 0.6849,
+      "step": 4734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5376415630335334e-06,
+      "loss": 0.7291,
+      "step": 4735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533615289538397e-06,
+      "loss": 0.7316,
+      "step": 4736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529589920309206e-06,
+      "loss": 0.7189,
+      "step": 4737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525565456160945e-06,
+      "loss": 0.6926,
+      "step": 4738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521541897908405e-06,
+      "loss": 0.7232,
+      "step": 4739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517519246366198e-06,
+      "loss": 0.5936,
+      "step": 4740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51349750234875e-06,
+      "loss": 0.7406,
+      "step": 4741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509476666670301e-06,
+      "loss": 0.7139,
+      "step": 4742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505456740144921e-06,
+      "loss": 0.7166,
+      "step": 4743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50143772358648e-06,
+      "loss": 0.7191,
+      "step": 4744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4974196178086695e-06,
+      "loss": 0.7106,
+      "step": 4745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4934024236249974e-06,
+      "loss": 0.7395,
+      "step": 4746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489386141848783e-06,
+      "loss": 0.6851,
+      "step": 4747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485370773293171e-06,
+      "loss": 0.7496,
+      "step": 4748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481356318771109e-06,
+      "loss": 0.6303,
+      "step": 4749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4773427790953695e-06,
+      "loss": 0.7704,
+      "step": 4750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473330155078531e-06,
+      "loss": 0.2393,
+      "step": 4751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469318447532989e-06,
+      "loss": 0.7627,
+      "step": 4752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465307657270966e-06,
+      "loss": 0.7686,
+      "step": 4753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461297785104474e-06,
+      "loss": 0.6969,
+      "step": 4754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457288831845364e-06,
+      "loss": 0.749,
+      "step": 4755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4532807983052825e-06,
+      "loss": 0.7509,
+      "step": 4756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449273685295711e-06,
+      "loss": 0.6291,
+      "step": 4757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445267493627915e-06,
+      "loss": 0.7127,
+      "step": 4758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4412622241129934e-06,
+      "loss": 0.8191,
+      "step": 4759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4372578775618615e-06,
+      "loss": 0.671,
+      "step": 4760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433254454785235e-06,
+      "loss": 0.7212,
+      "step": 4761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429251956593651e-06,
+      "loss": 0.7225,
+      "step": 4762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425250383797452e-06,
+      "loss": 0.7456,
+      "step": 4763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421249737206807e-06,
+      "loss": 0.6956,
+      "step": 4764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417250017631681e-06,
+      "loss": 0.7028,
+      "step": 4765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413251225881863e-06,
+      "loss": 0.7334,
+      "step": 4766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409253362766946e-06,
+      "loss": 0.225,
+      "step": 4767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40525642909634e-06,
+      "loss": 0.6743,
+      "step": 4768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401260425679266e-06,
+      "loss": 0.6844,
+      "step": 4769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397265353324761e-06,
+      "loss": 0.6829,
+      "step": 4770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393271212841662e-06,
+      "loss": 0.2425,
+      "step": 4771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389278005038627e-06,
+      "loss": 0.6652,
+      "step": 4772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3852857307241205e-06,
+      "loss": 0.7752,
+      "step": 4773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381294390706429e-06,
+      "loss": 0.7206,
+      "step": 4774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377303985793624e-06,
+      "loss": 0.2196,
+      "step": 4775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373314516793619e-06,
+      "loss": 0.7308,
+      "step": 4776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369325984514113e-06,
+      "loss": 0.7425,
+      "step": 4777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36533838976264e-06,
+      "loss": 0.6978,
+      "step": 4778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361351733346516e-06,
+      "loss": 0.742,
+      "step": 4779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357366016072884e-06,
+      "loss": 0.6586,
+      "step": 4780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3533812387486985e-06,
+      "loss": 0.6608,
+      "step": 4781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349397402180716e-06,
+      "loss": 0.6932,
+      "step": 4782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345414507175506e-06,
+      "loss": 0.7988,
+      "step": 4783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341432554539442e-06,
+      "loss": 0.7103,
+      "step": 4784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3374515450787216e-06,
+      "loss": 0.2215,
+      "step": 4785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333471479599336e-06,
+      "loss": 0.7668,
+      "step": 4786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329492358907091e-06,
+      "loss": 0.7733,
+      "step": 4787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3255141838076006e-06,
+      "loss": 0.7029,
+      "step": 4788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321536955106285e-06,
+      "loss": 0.7093,
+      "step": 4789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317560673608382e-06,
+      "loss": 0.7367,
+      "step": 4790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313585340118931e-06,
+      "loss": 0.7452,
+      "step": 4791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309610955442768e-06,
+      "loss": 0.699,
+      "step": 4792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305637520384561e-06,
+      "loss": 0.7293,
+      "step": 4793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301665035748765e-06,
+      "loss": 0.6954,
+      "step": 4794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297693502339662e-06,
+      "loss": 0.6477,
+      "step": 4795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2937229209613154e-06,
+      "loss": 0.7154,
+      "step": 4796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289753292417621e-06,
+      "loss": 0.7823,
+      "step": 4797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2857846175122684e-06,
+      "loss": 0.6853,
+      "step": 4798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281816897048756e-06,
+      "loss": 0.6728,
+      "step": 4799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277850131830391e-06,
+      "loss": 0.7836,
+      "step": 4800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2738843226602835e-06,
+      "loss": 0.7302,
+      "step": 4801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269919470341358e-06,
+      "loss": 0.6313,
+      "step": 4802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265955575676336e-06,
+      "loss": 0.7408,
+      "step": 4803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261992639467753e-06,
+      "loss": 0.6969,
+      "step": 4804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258030662517941e-06,
+      "loss": 0.2206,
+      "step": 4805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254069645629043e-06,
+      "loss": 0.2627,
+      "step": 4806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250109589603016e-06,
+      "loss": 0.6288,
+      "step": 4807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24615049524161e-06,
+      "loss": 0.7892,
+      "step": 4808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242192363346383e-06,
+      "loss": 0.7097,
+      "step": 4809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2382351947186995e-06,
+      "loss": 0.6684,
+      "step": 4810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234278990159734e-06,
+      "loss": 0.7562,
+      "step": 4811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230323750470459e-06,
+      "loss": 0.7178,
+      "step": 4812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226369476451652e-06,
+      "loss": 0.7141,
+      "step": 4813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222416168903898e-06,
+      "loss": 0.7228,
+      "step": 4814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218463828627581e-06,
+      "loss": 0.749,
+      "step": 4815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2145124564229025e-06,
+      "loss": 0.6883,
+      "step": 4816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210562053089854e-06,
+      "loss": 0.6516,
+      "step": 4817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206612619428235e-06,
+      "loss": 0.6731,
+      "step": 4818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202664156237649e-06,
+      "loss": 0.7354,
+      "step": 4819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198716664317501e-06,
+      "loss": 0.2032,
+      "step": 4820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194770144467013e-06,
+      "loss": 0.722,
+      "step": 4821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190824597485183e-06,
+      "loss": 0.6581,
+      "step": 4822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1868800241708396e-06,
+      "loss": 0.6996,
+      "step": 4823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1829364253225945e-06,
+      "loss": 0.7808,
+      "step": 4824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178993801738883e-06,
+      "loss": 0.2654,
+      "step": 4825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175052154217917e-06,
+      "loss": 0.7242,
+      "step": 4826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171111483557724e-06,
+      "loss": 0.6849,
+      "step": 4827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167171790556144e-06,
+      "loss": 0.7597,
+      "step": 4828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1632330760108015e-06,
+      "loss": 0.7602,
+      "step": 4829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159295340719133e-06,
+      "loss": 0.6886,
+      "step": 4830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155358585478372e-06,
+      "loss": 0.691,
+      "step": 4831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151422811085551e-06,
+      "loss": 0.7638,
+      "step": 4832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147488018337519e-06,
+      "loss": 0.6583,
+      "step": 4833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14355420803091e-06,
+      "loss": 0.6915,
+      "step": 4834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139621380962164e-06,
+      "loss": 0.7571,
+      "step": 4835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135689537927519e-06,
+      "loss": 0.7266,
+      "step": 4836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131758679723027e-06,
+      "loss": 0.7701,
+      "step": 4837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12782880714453e-06,
+      "loss": 0.7304,
+      "step": 4838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12389992098766e-06,
+      "loss": 0.6775,
+      "step": 4839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11997202204787e-06,
+      "loss": 0.7451,
+      "step": 4840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1160451111204e-06,
+      "loss": 0.7274,
+      "step": 4841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112119189000304e-06,
+      "loss": 0.7512,
+      "step": 4842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108194256482408e-06,
+      "loss": 0.6559,
+      "step": 4843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104270314361371e-06,
+      "loss": 0.7188,
+      "step": 4844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1003473634316294e-06,
+      "loss": 0.6989,
+      "step": 4845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096425404487425e-06,
+      "loss": 0.644,
+      "step": 4846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092504438322801e-06,
+      "loss": 0.7659,
+      "step": 4847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088584465731592e-06,
+      "loss": 0.7521,
+      "step": 4848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084665487507445e-06,
+      "loss": 0.692,
+      "step": 4849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080747504443795e-06,
+      "loss": 0.7665,
+      "step": 4850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076830517333879e-06,
+      "loss": 0.703,
+      "step": 4851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072914526970729e-06,
+      "loss": 0.6338,
+      "step": 4852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0689995341471765e-06,
+      "loss": 0.7346,
+      "step": 4853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.06508553965586e-06,
+      "loss": 0.693,
+      "step": 4854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061172544289204e-06,
+      "loss": 0.7392,
+      "step": 4855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057260548839434e-06,
+      "loss": 0.6604,
+      "step": 4856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05334955409857e-06,
+      "loss": 0.7229,
+      "step": 4857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049439560858442e-06,
+      "loss": 0.6971,
+      "step": 4858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.04553056991067e-06,
+      "loss": 0.741,
+      "step": 4859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041622582046658e-06,
+      "loss": 0.6716,
+      "step": 4860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037715598057629e-06,
+      "loss": 0.636,
+      "step": 4861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033809618734584e-06,
+      "loss": 0.74,
+      "step": 4862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0299046448683445e-06,
+      "loss": 0.7266,
+      "step": 4863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0260006772494915e-06,
+      "loss": 0.7214,
+      "step": 4864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0220977166684415e-06,
+      "loss": 0.7333,
+      "step": 4865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018195763915381e-06,
+      "loss": 0.6635,
+      "step": 4866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014294819780303e-06,
+      "loss": 0.7527,
+      "step": 4867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010394885052994e-06,
+      "loss": 0.6437,
+      "step": 4868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006495960523033e-06,
+      "loss": 0.721,
+      "step": 4869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002598046979803e-06,
+      "loss": 0.6836,
+      "step": 4870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998701145212476e-06,
+      "loss": 0.7265,
+      "step": 4871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994805256010019e-06,
+      "loss": 0.8219,
+      "step": 4872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9909103801611945e-06,
+      "loss": 0.66,
+      "step": 4873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987016518454558e-06,
+      "loss": 0.7402,
+      "step": 4874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983123671678469e-06,
+      "loss": 0.6905,
+      "step": 4875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979231840621072e-06,
+      "loss": 0.7192,
+      "step": 4876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975341026070307e-06,
+      "loss": 0.6781,
+      "step": 4877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971451228813911e-06,
+      "loss": 0.7665,
+      "step": 4878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967562449639409e-06,
+      "loss": 0.6732,
+      "step": 4879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963674689334134e-06,
+      "loss": 0.693,
+      "step": 4880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959787948685198e-06,
+      "loss": 0.6973,
+      "step": 4881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955902228479513e-06,
+      "loss": 0.7025,
+      "step": 4882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95201752950378e-06,
+      "loss": 0.7817,
+      "step": 4883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948133852544502e-06,
+      "loss": 0.6717,
+      "step": 4884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944251198387971e-06,
+      "loss": 0.6693,
+      "step": 4885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940369567820261e-06,
+      "loss": 0.6754,
+      "step": 4886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.936488961627257e-06,
+      "loss": 0.7212,
+      "step": 4887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932609380594622e-06,
+      "loss": 0.647,
+      "step": 4888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928730825507828e-06,
+      "loss": 0.7502,
+      "step": 4889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924853297152115e-06,
+      "loss": 0.7111,
+      "step": 4890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.920976796312539e-06,
+      "loss": 0.6843,
+      "step": 4891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917101323773933e-06,
+      "loss": 0.6975,
+      "step": 4892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91322688032093e-06,
+      "loss": 0.2482,
+      "step": 4893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909353466737947e-06,
+      "loss": 0.7316,
+      "step": 4894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905481083809197e-06,
+      "loss": 0.6919,
+      "step": 4895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901609732318689e-06,
+      "loss": 0.7196,
+      "step": 4896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897739413050216e-06,
+      "loss": 0.6935,
+      "step": 4897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893870126787363e-06,
+      "loss": 0.7155,
+      "step": 4898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890001874313509e-06,
+      "loss": 0.6251,
+      "step": 4899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8861346564118164e-06,
+      "loss": 0.6785,
+      "step": 4900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882268473865252e-06,
+      "loss": 0.7103,
+      "step": 4901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.87840332745656e-06,
+      "loss": 0.7286,
+      "step": 4902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8745392179682806e-06,
+      "loss": 0.7731,
+      "step": 4903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870676146182743e-06,
+      "loss": 0.7769,
+      "step": 4904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86681411288206e-06,
+      "loss": 0.665,
+      "step": 4905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862953118848155e-06,
+      "loss": 0.7335,
+      "step": 4906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859093164862708e-06,
+      "loss": 0.6968,
+      "step": 4907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855234251707221e-06,
+      "loss": 0.2427,
+      "step": 4908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8513763801629625e-06,
+      "loss": 0.7607,
+      "step": 4909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847519551011007e-06,
+      "loss": 0.731,
+      "step": 4910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843663765032204e-06,
+      "loss": 0.7426,
+      "step": 4911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839809023007192e-06,
+      "loss": 0.6298,
+      "step": 4912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835955325716415e-06,
+      "loss": 0.696,
+      "step": 4913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832102673940088e-06,
+      "loss": 0.7479,
+      "step": 4914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828251068458221e-06,
+      "loss": 0.7605,
+      "step": 4915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824400510050606e-06,
+      "loss": 0.699,
+      "step": 4916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820550999496841e-06,
+      "loss": 0.7315,
+      "step": 4917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81670253757629e-06,
+      "loss": 0.7372,
+      "step": 4918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8128551250681166e-06,
+      "loss": 0.681,
+      "step": 4919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809008762751269e-06,
+      "loss": 0.7056,
+      "step": 4920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805163451404481e-06,
+      "loss": 0.8167,
+      "step": 4921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801319191806282e-06,
+      "loss": 0.6618,
+      "step": 4922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797475984734978e-06,
+      "loss": 0.6935,
+      "step": 4923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7936338309686654e-06,
+      "loss": 0.7135,
+      "step": 4924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78979273128523e-06,
+      "loss": 0.7045,
+      "step": 4925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785952686462339e-06,
+      "loss": 0.7526,
+      "step": 4926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782113697277459e-06,
+      "loss": 0.7463,
+      "step": 4927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778275764507817e-06,
+      "loss": 0.747,
+      "step": 4928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774438888930457e-06,
+      "loss": 0.7362,
+      "step": 4929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7706030713221854e-06,
+      "loss": 0.761,
+      "step": 4930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766768312459613e-06,
+      "loss": 0.7319,
+      "step": 4931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762934613119116e-06,
+      "loss": 0.6738,
+      "step": 4932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7591019740768696e-06,
+      "loss": 0.7545,
+      "step": 4933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755270396108834e-06,
+      "loss": 0.6792,
+      "step": 4934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751439879990753e-06,
+      "loss": 0.6713,
+      "step": 4935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747610426498149e-06,
+      "loss": 0.7483,
+      "step": 4936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743782036406336e-06,
+      "loss": 0.7363,
+      "step": 4937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739954710490416e-06,
+      "loss": 0.6938,
+      "step": 4938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736128449525267e-06,
+      "loss": 0.7666,
+      "step": 4939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7323032542855566e-06,
+      "loss": 0.7593,
+      "step": 4940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728479125545734e-06,
+      "loss": 0.7341,
+      "step": 4941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7246560640800325e-06,
+      "loss": 0.6609,
+      "step": 4942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7208340706624745e-06,
+      "loss": 0.6994,
+      "step": 4943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71701314606686e-06,
+      "loss": 0.701,
+      "step": 4944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713193291066775e-06,
+      "loss": 0.7501,
+      "step": 4945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70937450643559e-06,
+      "loss": 0.2441,
+      "step": 4946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705556792946453e-06,
+      "loss": 0.7253,
+      "step": 4947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701740151372306e-06,
+      "loss": 0.7041,
+      "step": 4948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697924582485866e-06,
+      "loss": 0.6745,
+      "step": 4949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6941100870596315e-06,
+      "loss": 0.7193,
+      "step": 4950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69029666586589e-06,
+      "loss": 0.7885,
+      "step": 4951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686484319676703e-06,
+      "loss": 0.7113,
+      "step": 4952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68267304926393e-06,
+      "loss": 0.6575,
+      "step": 4953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678862855399189e-06,
+      "loss": 0.7876,
+      "step": 4954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675053738853904e-06,
+      "loss": 0.7108,
+      "step": 4955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671245700399262e-06,
+      "loss": 0.6717,
+      "step": 4956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667438740806252e-06,
+      "loss": 0.7232,
+      "step": 4957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663632860845619e-06,
+      "loss": 0.7491,
+      "step": 4958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659828061287906e-06,
+      "loss": 0.7514,
+      "step": 4959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65602434290344e-06,
+      "loss": 0.7308,
+      "step": 4960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652221706462321e-06,
+      "loss": 0.7294,
+      "step": 4961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64842015273443e-06,
+      "loss": 0.6908,
+      "step": 4962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644619682489429e-06,
+      "loss": 0.7246,
+      "step": 4963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640820296496768e-06,
+      "loss": 0.7189,
+      "step": 4964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637021995525672e-06,
+      "loss": 0.7189,
+      "step": 4965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6332247803451435e-06,
+      "loss": 0.7592,
+      "step": 4966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629428651723968e-06,
+      "loss": 0.6714,
+      "step": 4967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6256336104307094e-06,
+      "loss": 0.734,
+      "step": 4968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62183965723372e-06,
+      "loss": 0.6861,
+      "step": 4969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618046792901121e-06,
+      "loss": 0.7077,
+      "step": 4970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614255018200816e-06,
+      "loss": 0.6299,
+      "step": 4971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610464333900492e-06,
+      "loss": 0.7429,
+      "step": 4972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606674740767605e-06,
+      "loss": 0.7061,
+      "step": 4973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602886239569412e-06,
+      "loss": 0.6924,
+      "step": 4974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599098831072918e-06,
+      "loss": 0.6514,
+      "step": 4975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595312516044933e-06,
+      "loss": 0.698,
+      "step": 4976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591527295252031e-06,
+      "loss": 0.7004,
+      "step": 4977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587743169460579e-06,
+      "loss": 0.7169,
+      "step": 4978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5839601394367005e-06,
+      "loss": 0.7542,
+      "step": 4979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580178205946311e-06,
+      "loss": 0.6891,
+      "step": 4980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57639736975511e-06,
+      "loss": 0.6435,
+      "step": 4981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5726176316285605e-06,
+      "loss": 0.6088,
+      "step": 4982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568838992331912e-06,
+      "loss": 0.7624,
+      "step": 4983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565061452630188e-06,
+      "loss": 0.6895,
+      "step": 4984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561285013288189e-06,
+      "loss": 0.2132,
+      "step": 4985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5575096750705e-06,
+      "loss": 0.6938,
+      "step": 4986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553735438741473e-06,
+      "loss": 0.664,
+      "step": 4987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549962305065243e-06,
+      "loss": 0.6711,
+      "step": 4988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546190274805716e-06,
+      "loss": 0.7836,
+      "step": 4989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542419348726586e-06,
+      "loss": 0.7154,
+      "step": 4990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538649527591314e-06,
+      "loss": 0.7234,
+      "step": 4991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5348808121631325e-06,
+      "loss": 0.7439,
+      "step": 4992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531113203205065e-06,
+      "loss": 0.7376,
+      "step": 4993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527346701479896e-06,
+      "loss": 0.7391,
+      "step": 4994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523581307750206e-06,
+      "loss": 0.7803,
+      "step": 4995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519817022778321e-06,
+      "loss": 0.7389,
+      "step": 4996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516053847326372e-06,
+      "loss": 0.6822,
+      "step": 4997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512291782156249e-06,
+      "loss": 0.6628,
+      "step": 4998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508530828029619e-06,
+      "loss": 0.7033,
+      "step": 4999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5047709857079295e-06,
+      "loss": 0.727,
+      "step": 5000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5010122559523935e-06,
+      "loss": 0.7621,
+      "step": 5001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497254639524015e-06,
+      "loss": 0.7091,
+      "step": 5002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.493498137183555e-06,
+      "loss": 0.6455,
+      "step": 5003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489742749691559e-06,
+      "loss": 0.6333,
+      "step": 5004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485988477808344e-06,
+      "loss": 0.7352,
+      "step": 5005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482235322293996e-06,
+      "loss": 0.7054,
+      "step": 5006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47848328390839e-06,
+      "loss": 0.6743,
+      "step": 5007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47473236341116e-06,
+      "loss": 0.7516,
+      "step": 5008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47098256156172e-06,
+      "loss": 0.6427,
+      "step": 5009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46723387911925e-06,
+      "loss": 0.6521,
+      "step": 5010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46348631684272e-06,
+      "loss": 0.646,
+      "step": 5011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459739875490857e-06,
+      "loss": 0.6802,
+      "step": 5012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455994555822168e-06,
+      "loss": 0.6766,
+      "step": 5013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4522503585949325e-06,
+      "loss": 0.7049,
+      "step": 5014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448507284567197e-06,
+      "loss": 0.6784,
+      "step": 5015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444765334496793e-06,
+      "loss": 0.7239,
+      "step": 5016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441024509141319e-06,
+      "loss": 0.7265,
+      "step": 5017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4372848092581326e-06,
+      "loss": 0.6958,
+      "step": 5018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4335462356043835e-06,
+      "loss": 0.7121,
+      "step": 5019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429808788936979e-06,
+      "loss": 0.6695,
+      "step": 5020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426072470012614e-06,
+      "loss": 0.7365,
+      "step": 5021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422337279587733e-06,
+      "loss": 0.6453,
+      "step": 5022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418603218418573e-06,
+      "loss": 0.7212,
+      "step": 5023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4148702872611294e-06,
+      "loss": 0.7245,
+      "step": 5024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411138486871175e-06,
+      "loss": 0.7444,
+      "step": 5025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407407818004249e-06,
+      "loss": 0.7152,
+      "step": 5026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403678281415661e-06,
+      "loss": 0.7223,
+      "step": 5027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399949877860503e-06,
+      "loss": 0.6915,
+      "step": 5028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3962226080936245e-06,
+      "loss": 0.7236,
+      "step": 5029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392496472869649e-06,
+      "loss": 0.6904,
+      "step": 5030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388771472942974e-06,
+      "loss": 0.7383,
+      "step": 5031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385047609067758e-06,
+      "loss": 0.7211,
+      "step": 5032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381324881997944e-06,
+      "loss": 0.7348,
+      "step": 5033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377603292487234e-06,
+      "loss": 0.6549,
+      "step": 5034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373882841289102e-06,
+      "loss": 0.6831,
+      "step": 5035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370163529156788e-06,
+      "loss": 0.7027,
+      "step": 5036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366445356843312e-06,
+      "loss": 0.7385,
+      "step": 5037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3627283251014575e-06,
+      "loss": 0.6586,
+      "step": 5038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359012434683765e-06,
+      "loss": 0.7382,
+      "step": 5039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3552976863425654e-06,
+      "loss": 0.7432,
+      "step": 5040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351584080829942e-06,
+      "loss": 0.6847,
+      "step": 5041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347871618897763e-06,
+      "loss": 0.6703,
+      "step": 5042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344160301297642e-06,
+      "loss": 0.7036,
+      "step": 5043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340450128780981e-06,
+      "loss": 0.7153,
+      "step": 5044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3367411020989425e-06,
+      "loss": 0.6747,
+      "step": 5045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333033222002455e-06,
+      "loss": 0.7002,
+      "step": 5046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32932648924222e-06,
+      "loss": 0.7148,
+      "step": 5047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325620904568699e-06,
+      "loss": 0.7075,
+      "step": 5048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321916468732134e-06,
+      "loss": 0.6821,
+      "step": 5049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31821318248252e-06,
+      "loss": 0.7277,
+      "step": 5050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31451104656963e-06,
+      "loss": 0.6409,
+      "step": 5051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310810061742997e-06,
+      "loss": 0.6518,
+      "step": 5052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307110228751921e-06,
+      "loss": 0.7546,
+      "step": 5053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3034115483454795e-06,
+      "loss": 0.7299,
+      "step": 5054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2997140212725045e-06,
+      "loss": 0.7559,
+      "step": 5055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296017648281598e-06,
+      "loss": 0.7632,
+      "step": 5056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292322430121131e-06,
+      "loss": 0.7327,
+      "step": 5057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288628367539234e-06,
+      "loss": 0.667,
+      "step": 5058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2849354612838145e-06,
+      "loss": 0.7104,
+      "step": 5059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2812437121025375e-06,
+      "loss": 0.7659,
+      "step": 5060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277553120742835e-06,
+      "loss": 0.7305,
+      "step": 5061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273863687951902e-06,
+      "loss": 0.7204,
+      "step": 5062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270175414476716e-06,
+      "loss": 0.7289,
+      "step": 5063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266488301063991e-06,
+      "loss": 0.7432,
+      "step": 5064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262802348460226e-06,
+      "loss": 0.7129,
+      "step": 5065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259117557411683e-06,
+      "loss": 0.6952,
+      "step": 5066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255433928664385e-06,
+      "loss": 0.7331,
+      "step": 5067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251751462964122e-06,
+      "loss": 0.7099,
+      "step": 5068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248070161056441e-06,
+      "loss": 0.7192,
+      "step": 5069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24439002368667e-06,
+      "loss": 0.701,
+      "step": 5070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240711051599884e-06,
+      "loss": 0.7768,
+      "step": 5071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237033245540933e-06,
+      "loss": 0.7086,
+      "step": 5072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233356606254423e-06,
+      "loss": 0.7041,
+      "step": 5073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229681134484726e-06,
+      "loss": 0.7692,
+      "step": 5074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226006830975988e-06,
+      "loss": 0.6933,
+      "step": 5075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222333696472105e-06,
+      "loss": 0.6993,
+      "step": 5076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218661731716741e-06,
+      "loss": 0.7024,
+      "step": 5077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2149909374533235e-06,
+      "loss": 0.6722,
+      "step": 5078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21132131442504e-06,
+      "loss": 0.7643,
+      "step": 5079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20765286337485e-06,
+      "loss": 0.7258,
+      "step": 5080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203985585045466e-06,
+      "loss": 0.6679,
+      "step": 5081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2003194801793675e-06,
+      "loss": 0.6116,
+      "step": 5082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19665454951879e-06,
+      "loss": 0.7427,
+      "step": 5083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192990793805747e-06,
+      "loss": 0.7267,
+      "step": 5084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189328213782002e-06,
+      "loss": 0.7025,
+      "step": 5085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1856668101890706e-06,
+      "loss": 0.7073,
+      "step": 5086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1820065837682544e-06,
+      "loss": 0.7004,
+      "step": 5087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178347535260596e-06,
+      "loss": 0.6433,
+      "step": 5088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174689665406919e-06,
+      "loss": 0.7616,
+      "step": 5089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171032974947783e-06,
+      "loss": 0.7245,
+      "step": 5090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167377464623533e-06,
+      "loss": 0.712,
+      "step": 5091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163723135174262e-06,
+      "loss": 0.7574,
+      "step": 5092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160069987339828e-06,
+      "loss": 0.689,
+      "step": 5093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156418021859846e-06,
+      "loss": 0.7146,
+      "step": 5094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152767239473693e-06,
+      "loss": 0.6856,
+      "step": 5095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149117640920513e-06,
+      "loss": 0.6276,
+      "step": 5096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145469226939205e-06,
+      "loss": 0.7905,
+      "step": 5097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141821998268425e-06,
+      "loss": 0.6764,
+      "step": 5098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1381759556465925e-06,
+      "loss": 0.7589,
+      "step": 5099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134531099811885e-06,
+      "loss": 0.7067,
+      "step": 5100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130887431502248e-06,
+      "loss": 0.7006,
+      "step": 5101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1272449514553736e-06,
+      "loss": 0.6899,
+      "step": 5102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123603660408724e-06,
+      "loss": 0.7488,
+      "step": 5103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119963559099514e-06,
+      "loss": 0.6943,
+      "step": 5104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116324648264716e-06,
+      "loss": 0.6704,
+      "step": 5105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1126869286410765e-06,
+      "loss": 0.2503,
+      "step": 5106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109050400965075e-06,
+      "loss": 0.6903,
+      "step": 5107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105415065972978e-06,
+      "loss": 0.6489,
+      "step": 5108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101780924400785e-06,
+      "loss": 0.7839,
+      "step": 5109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098147976984278e-06,
+      "loss": 0.6545,
+      "step": 5110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0945162244589766e-06,
+      "loss": 0.708,
+      "step": 5111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090885667560164e-06,
+      "loss": 0.7176,
+      "step": 5112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087256307022892e-06,
+      "loss": 0.7608,
+      "step": 5113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083628143581959e-06,
+      "loss": 0.7022,
+      "step": 5114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080001177971925e-06,
+      "loss": 0.6552,
+      "step": 5115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076375410927101e-06,
+      "loss": 0.7385,
+      "step": 5116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072750843181572e-06,
+      "loss": 0.7563,
+      "step": 5117
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069127475469163e-06,
+      "loss": 0.6832,
+      "step": 5118
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065505308523462e-06,
+      "loss": 0.6406,
+      "step": 5119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061884343077817e-06,
+      "loss": 0.6747,
+      "step": 5120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058264579865323e-06,
+      "loss": 0.7417,
+      "step": 5121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0546460196188485e-06,
+      "loss": 0.7353,
+      "step": 5122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051028663071005e-06,
+      "loss": 0.7615,
+      "step": 5123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047412510954162e-06,
+      "loss": 0.7318,
+      "step": 5124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043797564000448e-06,
+      "loss": 0.7024,
+      "step": 5125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040183822941744e-06,
+      "loss": 0.6219,
+      "step": 5126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036571288509694e-06,
+      "loss": 0.6911,
+      "step": 5127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032959961435693e-06,
+      "loss": 0.678,
+      "step": 5128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029349842450888e-06,
+      "loss": 0.2293,
+      "step": 5129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025740932286187e-06,
+      "loss": 0.6929,
+      "step": 5130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0221332316722485e-06,
+      "loss": 0.696,
+      "step": 5131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018526741339499e-06,
+      "loss": 0.6807,
+      "step": 5132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014921462018094e-06,
+      "loss": 0.732,
+      "step": 5133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011317394437972e-06,
+      "loss": 0.7215,
+      "step": 5134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007714539328812e-06,
+      "loss": 0.7264,
+      "step": 5135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004112897420045e-06,
+      "loss": 0.7645,
+      "step": 5136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000512469440865e-06,
+      "loss": 0.6746,
+      "step": 5137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99691325612021e-06,
+      "loss": 0.7903,
+      "step": 5138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9933152581867865e-06,
+      "loss": 0.6561,
+      "step": 5139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989718476369041e-06,
+      "loss": 0.6432,
+      "step": 5140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986122911395183e-06,
+      "loss": 0.6625,
+      "step": 5141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982528563993165e-06,
+      "loss": 0.6541,
+      "step": 5142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978935434890707e-06,
+      "loss": 0.7567,
+      "step": 5143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975343524815275e-06,
+      "loss": 0.7085,
+      "step": 5144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971752834494086e-06,
+      "loss": 0.7074,
+      "step": 5145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968163364654112e-06,
+      "loss": 0.6855,
+      "step": 5146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964575116022077e-06,
+      "loss": 0.7116,
+      "step": 5147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960988089324465e-06,
+      "loss": 0.7276,
+      "step": 5148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957402285287502e-06,
+      "loss": 0.7969,
+      "step": 5149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953817704637174e-06,
+      "loss": 0.7208,
+      "step": 5150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950234348099214e-06,
+      "loss": 0.6955,
+      "step": 5151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9466522163991085e-06,
+      "loss": 0.7219,
+      "step": 5152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943071310262107e-06,
+      "loss": 0.7049,
+      "step": 5153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939491630413185e-06,
+      "loss": 0.7273,
+      "step": 5154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9359131775771e-06,
+      "loss": 0.7803,
+      "step": 5155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932335952478337e-06,
+      "loss": 0.6497,
+      "step": 5156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928759955841154e-06,
+      "loss": 0.7469,
+      "step": 5157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925185188389538e-06,
+      "loss": 0.7183,
+      "step": 5158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921611650847238e-06,
+      "loss": 0.7202,
+      "step": 5159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918039343937761e-06,
+      "loss": 0.7139,
+      "step": 5160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914468268384352e-06,
+      "loss": 0.7596,
+      "step": 5161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910898424910016e-06,
+      "loss": 0.6736,
+      "step": 5162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907329814237499e-06,
+      "loss": 0.694,
+      "step": 5163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90376243708931e-06,
+      "loss": 0.8007,
+      "step": 5164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900196294187699e-06,
+      "loss": 0.7049,
+      "step": 5165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896631386254669e-06,
+      "loss": 0.6592,
+      "step": 5166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893067714011971e-06,
+      "loss": 0.7676,
+      "step": 5167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889505278181106e-06,
+      "loss": 0.6964,
+      "step": 5168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885944079483331e-06,
+      "loss": 0.7237,
+      "step": 5169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882384118639649e-06,
+      "loss": 0.7285,
+      "step": 5170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8788253963708e-06,
+      "loss": 0.7327,
+      "step": 5171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875267913397296e-06,
+      "loss": 0.7203,
+      "step": 5172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871711670439376e-06,
+      "loss": 0.6871,
+      "step": 5173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8681566682170535e-06,
+      "loss": 0.7482,
+      "step": 5174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864602907450058e-06,
+      "loss": 0.7727,
+      "step": 5175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861050388857897e-06,
+      "loss": 0.7029,
+      "step": 5176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857499113159813e-06,
+      "loss": 0.7484,
+      "step": 5177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853949081074795e-06,
+      "loss": 0.2548,
+      "step": 5178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850400293321586e-06,
+      "loss": 0.6173,
+      "step": 5179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8468527506186715e-06,
+      "loss": 0.693,
+      "step": 5180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843306453684295e-06,
+      "loss": 0.6919,
+      "step": 5181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839761403236437e-06,
+      "loss": 0.7546,
+      "step": 5182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836217599992831e-06,
+      "loss": 0.7024,
+      "step": 5183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832675044670956e-06,
+      "loss": 0.7263,
+      "step": 5184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829133737988035e-06,
+      "loss": 0.6924,
+      "step": 5185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825593680661052e-06,
+      "loss": 0.7285,
+      "step": 5186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822054873406721e-06,
+      "loss": 0.6899,
+      "step": 5187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818517316941512e-06,
+      "loss": 0.6889,
+      "step": 5188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814981011981638e-06,
+      "loss": 0.2311,
+      "step": 5189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8114459592430663e-06,
+      "loss": 0.715,
+      "step": 5190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8079121594415004e-06,
+      "loss": 0.6759,
+      "step": 5191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804379613292397e-06,
+      "loss": 0.7747,
+      "step": 5192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8008483215109548e-06,
+      "loss": 0.7562,
+      "step": 5193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797318284812118e-06,
+      "loss": 0.6864,
+      "step": 5194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7937895039105864e-06,
+      "loss": 0.7293,
+      "step": 5195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.790261979520795e-06,
+      "loss": 0.7039,
+      "step": 5196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7867357123569258e-06,
+      "loss": 0.7407,
+      "step": 5197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78321070313291e-06,
+      "loss": 0.7032,
+      "step": 5198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779686952562419e-06,
+      "loss": 0.6253,
+      "step": 5199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.776164461358882e-06,
+      "loss": 0.7501,
+      "step": 5200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7726432302354498e-06,
+      "loss": 0.7424,
+      "step": 5201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7691232599050444e-06,
+      "loss": 0.7135,
+      "step": 5202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7656045510803153e-06,
+      "loss": 0.6928,
+      "step": 5203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762087104473662e-06,
+      "loss": 0.7732,
+      "step": 5204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7585709207972285e-06,
+      "loss": 0.7627,
+      "step": 5205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7550560007628988e-06,
+      "loss": 0.7435,
+      "step": 5206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751542345082312e-06,
+      "loss": 0.6548,
+      "step": 5207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748029954466841e-06,
+      "loss": 0.7862,
+      "step": 5208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7445188296276047e-06,
+      "loss": 0.6707,
+      "step": 5209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741008971275468e-06,
+      "loss": 0.2532,
+      "step": 5210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737500380121034e-06,
+      "loss": 0.6903,
+      "step": 5211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7339930568746606e-06,
+      "loss": 0.7592,
+      "step": 5212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730487002246438e-06,
+      "loss": 0.7368,
+      "step": 5213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7269822169462035e-06,
+      "loss": 0.7447,
+      "step": 5214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723478701683535e-06,
+      "loss": 0.7227,
+      "step": 5215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71997645716776e-06,
+      "loss": 0.7449,
+      "step": 5216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7164754841079466e-06,
+      "loss": 0.6946,
+      "step": 5217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.712975783212891e-06,
+      "loss": 0.7142,
+      "step": 5218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7094773551911558e-06,
+      "loss": 0.703,
+      "step": 5219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705980200751026e-06,
+      "loss": 0.7964,
+      "step": 5220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702484320600548e-06,
+      "loss": 0.7033,
+      "step": 5221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6989897154474843e-06,
+      "loss": 0.6912,
+      "step": 5222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6954963859993654e-06,
+      "loss": 0.6881,
+      "step": 5223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692004332963448e-06,
+      "loss": 0.7589,
+      "step": 5224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6885135570467358e-06,
+      "loss": 0.6274,
+      "step": 5225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6850240589559704e-06,
+      "loss": 0.7513,
+      "step": 5226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681535839397634e-06,
+      "loss": 0.7327,
+      "step": 5227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6780488990779616e-06,
+      "loss": 0.7495,
+      "step": 5228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674563238702915e-06,
+      "loss": 0.724,
+      "step": 5229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6710788589782033e-06,
+      "loss": 0.7015,
+      "step": 5230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6675957606092747e-06,
+      "loss": 0.7368,
+      "step": 5231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6641139443013163e-06,
+      "loss": 0.6934,
+      "step": 5232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6606334107592634e-06,
+      "loss": 0.6749,
+      "step": 5233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6571541606877836e-06,
+      "loss": 0.6639,
+      "step": 5234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653676194791288e-06,
+      "loss": 0.7656,
+      "step": 5235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650199513773921e-06,
+      "loss": 0.7079,
+      "step": 5236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6467241183395816e-06,
+      "loss": 0.7897,
+      "step": 5237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6432500091918997e-06,
+      "loss": 0.7196,
+      "step": 5238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6397771870342335e-06,
+      "loss": 0.7787,
+      "step": 5239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6363056525697038e-06,
+      "loss": 0.7585,
+      "step": 5240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6328354065011497e-06,
+      "loss": 0.71,
+      "step": 5241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62936644953117e-06,
+      "loss": 0.7402,
+      "step": 5242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6258987823620773e-06,
+      "loss": 0.6386,
+      "step": 5243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6224324056959472e-06,
+      "loss": 0.733,
+      "step": 5244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61896732023458e-06,
+      "loss": 0.7376,
+      "step": 5245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.615503526679518e-06,
+      "loss": 0.6469,
+      "step": 5246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6120410257320427e-06,
+      "loss": 0.712,
+      "step": 5247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6085798180931684e-06,
+      "loss": 0.751,
+      "step": 5248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.605119904463662e-06,
+      "loss": 0.7016,
+      "step": 5249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6016612855440148e-06,
+      "loss": 0.7471,
+      "step": 5250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598203962034459e-06,
+      "loss": 0.6273,
+      "step": 5251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5947479346349656e-06,
+      "loss": 0.7374,
+      "step": 5252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5912932040452398e-06,
+      "loss": 0.7685,
+      "step": 5253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587839770964736e-06,
+      "loss": 0.7438,
+      "step": 5254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5843876360926323e-06,
+      "loss": 0.673,
+      "step": 5255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580936800127851e-06,
+      "loss": 0.6756,
+      "step": 5256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5774872637690484e-06,
+      "loss": 0.6798,
+      "step": 5257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5740390277146152e-06,
+      "loss": 0.658,
+      "step": 5258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5705920926626913e-06,
+      "loss": 0.7074,
+      "step": 5259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5671464593111395e-06,
+      "loss": 0.6941,
+      "step": 5260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5637021283575635e-06,
+      "loss": 0.6584,
+      "step": 5261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5602591004993025e-06,
+      "loss": 0.7721,
+      "step": 5262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556817376433438e-06,
+      "loss": 0.7066,
+      "step": 5263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5533769568567844e-06,
+      "loss": 0.7193,
+      "step": 5264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549937842465879e-06,
+      "loss": 0.754,
+      "step": 5265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5465000339570166e-06,
+      "loss": 0.7085,
+      "step": 5266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543063532026211e-06,
+      "loss": 0.7569,
+      "step": 5267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5396283373692273e-06,
+      "loss": 0.6823,
+      "step": 5268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5361944506815436e-06,
+      "loss": 0.771,
+      "step": 5269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5327618726583945e-06,
+      "loss": 0.6988,
+      "step": 5270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5293306039947394e-06,
+      "loss": 0.7127,
+      "step": 5271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525900645385274e-06,
+      "loss": 0.7338,
+      "step": 5272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522471997524427e-06,
+      "loss": 0.6542,
+      "step": 5273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519044661106362e-06,
+      "loss": 0.7708,
+      "step": 5274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515618636824987e-06,
+      "loss": 0.739,
+      "step": 5275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5121939253739302e-06,
+      "loss": 0.6539,
+      "step": 5276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508770527446561e-06,
+      "loss": 0.6783,
+      "step": 5277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505348443735982e-06,
+      "loss": 0.6803,
+      "step": 5278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5019276749350262e-06,
+      "loss": 0.7063,
+      "step": 5279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.498508221736271e-06,
+      "loss": 0.7389,
+      "step": 5280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4950900848320144e-06,
+      "loss": 0.7026,
+      "step": 5281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491673264914297e-06,
+      "loss": 0.7635,
+      "step": 5282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4882577626748883e-06,
+      "loss": 0.802,
+      "step": 5283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484843578805287e-06,
+      "loss": 0.7379,
+      "step": 5284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481430713996743e-06,
+      "loss": 0.7228,
+      "step": 5285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478019168940211e-06,
+      "loss": 0.7072,
+      "step": 5286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474608944326404e-06,
+      "loss": 0.7132,
+      "step": 5287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4712000408457514e-06,
+      "loss": 0.7673,
+      "step": 5288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4677924591884305e-06,
+      "loss": 0.7343,
+      "step": 5289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.464386200044331e-06,
+      "loss": 0.6595,
+      "step": 5290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.460981264103086e-06,
+      "loss": 0.7702,
+      "step": 5291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.457577652054067e-06,
+      "loss": 0.7058,
+      "step": 5292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4541753645863686e-06,
+      "loss": 0.7136,
+      "step": 5293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450774402388816e-06,
+      "loss": 0.7165,
+      "step": 5294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447374766149968e-06,
+      "loss": 0.6895,
+      "step": 5295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4439764565581237e-06,
+      "loss": 0.7343,
+      "step": 5296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440579474301301e-06,
+      "loss": 0.7701,
+      "step": 5297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437183820067255e-06,
+      "loss": 0.2472,
+      "step": 5298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43378949454347e-06,
+      "loss": 0.2457,
+      "step": 5299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4303964984171625e-06,
+      "loss": 0.6287,
+      "step": 5300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4270048323752823e-06,
+      "loss": 0.6579,
+      "step": 5301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423614497104507e-06,
+      "loss": 0.6971,
+      "step": 5302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4202254932912437e-06,
+      "loss": 0.6677,
+      "step": 5303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4168378216216323e-06,
+      "loss": 0.7149,
+      "step": 5304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413451482781538e-06,
+      "loss": 0.7256,
+      "step": 5305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4100664774565707e-06,
+      "loss": 0.603,
+      "step": 5306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.406682806332046e-06,
+      "loss": 0.6965,
+      "step": 5307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.403300470093033e-06,
+      "loss": 0.758,
+      "step": 5308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3999194694243153e-06,
+      "loss": 0.7538,
+      "step": 5309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3965398050104213e-06,
+      "loss": 0.7082,
+      "step": 5310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3931614775355894e-06,
+      "loss": 0.8172,
+      "step": 5311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3897844876837948e-06,
+      "loss": 0.6261,
+      "step": 5312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386408836138754e-06,
+      "loss": 0.6279,
+      "step": 5313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.383034523583898e-06,
+      "loss": 0.6947,
+      "step": 5314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.37966155070239e-06,
+      "loss": 0.7559,
+      "step": 5315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3762899181771225e-06,
+      "loss": 0.7454,
+      "step": 5316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3729196266907225e-06,
+      "loss": 0.7109,
+      "step": 5317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369550676925538e-06,
+      "loss": 0.7315,
+      "step": 5318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3661830695636466e-06,
+      "loss": 0.6332,
+      "step": 5319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362816805286857e-06,
+      "loss": 0.6153,
+      "step": 5320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359451884776701e-06,
+      "loss": 0.6509,
+      "step": 5321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3560883087144468e-06,
+      "loss": 0.6991,
+      "step": 5322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352726077781083e-06,
+      "loss": 0.684,
+      "step": 5323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3493651926573276e-06,
+      "loss": 0.7224,
+      "step": 5324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3460056540236286e-06,
+      "loss": 0.2464,
+      "step": 5325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3426474625601535e-06,
+      "loss": 0.7195,
+      "step": 5326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33929061894681e-06,
+      "loss": 0.693,
+      "step": 5327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335935123863224e-06,
+      "loss": 0.7225,
+      "step": 5328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3325809779887496e-06,
+      "loss": 0.7038,
+      "step": 5329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3292281820024697e-06,
+      "loss": 0.7315,
+      "step": 5330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325876736583187e-06,
+      "loss": 0.6342,
+      "step": 5331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3225266424094483e-06,
+      "loss": 0.7384,
+      "step": 5332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3191779001595003e-06,
+      "loss": 0.71,
+      "step": 5333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3158305105113407e-06,
+      "loss": 0.7648,
+      "step": 5334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3124844741426787e-06,
+      "loss": 0.6814,
+      "step": 5335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.309139791730961e-06,
+      "loss": 0.6875,
+      "step": 5336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305796463953346e-06,
+      "loss": 0.7267,
+      "step": 5337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302454491486723e-06,
+      "loss": 0.7068,
+      "step": 5338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.299113875007717e-06,
+      "loss": 0.7281,
+      "step": 5339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295774615192667e-06,
+      "loss": 0.7227,
+      "step": 5340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2924367127176404e-06,
+      "loss": 0.7339,
+      "step": 5341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2891001682584265e-06,
+      "loss": 0.8024,
+      "step": 5342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2857649824905513e-06,
+      "loss": 0.7157,
+      "step": 5343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282431156089254e-06,
+      "loss": 0.7657,
+      "step": 5344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2790986897295007e-06,
+      "loss": 0.7175,
+      "step": 5345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275767584085987e-06,
+      "loss": 0.707,
+      "step": 5346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2724378398331238e-06,
+      "loss": 0.6993,
+      "step": 5347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26910945764506e-06,
+      "loss": 0.6981,
+      "step": 5348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.265782438195657e-06,
+      "loss": 0.7285,
+      "step": 5349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2624567821585064e-06,
+      "loss": 0.7506,
+      "step": 5350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259132490206919e-06,
+      "loss": 0.7697,
+      "step": 5351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255809563013932e-06,
+      "loss": 0.6897,
+      "step": 5352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252488001252314e-06,
+      "loss": 0.6617,
+      "step": 5353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2491678055945377e-06,
+      "loss": 0.7572,
+      "step": 5354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24584897671282e-06,
+      "loss": 0.678,
+      "step": 5355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.242531515279087e-06,
+      "loss": 0.6801,
+      "step": 5356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239215421965002e-06,
+      "loss": 0.7211,
+      "step": 5357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2359006974419336e-06,
+      "loss": 0.6887,
+      "step": 5358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2325873423809817e-06,
+      "loss": 0.7162,
+      "step": 5359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2292753574529754e-06,
+      "loss": 0.6172,
+      "step": 5360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225964743328458e-06,
+      "loss": 0.7625,
+      "step": 5361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.222655500677697e-06,
+      "loss": 0.6805,
+      "step": 5362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.219347630170684e-06,
+      "loss": 0.7194,
+      "step": 5363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216041132477128e-06,
+      "loss": 0.2359,
+      "step": 5364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2127360082664693e-06,
+      "loss": 0.7396,
+      "step": 5365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2094322582078628e-06,
+      "loss": 0.6574,
+      "step": 5366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.206129882970186e-06,
+      "loss": 0.7992,
+      "step": 5367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2028288832220356e-06,
+      "loss": 0.6722,
+      "step": 5368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199529259631741e-06,
+      "loss": 0.7654,
+      "step": 5369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1962310128673446e-06,
+      "loss": 0.732,
+      "step": 5370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.192934143596601e-06,
+      "loss": 0.7126,
+      "step": 5371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189638652487005e-06,
+      "loss": 0.6123,
+      "step": 5372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186344540205756e-06,
+      "loss": 0.2582,
+      "step": 5373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1830518074197923e-06,
+      "loss": 0.7906,
+      "step": 5374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1797604547957484e-06,
+      "loss": 0.6671,
+      "step": 5375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1764704830000027e-06,
+      "loss": 0.7113,
+      "step": 5376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1731818926986403e-06,
+      "loss": 0.7532,
+      "step": 5377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1698946845574706e-06,
+      "loss": 0.6532,
+      "step": 5378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166608859242023e-06,
+      "loss": 0.726,
+      "step": 5379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1633244174175447e-06,
+      "loss": 0.7633,
+      "step": 5380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1600413597490097e-06,
+      "loss": 0.7155,
+      "step": 5381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156759686901105e-06,
+      "loss": 0.7073,
+      "step": 5382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1534793995382398e-06,
+      "loss": 0.7565,
+      "step": 5383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1502004983245413e-06,
+      "loss": 0.6193,
+      "step": 5384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1469229839238545e-06,
+      "loss": 0.6767,
+      "step": 5385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1436468569997526e-06,
+      "loss": 0.7549,
+      "step": 5386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140372118215517e-06,
+      "loss": 0.6505,
+      "step": 5387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137098768234155e-06,
+      "loss": 0.7327,
+      "step": 5388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1338268077183842e-06,
+      "loss": 0.7409,
+      "step": 5389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1305562373306564e-06,
+      "loss": 0.6427,
+      "step": 5390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127287057733127e-06,
+      "loss": 0.756,
+      "step": 5391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124019269587677e-06,
+      "loss": 0.2582,
+      "step": 5392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120752873555902e-06,
+      "loss": 0.7179,
+      "step": 5393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1174878702991175e-06,
+      "loss": 0.715,
+      "step": 5394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114224260478361e-06,
+      "loss": 0.7003,
+      "step": 5395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1109620447543852e-06,
+      "loss": 0.7233,
+      "step": 5396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1077012237876502e-06,
+      "loss": 0.7277,
+      "step": 5397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104441798238352e-06,
+      "loss": 0.6714,
+      "step": 5398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101183768766389e-06,
+      "loss": 0.7661,
+      "step": 5399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0979271360313934e-06,
+      "loss": 0.7357,
+      "step": 5400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0946719006926908e-06,
+      "loss": 0.7575,
+      "step": 5401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091418063409347e-06,
+      "loss": 0.662,
+      "step": 5402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0881656248401315e-06,
+      "loss": 0.6586,
+      "step": 5403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084914585643536e-06,
+      "loss": 0.6737,
+      "step": 5404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081664946477765e-06,
+      "loss": 0.7215,
+      "step": 5405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.07841670800074e-06,
+      "loss": 0.5802,
+      "step": 5406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075169870870106e-06,
+      "loss": 0.7525,
+      "step": 5407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0719244357432163e-06,
+      "loss": 0.736,
+      "step": 5408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068680403277142e-06,
+      "loss": 0.7059,
+      "step": 5409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0654377741286735e-06,
+      "loss": 0.6913,
+      "step": 5410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0621965489543082e-06,
+      "loss": 0.7852,
+      "step": 5411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058956728410275e-06,
+      "loss": 0.7098,
+      "step": 5412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0557183131525037e-06,
+      "loss": 0.743,
+      "step": 5413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0524813038366475e-06,
+      "loss": 0.7311,
+      "step": 5414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0492457011180675e-06,
+      "loss": 0.5818,
+      "step": 5415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0460115056518523e-06,
+      "loss": 0.6607,
+      "step": 5416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.042778718092798e-06,
+      "loss": 0.7187,
+      "step": 5417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0395473390954066e-06,
+      "loss": 0.7482,
+      "step": 5418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0363173693139136e-06,
+      "loss": 0.738,
+      "step": 5419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0330888094022536e-06,
+      "loss": 0.6564,
+      "step": 5420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0298616600140928e-06,
+      "loss": 0.6834,
+      "step": 5421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026635921802786e-06,
+      "loss": 0.7237,
+      "step": 5422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0234115954214293e-06,
+      "loss": 0.7008,
+      "step": 5423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0201886815228165e-06,
+      "loss": 0.6541,
+      "step": 5424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0169671807594604e-06,
+      "loss": 0.6566,
+      "step": 5425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0137470937835865e-06,
+      "loss": 0.6167,
+      "step": 5426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010528421247133e-06,
+      "loss": 0.7148,
+      "step": 5427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0073111638017594e-06,
+      "loss": 0.77,
+      "step": 5428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.00409532209883e-06,
+      "loss": 0.7798,
+      "step": 5429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0008808967894234e-06,
+      "loss": 0.7437,
+      "step": 5430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997667888524336e-06,
+      "loss": 0.7085,
+      "step": 5431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9944562979540704e-06,
+      "loss": 0.7142,
+      "step": 5432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9912461257288517e-06,
+      "loss": 0.7122,
+      "step": 5433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988037372498611e-06,
+      "loss": 0.7348,
+      "step": 5434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9848300389129936e-06,
+      "loss": 0.7601,
+      "step": 5435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9816241256213572e-06,
+      "loss": 0.6019,
+      "step": 5436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978419633272769e-06,
+      "loss": 0.7589,
+      "step": 5437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.97521656251602e-06,
+      "loss": 0.6588,
+      "step": 5438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9720149139995945e-06,
+      "loss": 0.7445,
+      "step": 5439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9688146883717083e-06,
+      "loss": 0.7296,
+      "step": 5440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965615886280273e-06,
+      "loss": 0.7672,
+      "step": 5441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9624185083729296e-06,
+      "loss": 0.6887,
+      "step": 5442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9592225552970123e-06,
+      "loss": 0.7424,
+      "step": 5443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9560280276995724e-06,
+      "loss": 0.6585,
+      "step": 5444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9528349262273824e-06,
+      "loss": 0.6986,
+      "step": 5445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.949643251526916e-06,
+      "loss": 0.7839,
+      "step": 5446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9464530042443617e-06,
+      "loss": 0.748,
+      "step": 5447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9432641850256137e-06,
+      "loss": 0.7435,
+      "step": 5448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9400767945162877e-06,
+      "loss": 0.6665,
+      "step": 5449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936890833361702e-06,
+      "loss": 0.7192,
+      "step": 5450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9337063022068867e-06,
+      "loss": 0.7676,
+      "step": 5451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9305232016965846e-06,
+      "loss": 0.7003,
+      "step": 5452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927341532475242e-06,
+      "loss": 0.6588,
+      "step": 5453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924161295187028e-06,
+      "loss": 0.7235,
+      "step": 5454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9209824904758134e-06,
+      "loss": 0.6793,
+      "step": 5455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9178051189851776e-06,
+      "loss": 0.7178,
+      "step": 5456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9146291813584126e-06,
+      "loss": 0.6489,
+      "step": 5457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911454678238519e-06,
+      "loss": 0.6925,
+      "step": 5458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9082816102682123e-06,
+      "loss": 0.7935,
+      "step": 5459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9051099780899108e-06,
+      "loss": 0.7113,
+      "step": 5460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901939782345743e-06,
+      "loss": 0.7362,
+      "step": 5461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.898771023677547e-06,
+      "loss": 0.7559,
+      "step": 5462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8956037027268758e-06,
+      "loss": 0.7576,
+      "step": 5463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8924378201349857e-06,
+      "loss": 0.6629,
+      "step": 5464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889273376542835e-06,
+      "loss": 0.7068,
+      "step": 5465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8861103725911067e-06,
+      "loss": 0.6826,
+      "step": 5466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8829488089201763e-06,
+      "loss": 0.6244,
+      "step": 5467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8797886861701462e-06,
+      "loss": 0.669,
+      "step": 5468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876630004980803e-06,
+      "loss": 0.6669,
+      "step": 5469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873472765991664e-06,
+      "loss": 0.7866,
+      "step": 5470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870316969841941e-06,
+      "loss": 0.7401,
+      "step": 5471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867162617170559e-06,
+      "loss": 0.7362,
+      "step": 5472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864009708616149e-06,
+      "loss": 0.2323,
+      "step": 5473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8608582448170464e-06,
+      "loss": 0.6881,
+      "step": 5474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857708226411304e-06,
+      "loss": 0.6896,
+      "step": 5475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8545596540366737e-06,
+      "loss": 0.6759,
+      "step": 5476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851412528330616e-06,
+      "loss": 0.641,
+      "step": 5477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848266849930299e-06,
+      "loss": 0.6451,
+      "step": 5478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8451226194725958e-06,
+      "loss": 0.6112,
+      "step": 5479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841979837594092e-06,
+      "loss": 0.7409,
+      "step": 5480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838838504931076e-06,
+      "loss": 0.7537,
+      "step": 5481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8356986221195436e-06,
+      "loss": 0.2401,
+      "step": 5482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832560189795195e-06,
+      "loss": 0.6993,
+      "step": 5483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829423208593435e-06,
+      "loss": 0.6929,
+      "step": 5484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8262876791493876e-06,
+      "loss": 0.7109,
+      "step": 5485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8231536020978623e-06,
+      "loss": 0.7728,
+      "step": 5486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8200209780733943e-06,
+      "loss": 0.6441,
+      "step": 5487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816889807710208e-06,
+      "loss": 0.6759,
+      "step": 5488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8137600916422537e-06,
+      "loss": 0.707,
+      "step": 5489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810631830503162e-06,
+      "loss": 0.7698,
+      "step": 5490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8075050249262846e-06,
+      "loss": 0.673,
+      "step": 5491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8043796755446806e-06,
+      "loss": 0.7199,
+      "step": 5492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.801255782991107e-06,
+      "loss": 0.7552,
+      "step": 5493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7981333478980276e-06,
+      "loss": 0.6397,
+      "step": 5494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7950123708976094e-06,
+      "loss": 0.7511,
+      "step": 5495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7918928526217326e-06,
+      "loss": 0.7041,
+      "step": 5496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.788774793701973e-06,
+      "loss": 0.6427,
+      "step": 5497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7856581947696137e-06,
+      "loss": 0.7181,
+      "step": 5498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7825430564556423e-06,
+      "loss": 0.6824,
+      "step": 5499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779429379390749e-06,
+      "loss": 0.674,
+      "step": 5500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7763171642053343e-06,
+      "loss": 0.6985,
+      "step": 5501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7732064115294964e-06,
+      "loss": 0.7189,
+      "step": 5502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7700971219930407e-06,
+      "loss": 0.634,
+      "step": 5503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766989296225473e-06,
+      "loss": 0.7274,
+      "step": 5504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7638829348560027e-06,
+      "loss": 0.6754,
+      "step": 5505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7607780385135553e-06,
+      "loss": 0.6212,
+      "step": 5506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.757674607826735e-06,
+      "loss": 0.7042,
+      "step": 5507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754572643423874e-06,
+      "loss": 0.7184,
+      "step": 5508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7514721459329898e-06,
+      "loss": 0.701,
+      "step": 5509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7483731159818215e-06,
+      "loss": 0.6775,
+      "step": 5510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7452755541977892e-06,
+      "loss": 0.6951,
+      "step": 5511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7421794612080268e-06,
+      "loss": 0.7653,
+      "step": 5512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7390848376393774e-06,
+      "loss": 0.6956,
+      "step": 5513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7359916841183754e-06,
+      "loss": 0.6903,
+      "step": 5514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7329000012712615e-06,
+      "loss": 0.7606,
+      "step": 5515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7298097897239806e-06,
+      "loss": 0.6869,
+      "step": 5516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7267210501021746e-06,
+      "loss": 0.6556,
+      "step": 5517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7236337830311965e-06,
+      "loss": 0.2191,
+      "step": 5518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720547989136092e-06,
+      "loss": 0.7169,
+      "step": 5519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7174636690416135e-06,
+      "loss": 0.728,
+      "step": 5520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.714380823372209e-06,
+      "loss": 0.6843,
+      "step": 5521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.71129945275204e-06,
+      "loss": 0.6661,
+      "step": 5522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7082195578049586e-06,
+      "loss": 0.626,
+      "step": 5523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7051411391545212e-06,
+      "loss": 0.7043,
+      "step": 5524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7020641974239857e-06,
+      "loss": 0.7437,
+      "step": 5525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6989887332363074e-06,
+      "loss": 0.221,
+      "step": 5526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6959147472141533e-06,
+      "loss": 0.7912,
+      "step": 5527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69284223997988e-06,
+      "loss": 0.7276,
+      "step": 5528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6897712121555485e-06,
+      "loss": 0.2402,
+      "step": 5529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6867016643629195e-06,
+      "loss": 0.246,
+      "step": 5530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683633597223454e-06,
+      "loss": 0.7787,
+      "step": 5531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6805670113583204e-06,
+      "loss": 0.6571,
+      "step": 5532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67750190738837e-06,
+      "loss": 0.7225,
+      "step": 5533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6744382859341745e-06,
+      "loss": 0.7622,
+      "step": 5534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6713761476159894e-06,
+      "loss": 0.7233,
+      "step": 5535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668315493053786e-06,
+      "loss": 0.7432,
+      "step": 5536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6652563228672146e-06,
+      "loss": 0.7071,
+      "step": 5537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662198637675637e-06,
+      "loss": 0.6971,
+      "step": 5538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65914243809812e-06,
+      "loss": 0.687,
+      "step": 5539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65608772475342e-06,
+      "loss": 0.6119,
+      "step": 5540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6530344982599953e-06,
+      "loss": 0.6969,
+      "step": 5541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6499827592359993e-06,
+      "loss": 0.6605,
+      "step": 5542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6469325082992946e-06,
+      "loss": 0.712,
+      "step": 5543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.643883746067434e-06,
+      "loss": 0.7852,
+      "step": 5544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6408364731576695e-06,
+      "loss": 0.6867,
+      "step": 5545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6377906901869553e-06,
+      "loss": 0.7265,
+      "step": 5546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6347463977719367e-06,
+      "loss": 0.743,
+      "step": 5547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6317035965289704e-06,
+      "loss": 0.7328,
+      "step": 5548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6286622870741006e-06,
+      "loss": 0.6825,
+      "step": 5549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625622470023066e-06,
+      "loss": 0.7256,
+      "step": 5550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622584145991315e-06,
+      "loss": 0.726,
+      "step": 5551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6195473155939834e-06,
+      "loss": 0.6739,
+      "step": 5552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616511979445918e-06,
+      "loss": 0.6746,
+      "step": 5553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6134781381616416e-06,
+      "loss": 0.678,
+      "step": 5554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6104457923553974e-06,
+      "loss": 0.2398,
+      "step": 5555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60741494264111e-06,
+      "loss": 0.6797,
+      "step": 5556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6043855896324076e-06,
+      "loss": 0.7473,
+      "step": 5557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6013577339426143e-06,
+      "loss": 0.7102,
+      "step": 5558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5983313761847485e-06,
+      "loss": 0.6755,
+      "step": 5559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595306516971532e-06,
+      "loss": 0.2452,
+      "step": 5560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.592283156915377e-06,
+      "loss": 0.7864,
+      "step": 5561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5892612966283935e-06,
+      "loss": 0.7106,
+      "step": 5562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5862409367223905e-06,
+      "loss": 0.6741,
+      "step": 5563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5832220778088657e-06,
+      "loss": 0.6815,
+      "step": 5564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5802047204990253e-06,
+      "loss": 0.6944,
+      "step": 5565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5771888654037626e-06,
+      "loss": 0.6815,
+      "step": 5566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574174513133667e-06,
+      "loss": 0.7708,
+      "step": 5567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5711616642990246e-06,
+      "loss": 0.6368,
+      "step": 5568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5681503195098212e-06,
+      "loss": 0.6785,
+      "step": 5569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5651404793757372e-06,
+      "loss": 0.6879,
+      "step": 5570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5621321445061354e-06,
+      "loss": 0.7359,
+      "step": 5571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559125315510095e-06,
+      "loss": 0.6787,
+      "step": 5572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.556119992996372e-06,
+      "loss": 0.6756,
+      "step": 5573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553116177573435e-06,
+      "loss": 0.7875,
+      "step": 5574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5501138698494256e-06,
+      "loss": 0.7486,
+      "step": 5575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547113070432201e-06,
+      "loss": 0.7452,
+      "step": 5576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5441137799293023e-06,
+      "loss": 0.7454,
+      "step": 5577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541115998947965e-06,
+      "loss": 0.723,
+      "step": 5578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.538119728095123e-06,
+      "loss": 0.7378,
+      "step": 5579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.535124967977397e-06,
+      "loss": 0.2268,
+      "step": 5580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5321317192011164e-06,
+      "loss": 0.724,
+      "step": 5581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5291399823722897e-06,
+      "loss": 0.6835,
+      "step": 5582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5261497580966267e-06,
+      "loss": 0.7419,
+      "step": 5583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5231610469795286e-06,
+      "loss": 0.7324,
+      "step": 5584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5201738496260873e-06,
+      "loss": 0.2325,
+      "step": 5585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5171881666411e-06,
+      "loss": 0.7452,
+      "step": 5586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5142039986290447e-06,
+      "loss": 0.6876,
+      "step": 5587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5112213461940984e-06,
+      "loss": 0.7231,
+      "step": 5588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508240209940127e-06,
+      "loss": 0.7315,
+      "step": 5589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505260590470694e-06,
+      "loss": 0.7194,
+      "step": 5590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5022824883890564e-06,
+      "loss": 0.2452,
+      "step": 5591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4993059042981617e-06,
+      "loss": 0.6515,
+      "step": 5592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496330838800648e-06,
+      "loss": 0.7276,
+      "step": 5593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493357292498846e-06,
+      "loss": 0.7677,
+      "step": 5594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.490385265994788e-06,
+      "loss": 0.6684,
+      "step": 5595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48741475989019e-06,
+      "loss": 0.754,
+      "step": 5596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.484445774786455e-06,
+      "loss": 0.7854,
+      "step": 5597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481478311284691e-06,
+      "loss": 0.6809,
+      "step": 5598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478512369985687e-06,
+      "loss": 0.7086,
+      "step": 5599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4755479514899394e-06,
+      "loss": 0.7152,
+      "step": 5600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4725850563976107e-06,
+      "loss": 0.7379,
+      "step": 5601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4696236853085808e-06,
+      "loss": 0.6784,
+      "step": 5602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4666638388224062e-06,
+      "loss": 0.6965,
+      "step": 5603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4637055175383384e-06,
+      "loss": 0.7502,
+      "step": 5604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4607487220553205e-06,
+      "loss": 0.7343,
+      "step": 5605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4577934529719826e-06,
+      "loss": 0.634,
+      "step": 5606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.454839710886657e-06,
+      "loss": 0.724,
+      "step": 5607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451887496397356e-06,
+      "loss": 0.6599,
+      "step": 5608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.448936810101786e-06,
+      "loss": 0.6527,
+      "step": 5609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4459876525973438e-06,
+      "loss": 0.6973,
+      "step": 5610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4430400244811135e-06,
+      "loss": 0.2657,
+      "step": 5611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44009392634988e-06,
+      "loss": 0.7383,
+      "step": 5612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437149358800108e-06,
+      "loss": 0.7145,
+      "step": 5613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434206322427957e-06,
+      "loss": 0.696,
+      "step": 5614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43126481782927e-06,
+      "loss": 0.752,
+      "step": 5615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428324845599592e-06,
+      "loss": 0.7379,
+      "step": 5616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4253864063341517e-06,
+      "loss": 0.7318,
+      "step": 5617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422449500627857e-06,
+      "loss": 0.5708,
+      "step": 5618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4195141290753243e-06,
+      "loss": 0.6587,
+      "step": 5619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416580292270844e-06,
+      "loss": 0.7586,
+      "step": 5620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.41364799080841e-06,
+      "loss": 0.2669,
+      "step": 5621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4107172252816856e-06,
+      "loss": 0.7442,
+      "step": 5622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4077879962840446e-06,
+      "loss": 0.6798,
+      "step": 5623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404860304408535e-06,
+      "loss": 0.7787,
+      "step": 5624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4019341502478997e-06,
+      "loss": 0.7741,
+      "step": 5625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3990095343945694e-06,
+      "loss": 0.7207,
+      "step": 5626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3960864574406593e-06,
+      "loss": 0.6399,
+      "step": 5627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3931649199779814e-06,
+      "loss": 0.737,
+      "step": 5628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3902449225980316e-06,
+      "loss": 0.6814,
+      "step": 5629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3873264658919902e-06,
+      "loss": 0.6987,
+      "step": 5630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3844095504507313e-06,
+      "loss": 0.7022,
+      "step": 5631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3814941768648105e-06,
+      "loss": 0.6026,
+      "step": 5632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3785803457244814e-06,
+      "loss": 0.7221,
+      "step": 5633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375668057619678e-06,
+      "loss": 0.7121,
+      "step": 5634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3727573131400218e-06,
+      "loss": 0.721,
+      "step": 5635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.369848112874823e-06,
+      "loss": 0.6522,
+      "step": 5636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.366940457413077e-06,
+      "loss": 0.6687,
+      "step": 5637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364034347343478e-06,
+      "loss": 0.7349,
+      "step": 5638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3611297832543855e-06,
+      "loss": 0.7463,
+      "step": 5639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3582267657338686e-06,
+      "loss": 0.6306,
+      "step": 5640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355325295369666e-06,
+      "loss": 0.7171,
+      "step": 5641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352425372749221e-06,
+      "loss": 0.7279,
+      "step": 5642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3495269984596437e-06,
+      "loss": 0.701,
+      "step": 5643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3466301730877384e-06,
+      "loss": 0.7156,
+      "step": 5644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3437348972200058e-06,
+      "loss": 0.7192,
+      "step": 5645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3408411714426192e-06,
+      "loss": 0.6879,
+      "step": 5646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3379489963414457e-06,
+      "loss": 0.694,
+      "step": 5647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3350583725020327e-06,
+      "loss": 0.6992,
+      "step": 5648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3321693005096224e-06,
+      "loss": 0.6799,
+      "step": 5649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329281780949134e-06,
+      "loss": 0.7517,
+      "step": 5650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.326395814405176e-06,
+      "loss": 0.7205,
+      "step": 5651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323511401462043e-06,
+      "loss": 0.6355,
+      "step": 5652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320628542703711e-06,
+      "loss": 0.2378,
+      "step": 5653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31774723871385e-06,
+      "loss": 0.7088,
+      "step": 5654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3148674900758083e-06,
+      "loss": 0.6903,
+      "step": 5655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31198929737262e-06,
+      "loss": 0.7229,
+      "step": 5656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.309112661187005e-06,
+      "loss": 0.704,
+      "step": 5657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306237582101366e-06,
+      "loss": 0.6509,
+      "step": 5658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.303364060697797e-06,
+      "loss": 0.6914,
+      "step": 5659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3004920975580713e-06,
+      "loss": 0.703,
+      "step": 5660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2976216932636463e-06,
+      "loss": 0.6854,
+      "step": 5661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294752848395665e-06,
+      "loss": 0.704,
+      "step": 5662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2918855635349524e-06,
+      "loss": 0.6923,
+      "step": 5663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.289019839262029e-06,
+      "loss": 0.7364,
+      "step": 5664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286155676157078e-06,
+      "loss": 0.665,
+      "step": 5665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283293074799988e-06,
+      "loss": 0.6893,
+      "step": 5666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280432035770318e-06,
+      "loss": 0.6764,
+      "step": 5667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2775725596473217e-06,
+      "loss": 0.685,
+      "step": 5668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2747146470099224e-06,
+      "loss": 0.7742,
+      "step": 5669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271858298436733e-06,
+      "loss": 0.6851,
+      "step": 5670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2690035145060573e-06,
+      "loss": 0.7502,
+      "step": 5671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2661502957958737e-06,
+      "loss": 0.6623,
+      "step": 5672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2632986428838454e-06,
+      "loss": 0.746,
+      "step": 5673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2604485563473156e-06,
+      "loss": 0.6892,
+      "step": 5674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.257600036763321e-06,
+      "loss": 0.6699,
+      "step": 5675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25475308470857e-06,
+      "loss": 0.7532,
+      "step": 5676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251907700759457e-06,
+      "loss": 0.6714,
+      "step": 5677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249063885492062e-06,
+      "loss": 0.6219,
+      "step": 5678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246221639482139e-06,
+      "loss": 0.7188,
+      "step": 5679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2433809633051386e-06,
+      "loss": 0.6835,
+      "step": 5680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2405418575361804e-06,
+      "loss": 0.7745,
+      "step": 5681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2377043227500727e-06,
+      "loss": 0.6859,
+      "step": 5682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234868359521302e-06,
+      "loss": 0.7127,
+      "step": 5683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2320339684240357e-06,
+      "loss": 0.7224,
+      "step": 5684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2292011500321363e-06,
+      "loss": 0.7683,
+      "step": 5685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2263699049191236e-06,
+      "loss": 0.7183,
+      "step": 5686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2235402336582222e-06,
+      "loss": 0.7297,
+      "step": 5687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2207121368223226e-06,
+      "loss": 0.6979,
+      "step": 5688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217885614984011e-06,
+      "loss": 0.7009,
+      "step": 5689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2150606687155385e-06,
+      "loss": 0.6493,
+      "step": 5690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212237298588843e-06,
+      "loss": 0.6356,
+      "step": 5691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2094155051755507e-06,
+      "loss": 0.6931,
+      "step": 5692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206595289046961e-06,
+      "loss": 0.7822,
+      "step": 5693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2037766507740566e-06,
+      "loss": 0.6828,
+      "step": 5694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.200959590927497e-06,
+      "loss": 0.7065,
+      "step": 5695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198144110077629e-06,
+      "loss": 0.6625,
+      "step": 5696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195330208794476e-06,
+      "loss": 0.6767,
+      "step": 5697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192517887647739e-06,
+      "loss": 0.7275,
+      "step": 5698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1897071472068033e-06,
+      "loss": 0.2582,
+      "step": 5699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186897988040728e-06,
+      "loss": 0.6854,
+      "step": 5700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1840904107182637e-06,
+      "loss": 0.6744,
+      "step": 5701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1812844158078307e-06,
+      "loss": 0.76,
+      "step": 5702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17848000387753e-06,
+      "loss": 0.6558,
+      "step": 5703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1756771754951467e-06,
+      "loss": 0.7325,
+      "step": 5704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1728759312281378e-06,
+      "loss": 0.6742,
+      "step": 5705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170076271643652e-06,
+      "loss": 0.7123,
+      "step": 5706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1672781973085012e-06,
+      "loss": 0.6398,
+      "step": 5707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1644817087891903e-06,
+      "loss": 0.6304,
+      "step": 5708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161686806651895e-06,
+      "loss": 0.7489,
+      "step": 5709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1588934914624728e-06,
+      "loss": 0.733,
+      "step": 5710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156101763786459e-06,
+      "loss": 0.6869,
+      "step": 5711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1533116241890662e-06,
+      "loss": 0.6917,
+      "step": 5712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150523073235191e-06,
+      "loss": 0.6859,
+      "step": 5713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1477361114894034e-06,
+      "loss": 0.7411,
+      "step": 5714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144950739515951e-06,
+      "loss": 0.7287,
+      "step": 5715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142166957878763e-06,
+      "loss": 0.6978,
+      "step": 5716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1393847671414402e-06,
+      "loss": 0.6758,
+      "step": 5717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136604167867273e-06,
+      "loss": 0.8019,
+      "step": 5718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1338251606192205e-06,
+      "loss": 0.7192,
+      "step": 5719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13104774595992e-06,
+      "loss": 0.6119,
+      "step": 5720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128271924451685e-06,
+      "loss": 0.7327,
+      "step": 5721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1254976966565157e-06,
+      "loss": 0.7505,
+      "step": 5722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.122725063136081e-06,
+      "loss": 0.2359,
+      "step": 5723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119954024451728e-06,
+      "loss": 0.7365,
+      "step": 5724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1171845811644833e-06,
+      "loss": 0.7209,
+      "step": 5725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1144167338350462e-06,
+      "loss": 0.2249,
+      "step": 5726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1116504830238017e-06,
+      "loss": 0.7257,
+      "step": 5727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.108885829290804e-06,
+      "loss": 0.7828,
+      "step": 5728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1061227731957844e-06,
+      "loss": 0.7511,
+      "step": 5729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1033613152981548e-06,
+      "loss": 0.6581,
+      "step": 5730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1006014561569955e-06,
+      "loss": 0.7084,
+      "step": 5731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0978431963310773e-06,
+      "loss": 0.7056,
+      "step": 5732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0950865363788286e-06,
+      "loss": 0.8311,
+      "step": 5733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0923314768583723e-06,
+      "loss": 0.6596,
+      "step": 5734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0895780183274925e-06,
+      "loss": 0.6698,
+      "step": 5735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0868261613436646e-06,
+      "loss": 0.7648,
+      "step": 5736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0840759064640213e-06,
+      "loss": 0.6762,
+      "step": 5737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0813272542453798e-06,
+      "loss": 0.6916,
+      "step": 5738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0785802052442407e-06,
+      "loss": 0.7174,
+      "step": 5739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0758347600167683e-06,
+      "loss": 0.2262,
+      "step": 5740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0730909191188077e-06,
+      "loss": 0.6462,
+      "step": 5741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070348683105876e-06,
+      "loss": 0.7148,
+      "step": 5742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0676080525331677e-06,
+      "loss": 0.7084,
+      "step": 5743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064869027955554e-06,
+      "loss": 0.6887,
+      "step": 5744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.062131609927578e-06,
+      "loss": 0.2408,
+      "step": 5745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0593957990034574e-06,
+      "loss": 0.7268,
+      "step": 5746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056661595737084e-06,
+      "loss": 0.7252,
+      "step": 5747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.053929000682031e-06,
+      "loss": 0.6747,
+      "step": 5748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05119801439154e-06,
+      "loss": 0.6489,
+      "step": 5749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0484686374185194e-06,
+      "loss": 0.7322,
+      "step": 5750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045740870315568e-06,
+      "loss": 0.7154,
+      "step": 5751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0430147136349453e-06,
+      "loss": 0.72,
+      "step": 5752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040290167928598e-06,
+      "loss": 0.757,
+      "step": 5753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0375672337481277e-06,
+      "loss": 0.6904,
+      "step": 5754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0348459116448303e-06,
+      "loss": 0.6704,
+      "step": 5755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0321262021696607e-06,
+      "loss": 0.6619,
+      "step": 5756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029408105873254e-06,
+      "loss": 0.6402,
+      "step": 5757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0266916233059163e-06,
+      "loss": 0.7402,
+      "step": 5758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0239767550176247e-06,
+      "loss": 0.6513,
+      "step": 5759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0212635015580394e-06,
+      "loss": 0.7015,
+      "step": 5760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0185518634764823e-06,
+      "loss": 0.6688,
+      "step": 5761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015841841321953e-06,
+      "loss": 0.7317,
+      "step": 5762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0131334356431243e-06,
+      "loss": 0.6392,
+      "step": 5763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0104266469883373e-06,
+      "loss": 0.7209,
+      "step": 5764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0077214759056162e-06,
+      "loss": 0.713,
+      "step": 5765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005017922942647e-06,
+      "loss": 0.7515,
+      "step": 5766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0023159886467934e-06,
+      "loss": 0.7226,
+      "step": 5767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999615673565086e-06,
+      "loss": 0.6486,
+      "step": 5768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996916978244239e-06,
+      "loss": 0.7808,
+      "step": 5769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9942199032306253e-06,
+      "loss": 0.5691,
+      "step": 5770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991524449070299e-06,
+      "loss": 0.7119,
+      "step": 5771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98883061630898e-06,
+      "loss": 0.2368,
+      "step": 5772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986138405492063e-06,
+      "loss": 0.6262,
+      "step": 5773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98344781716462e-06,
+      "loss": 0.6963,
+      "step": 5774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980758851871377e-06,
+      "loss": 0.6452,
+      "step": 5775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978071510156754e-06,
+      "loss": 0.6469,
+      "step": 5776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9753857925648256e-06,
+      "loss": 0.6516,
+      "step": 5777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9727016996393457e-06,
+      "loss": 0.6962,
+      "step": 5778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9700192319237354e-06,
+      "loss": 0.7865,
+      "step": 5779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9673383899610843e-06,
+      "loss": 0.6683,
+      "step": 5780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9646591742941647e-06,
+      "loss": 0.6827,
+      "step": 5781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961981585465407e-06,
+      "loss": 0.6883,
+      "step": 5782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9593056240169175e-06,
+      "loss": 0.7114,
+      "step": 5783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9566312904904717e-06,
+      "loss": 0.6922,
+      "step": 5784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9539585854275143e-06,
+      "loss": 0.766,
+      "step": 5785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9512875093691685e-06,
+      "loss": 0.7008,
+      "step": 5786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948618062856217e-06,
+      "loss": 0.6906,
+      "step": 5787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945950246429117e-06,
+      "loss": 0.6437,
+      "step": 5788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9432840606279956e-06,
+      "loss": 0.7178,
+      "step": 5789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940619505992649e-06,
+      "loss": 0.6369,
+      "step": 5790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9379565830625467e-06,
+      "loss": 0.7668,
+      "step": 5791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9352952923768244e-06,
+      "loss": 0.6443,
+      "step": 5792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932635634474287e-06,
+      "loss": 0.6979,
+      "step": 5793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9299776098934077e-06,
+      "loss": 0.6798,
+      "step": 5794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9273212191723366e-06,
+      "loss": 0.6985,
+      "step": 5795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9246664628488863e-06,
+      "loss": 0.7647,
+      "step": 5796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9220133414605335e-06,
+      "loss": 0.703,
+      "step": 5797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919361855544437e-06,
+      "loss": 0.2244,
+      "step": 5798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9167120056374135e-06,
+      "loss": 0.7335,
+      "step": 5799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9140637922759586e-06,
+      "loss": 0.6992,
+      "step": 5800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911417215996222e-06,
+      "loss": 0.6968,
+      "step": 5801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087722773340367e-06,
+      "loss": 0.676,
+      "step": 5802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906128976824897e-06,
+      "loss": 0.712,
+      "step": 5803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9034873150039668e-06,
+      "loss": 0.7078,
+      "step": 5804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9008472924060751e-06,
+      "loss": 0.6522,
+      "step": 5805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898208909565722e-06,
+      "loss": 0.686,
+      "step": 5806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.89557216701708e-06,
+      "loss": 0.6929,
+      "step": 5807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8929370652939827e-06,
+      "loss": 0.6344,
+      "step": 5808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8903036049299328e-06,
+      "loss": 0.7227,
+      "step": 5809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8876717864581018e-06,
+      "loss": 0.5824,
+      "step": 5810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8850416104113278e-06,
+      "loss": 0.7706,
+      "step": 5811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8824130773221195e-06,
+      "loss": 0.7287,
+      "step": 5812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8797861877226508e-06,
+      "loss": 0.7314,
+      "step": 5813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8771609421447624e-06,
+      "loss": 0.2271,
+      "step": 5814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8745373411199609e-06,
+      "loss": 0.7613,
+      "step": 5815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8719153851794192e-06,
+      "loss": 0.7285,
+      "step": 5816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8692950748539896e-06,
+      "loss": 0.6964,
+      "step": 5817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8666764106741686e-06,
+      "loss": 0.6206,
+      "step": 5818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.86405939317014e-06,
+      "loss": 0.7138,
+      "step": 5819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8614440228717412e-06,
+      "loss": 0.7148,
+      "step": 5820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8588303003084885e-06,
+      "loss": 0.7062,
+      "step": 5821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8562182260095507e-06,
+      "loss": 0.6933,
+      "step": 5822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8536078005037682e-06,
+      "loss": 0.2518,
+      "step": 5823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8509990243196541e-06,
+      "loss": 0.7873,
+      "step": 5824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8483918979853798e-06,
+      "loss": 0.6614,
+      "step": 5825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8457864220287846e-06,
+      "loss": 0.6713,
+      "step": 5826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843182596977372e-06,
+      "loss": 0.7764,
+      "step": 5827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8405804233583191e-06,
+      "loss": 0.6026,
+      "step": 5828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837979901698459e-06,
+      "loss": 0.7528,
+      "step": 5829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.835381032524296e-06,
+      "loss": 0.2478,
+      "step": 5830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8327838163619971e-06,
+      "loss": 0.7336,
+      "step": 5831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8301882537373927e-06,
+      "loss": 0.7756,
+      "step": 5832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827594345175988e-06,
+      "loss": 0.6866,
+      "step": 5833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.825002091202942e-06,
+      "loss": 0.7433,
+      "step": 5834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8224114923430857e-06,
+      "loss": 0.7222,
+      "step": 5835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198225491209109e-06,
+      "loss": 0.6975,
+      "step": 5836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8172352620605749e-06,
+      "loss": 0.6969,
+      "step": 5837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8146496316859052e-06,
+      "loss": 0.6462,
+      "step": 5838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8120656585203877e-06,
+      "loss": 0.6647,
+      "step": 5839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8094833430871729e-06,
+      "loss": 0.2266,
+      "step": 5840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8069026859090765e-06,
+      "loss": 0.6628,
+      "step": 5841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8043236875085834e-06,
+      "loss": 0.6944,
+      "step": 5842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801746348407839e-06,
+      "loss": 0.6542,
+      "step": 5843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799170669128646e-06,
+      "loss": 0.6614,
+      "step": 5844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7965966501924825e-06,
+      "loss": 0.7406,
+      "step": 5845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7940242921204842e-06,
+      "loss": 0.7943,
+      "step": 5846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7914535954334511e-06,
+      "loss": 0.6892,
+      "step": 5847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7888845606518457e-06,
+      "loss": 0.6646,
+      "step": 5848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7863171882958009e-06,
+      "loss": 0.6479,
+      "step": 5849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783751478885105e-06,
+      "loss": 0.7039,
+      "step": 5850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.781187432939212e-06,
+      "loss": 0.7231,
+      "step": 5851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7786250509772408e-06,
+      "loss": 0.7436,
+      "step": 5852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7760643335179684e-06,
+      "loss": 0.2384,
+      "step": 5853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735052810798437e-06,
+      "loss": 0.7174,
+      "step": 5854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7709478941809721e-06,
+      "loss": 0.6744,
+      "step": 5855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768392173339123e-06,
+      "loss": 0.6505,
+      "step": 5856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7658381190717278e-06,
+      "loss": 0.6985,
+      "step": 5857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7632857318958785e-06,
+      "loss": 0.6154,
+      "step": 5858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7607350123283384e-06,
+      "loss": 0.2175,
+      "step": 5859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7581859608855245e-06,
+      "loss": 0.7635,
+      "step": 5860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7556385780835172e-06,
+      "loss": 0.6531,
+      "step": 5861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.753092864438063e-06,
+      "loss": 0.6571,
+      "step": 5862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7505488204645648e-06,
+      "loss": 0.7149,
+      "step": 5863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7480064466780966e-06,
+      "loss": 0.7113,
+      "step": 5864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7454657435933798e-06,
+      "loss": 0.7149,
+      "step": 5865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7429267117248138e-06,
+      "loss": 0.7086,
+      "step": 5866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7403893515864468e-06,
+      "loss": 0.6713,
+      "step": 5867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7378536636920008e-06,
+      "loss": 0.7695,
+      "step": 5868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7353196485548441e-06,
+      "loss": 0.6679,
+      "step": 5869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7327873066880163e-06,
+      "loss": 0.6617,
+      "step": 5870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7302566386042185e-06,
+      "loss": 0.6979,
+      "step": 5871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.727727644815811e-06,
+      "loss": 0.6653,
+      "step": 5872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7252003258348137e-06,
+      "loss": 0.2631,
+      "step": 5873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7226746821729056e-06,
+      "loss": 0.6708,
+      "step": 5874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7201507143414341e-06,
+      "loss": 0.5876,
+      "step": 5875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7176284228514018e-06,
+      "loss": 0.6239,
+      "step": 5876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.715107808213472e-06,
+      "loss": 0.7321,
+      "step": 5877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.712588870937969e-06,
+      "loss": 0.6867,
+      "step": 5878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7100716115348759e-06,
+      "loss": 0.706,
+      "step": 5879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075560305138417e-06,
+      "loss": 0.7642,
+      "step": 5880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705042128384171e-06,
+      "loss": 0.7351,
+      "step": 5881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025299056548284e-06,
+      "loss": 0.7414,
+      "step": 5882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7000193628344407e-06,
+      "loss": 0.7184,
+      "step": 5883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6975105004312887e-06,
+      "loss": 0.7008,
+      "step": 5884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6950033189533278e-06,
+      "loss": 0.2265,
+      "step": 5885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6924978189081498e-06,
+      "loss": 0.7038,
+      "step": 5886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6899940008030291e-06,
+      "loss": 0.6261,
+      "step": 5887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6874918651448845e-06,
+      "loss": 0.735,
+      "step": 5888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6849914124403055e-06,
+      "loss": 0.7087,
+      "step": 5889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6824926431955292e-06,
+      "loss": 0.7199,
+      "step": 5890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.679995557916455e-06,
+      "loss": 0.6771,
+      "step": 5891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6775001571086503e-06,
+      "loss": 0.6852,
+      "step": 5892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6750064412773315e-06,
+      "loss": 0.7148,
+      "step": 5893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6725144109273783e-06,
+      "loss": 0.6767,
+      "step": 5894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6700240665633283e-06,
+      "loss": 0.712,
+      "step": 5895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6675354086893737e-06,
+      "loss": 0.6265,
+      "step": 5896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6650484378093745e-06,
+      "loss": 0.6525,
+      "step": 5897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6625631544268416e-06,
+      "loss": 0.7064,
+      "step": 5898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660079559044946e-06,
+      "loss": 0.7382,
+      "step": 5899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.657597652166516e-06,
+      "loss": 0.6403,
+      "step": 5900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6551174342940424e-06,
+      "loss": 0.6952,
+      "step": 5901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.652638905929671e-06,
+      "loss": 0.6995,
+      "step": 5902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.650162067575204e-06,
+      "loss": 0.2204,
+      "step": 5903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6476869197321033e-06,
+      "loss": 0.6108,
+      "step": 5904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645213462901486e-06,
+      "loss": 0.237,
+      "step": 5905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6427416975841348e-06,
+      "loss": 0.75,
+      "step": 5906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6402716242804795e-06,
+      "loss": 0.6835,
+      "step": 5907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6378032434906144e-06,
+      "loss": 0.751,
+      "step": 5908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6353365557142876e-06,
+      "loss": 0.6458,
+      "step": 5909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.632871561450904e-06,
+      "loss": 0.6881,
+      "step": 5910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6304082611995342e-06,
+      "loss": 0.6606,
+      "step": 5911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6279466554588886e-06,
+      "loss": 0.7379,
+      "step": 5912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6254867447273526e-06,
+      "loss": 0.6975,
+      "step": 5913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6230285295029557e-06,
+      "loss": 0.6768,
+      "step": 5914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6205720102833976e-06,
+      "loss": 0.6439,
+      "step": 5915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6181171875660162e-06,
+      "loss": 0.7395,
+      "step": 5916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6156640618478192e-06,
+      "loss": 0.6756,
+      "step": 5917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6132126336254695e-06,
+      "loss": 0.7096,
+      "step": 5918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610762903395282e-06,
+      "loss": 0.6453,
+      "step": 5919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6083148716532316e-06,
+      "loss": 0.698,
+      "step": 5920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6058685388949436e-06,
+      "loss": 0.7287,
+      "step": 5921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6034239056157097e-06,
+      "loss": 0.7192,
+      "step": 5922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6009809723104685e-06,
+      "loss": 0.7697,
+      "step": 5923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5985397394738168e-06,
+      "loss": 0.7069,
+      "step": 5924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5961002076000077e-06,
+      "loss": 0.7014,
+      "step": 5925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5936623771829473e-06,
+      "loss": 0.7014,
+      "step": 5926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5912262487162045e-06,
+      "loss": 0.6312,
+      "step": 5927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5887918226929989e-06,
+      "loss": 0.7069,
+      "step": 5928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5863590996061983e-06,
+      "loss": 0.2287,
+      "step": 5929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5839280799483392e-06,
+      "loss": 0.6843,
+      "step": 5930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.581498764211603e-06,
+      "loss": 0.7467,
+      "step": 5931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5790711528878356e-06,
+      "loss": 0.755,
+      "step": 5932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5766452464685245e-06,
+      "loss": 0.7212,
+      "step": 5933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5742210454448258e-06,
+      "loss": 0.6935,
+      "step": 5934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5717985503075407e-06,
+      "loss": 0.6774,
+      "step": 5935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5693777615471306e-06,
+      "loss": 0.6524,
+      "step": 5936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5669586796537084e-06,
+      "loss": 0.7797,
+      "step": 5937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5645413051170388e-06,
+      "loss": 0.6445,
+      "step": 5938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5621256384265504e-06,
+      "loss": 0.6527,
+      "step": 5939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5597116800713186e-06,
+      "loss": 0.7085,
+      "step": 5940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5572994305400713e-06,
+      "loss": 0.7473,
+      "step": 5941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5548888903211967e-06,
+      "loss": 0.7663,
+      "step": 5942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.552480059902729e-06,
+      "loss": 0.7029,
+      "step": 5943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5500729397723668e-06,
+      "loss": 0.2271,
+      "step": 5944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5476675304174549e-06,
+      "loss": 0.7164,
+      "step": 5945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5452638323249912e-06,
+      "loss": 0.6914,
+      "step": 5946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.542861845981628e-06,
+      "loss": 0.7498,
+      "step": 5947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5404615718736782e-06,
+      "loss": 0.6282,
+      "step": 5948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538063010487101e-06,
+      "loss": 0.7587,
+      "step": 5949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535666162307503e-06,
+      "loss": 0.7282,
+      "step": 5950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533271027820159e-06,
+      "loss": 0.6731,
+      "step": 5951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5308776075099829e-06,
+      "loss": 0.6863,
+      "step": 5952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528485901861555e-06,
+      "loss": 0.7849,
+      "step": 5953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5260959113590924e-06,
+      "loss": 0.7572,
+      "step": 5954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5237076364864788e-06,
+      "loss": 0.7348,
+      "step": 5955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5213210777272448e-06,
+      "loss": 0.7115,
+      "step": 5956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.518936235564573e-06,
+      "loss": 0.6664,
+      "step": 5957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5165531104812991e-06,
+      "loss": 0.7056,
+      "step": 5958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.514171702959909e-06,
+      "loss": 0.6578,
+      "step": 5959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5117920134825492e-06,
+      "loss": 0.6905,
+      "step": 5960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.50941404253101e-06,
+      "loss": 0.7402,
+      "step": 5961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5070377905867361e-06,
+      "loss": 0.7721,
+      "step": 5962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5046632581308241e-06,
+      "loss": 0.6727,
+      "step": 5963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.50229044564402e-06,
+      "loss": 0.6914,
+      "step": 5964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499919353606729e-06,
+      "loss": 0.714,
+      "step": 5965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4975499824990025e-06,
+      "loss": 0.7427,
+      "step": 5966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4951823328005433e-06,
+      "loss": 0.7274,
+      "step": 5967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4928164049907068e-06,
+      "loss": 0.725,
+      "step": 5968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4904521995484967e-06,
+      "loss": 0.2031,
+      "step": 5969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4880897169525753e-06,
+      "loss": 0.7412,
+      "step": 5970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4857289576812505e-06,
+      "loss": 0.7622,
+      "step": 5971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.483369922212483e-06,
+      "loss": 0.6551,
+      "step": 5972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4810126110238798e-06,
+      "loss": 0.6686,
+      "step": 5973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4786570245927101e-06,
+      "loss": 0.6931,
+      "step": 5974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4763031633958847e-06,
+      "loss": 0.6258,
+      "step": 5975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.47395102790996e-06,
+      "loss": 0.7071,
+      "step": 5976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4716006186111598e-06,
+      "loss": 0.6768,
+      "step": 5977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469251935975341e-06,
+      "loss": 0.7505,
+      "step": 5978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.466904980478029e-06,
+      "loss": 0.6946,
+      "step": 5979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4645597525943779e-06,
+      "loss": 0.6489,
+      "step": 5980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4622162527992101e-06,
+      "loss": 0.6888,
+      "step": 5981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4598744815669908e-06,
+      "loss": 0.7218,
+      "step": 5982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4575344393718338e-06,
+      "loss": 0.6981,
+      "step": 5983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455196126687507e-06,
+      "loss": 0.6699,
+      "step": 5984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4528595439874238e-06,
+      "loss": 0.7063,
+      "step": 5985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4505246917446525e-06,
+      "loss": 0.7754,
+      "step": 5986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448191570431906e-06,
+      "loss": 0.7878,
+      "step": 5987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445860180521551e-06,
+      "loss": 0.6747,
+      "step": 5988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435305224856e-06,
+      "loss": 0.7503,
+      "step": 5989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4412025967957133e-06,
+      "loss": 0.6352,
+      "step": 5990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4388764039232106e-06,
+      "loss": 0.7431,
+      "step": 5991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4365519443390496e-06,
+      "loss": 0.6898,
+      "step": 5992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4342292185138418e-06,
+      "loss": 0.7622,
+      "step": 5993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4319082269178453e-06,
+      "loss": 0.7304,
+      "step": 5994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.429588970020973e-06,
+      "loss": 0.7223,
+      "step": 5995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4272714482927819e-06,
+      "loss": 0.6448,
+      "step": 5996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4249556622024741e-06,
+      "loss": 0.6966,
+      "step": 5997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.422641612218908e-06,
+      "loss": 0.7133,
+      "step": 5998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4203292988105844e-06,
+      "loss": 0.6344,
+      "step": 5999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4180187224456621e-06,
+      "loss": 0.7365,
+      "step": 6000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4157098835919314e-06,
+      "loss": 0.7281,
+      "step": 6001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4134027827168472e-06,
+      "loss": 0.7568,
+      "step": 6002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4110974202875049e-06,
+      "loss": 0.7432,
+      "step": 6003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.408793796770649e-06,
+      "loss": 0.7133,
+      "step": 6004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4064919126326715e-06,
+      "loss": 0.7275,
+      "step": 6005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40419176833961e-06,
+      "loss": 0.6966,
+      "step": 6006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4018933643571575e-06,
+      "loss": 0.7354,
+      "step": 6007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3995967011506473e-06,
+      "loss": 0.6432,
+      "step": 6008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3973017791850641e-06,
+      "loss": 0.7333,
+      "step": 6009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395008598925036e-06,
+      "loss": 0.7571,
+      "step": 6010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3927171608348423e-06,
+      "loss": 0.7296,
+      "step": 6011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.39042746537841e-06,
+      "loss": 0.6782,
+      "step": 6012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3881395130193108e-06,
+      "loss": 0.6889,
+      "step": 6013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3858533042207634e-06,
+      "loss": 0.6907,
+      "step": 6014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.383568839445636e-06,
+      "loss": 0.2277,
+      "step": 6015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3812861191564397e-06,
+      "loss": 0.645,
+      "step": 6016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3790051438153408e-06,
+      "loss": 0.6567,
+      "step": 6017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3767259138841372e-06,
+      "loss": 0.671,
+      "step": 6018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3744484298242911e-06,
+      "loss": 0.6308,
+      "step": 6019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372172692096896e-06,
+      "loss": 0.5906,
+      "step": 6020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3698987011627075e-06,
+      "loss": 0.6989,
+      "step": 6021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3676264574821108e-06,
+      "loss": 0.2628,
+      "step": 6022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3653559615151456e-06,
+      "loss": 0.6892,
+      "step": 6023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3630872137215012e-06,
+      "loss": 0.683,
+      "step": 6024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3608202145605075e-06,
+      "loss": 0.7005,
+      "step": 6025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3585549644911422e-06,
+      "loss": 0.6443,
+      "step": 6026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.356291463972027e-06,
+      "loss": 0.8398,
+      "step": 6027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3540297134614332e-06,
+      "loss": 0.7308,
+      "step": 6028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3517697134172758e-06,
+      "loss": 0.7106,
+      "step": 6029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3495114642971142e-06,
+      "loss": 0.6934,
+      "step": 6030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3472549665581546e-06,
+      "loss": 0.6926,
+      "step": 6031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3450002206572465e-06,
+      "loss": 0.7091,
+      "step": 6032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3427472270508912e-06,
+      "loss": 0.7277,
+      "step": 6033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.340495986195227e-06,
+      "loss": 0.7736,
+      "step": 6034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.338246498546042e-06,
+      "loss": 0.7531,
+      "step": 6035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.335998764558768e-06,
+      "loss": 0.6752,
+      "step": 6036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3337527846884812e-06,
+      "loss": 0.6977,
+      "step": 6037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3315085593899057e-06,
+      "loss": 0.7023,
+      "step": 6038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3292660891174081e-06,
+      "loss": 0.6907,
+      "step": 6039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3270253743249984e-06,
+      "loss": 0.6558,
+      "step": 6040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3247864154663337e-06,
+      "loss": 0.6757,
+      "step": 6041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3225492129947104e-06,
+      "loss": 0.6602,
+      "step": 6042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3203137673630827e-06,
+      "loss": 0.7047,
+      "step": 6043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3180800790240288e-06,
+      "loss": 0.699,
+      "step": 6044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.315848148429789e-06,
+      "loss": 0.717,
+      "step": 6045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3136179760322366e-06,
+      "loss": 0.6871,
+      "step": 6046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3113895622828998e-06,
+      "loss": 0.7229,
+      "step": 6047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309162907632938e-06,
+      "loss": 0.6771,
+      "step": 6048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306938012533161e-06,
+      "loss": 0.733,
+      "step": 6049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3047148774340246e-06,
+      "loss": 0.6875,
+      "step": 6050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3024935027856245e-06,
+      "loss": 0.6653,
+      "step": 6051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3002738890376999e-06,
+      "loss": 0.6756,
+      "step": 6052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2980560366396345e-06,
+      "loss": 0.7325,
+      "step": 6053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958399460404581e-06,
+      "loss": 0.7528,
+      "step": 6054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2936256176888394e-06,
+      "loss": 0.7435,
+      "step": 6055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2914130520330937e-06,
+      "loss": 0.7143,
+      "step": 6056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2892022495211765e-06,
+      "loss": 0.6511,
+      "step": 6057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2869932106006845e-06,
+      "loss": 0.7303,
+      "step": 6058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2847859357188674e-06,
+      "loss": 0.751,
+      "step": 6059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2825804253226059e-06,
+      "loss": 0.67,
+      "step": 6060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2803766798584315e-06,
+      "loss": 0.7111,
+      "step": 6061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.278174699772513e-06,
+      "loss": 0.6861,
+      "step": 6062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2759744855106637e-06,
+      "loss": 0.6476,
+      "step": 6063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273776037518345e-06,
+      "loss": 0.7107,
+      "step": 6064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2715793562406475e-06,
+      "loss": 0.7011,
+      "step": 6065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2693844421223179e-06,
+      "loss": 0.714,
+      "step": 6066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267191295607736e-06,
+      "loss": 0.6495,
+      "step": 6067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2649999171409333e-06,
+      "loss": 0.7159,
+      "step": 6068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2628103071655706e-06,
+      "loss": 0.7183,
+      "step": 6069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2606224661249566e-06,
+      "loss": 0.6376,
+      "step": 6070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258436394462047e-06,
+      "loss": 0.7779,
+      "step": 6071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2562520926194333e-06,
+      "loss": 0.6942,
+      "step": 6072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2540695610393493e-06,
+      "loss": 0.7186,
+      "step": 6073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2518888001636686e-06,
+      "loss": 0.7305,
+      "step": 6074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2497098104339145e-06,
+      "loss": 0.6462,
+      "step": 6075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247532592291243e-06,
+      "loss": 0.6621,
+      "step": 6076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2453571461764547e-06,
+      "loss": 0.6396,
+      "step": 6077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2431834725299919e-06,
+      "loss": 0.6368,
+      "step": 6078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2410115717919335e-06,
+      "loss": 0.2153,
+      "step": 6079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2388414444020091e-06,
+      "loss": 0.6343,
+      "step": 6080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366730907995838e-06,
+      "loss": 0.7045,
+      "step": 6081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2345065114236564e-06,
+      "loss": 0.7389,
+      "step": 6082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2323417067128796e-06,
+      "loss": 0.762,
+      "step": 6083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2301786771055358e-06,
+      "loss": 0.6785,
+      "step": 6084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2280174230395604e-06,
+      "loss": 0.6728,
+      "step": 6085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2258579449525122e-06,
+      "loss": 0.66,
+      "step": 6086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.223700243281607e-06,
+      "loss": 0.7512,
+      "step": 6087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.221544318463691e-06,
+      "loss": 0.2273,
+      "step": 6088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.219390170935254e-06,
+      "loss": 0.7638,
+      "step": 6089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.217237801132427e-06,
+      "loss": 0.6966,
+      "step": 6090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215087209490975e-06,
+      "loss": 0.75,
+      "step": 6091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2129383964463127e-06,
+      "loss": 0.718,
+      "step": 6092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2107913624334877e-06,
+      "loss": 0.7043,
+      "step": 6093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2086461078871904e-06,
+      "loss": 0.6436,
+      "step": 6094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2065026332417474e-06,
+      "loss": 0.6138,
+      "step": 6095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2043609389311283e-06,
+      "loss": 0.7076,
+      "step": 6096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2022210253889421e-06,
+      "loss": 0.7629,
+      "step": 6097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2000828930484376e-06,
+      "loss": 0.7323,
+      "step": 6098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1979465423424995e-06,
+      "loss": 0.6837,
+      "step": 6099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1958119737036544e-06,
+      "loss": 0.6468,
+      "step": 6100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1936791875640696e-06,
+      "loss": 0.6679,
+      "step": 6101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19154818435555e-06,
+      "loss": 0.75,
+      "step": 6102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1894189645095377e-06,
+      "loss": 0.7144,
+      "step": 6103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1872915284571151e-06,
+      "loss": 0.6623,
+      "step": 6104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185165876629003e-06,
+      "loss": 0.6756,
+      "step": 6105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183042009455565e-06,
+      "loss": 0.6716,
+      "step": 6106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1809199273667993e-06,
+      "loss": 0.6906,
+      "step": 6107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.178799630792341e-06,
+      "loss": 0.6754,
+      "step": 6108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176681120161468e-06,
+      "loss": 0.5813,
+      "step": 6109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1745643959030928e-06,
+      "loss": 0.7045,
+      "step": 6110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1724494584457723e-06,
+      "loss": 0.6986,
+      "step": 6111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1703363082176922e-06,
+      "loss": 0.6831,
+      "step": 6112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1682249456466866e-06,
+      "loss": 0.6949,
+      "step": 6113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.166115371160218e-06,
+      "loss": 0.7112,
+      "step": 6114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1640075851853982e-06,
+      "loss": 0.7452,
+      "step": 6115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1619015881489638e-06,
+      "loss": 0.6823,
+      "step": 6116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597973804772966e-06,
+      "loss": 0.6477,
+      "step": 6117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1576949625964195e-06,
+      "loss": 0.7114,
+      "step": 6118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1555943349319865e-06,
+      "loss": 0.705,
+      "step": 6119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1534954979092917e-06,
+      "loss": 0.6685,
+      "step": 6120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151398451953265e-06,
+      "loss": 0.7187,
+      "step": 6121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.149303197488475e-06,
+      "loss": 0.7597,
+      "step": 6122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.147209734939131e-06,
+      "loss": 0.7002,
+      "step": 6123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1451180647290738e-06,
+      "loss": 0.6837,
+      "step": 6124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143028187281784e-06,
+      "loss": 0.7066,
+      "step": 6125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1409401030203771e-06,
+      "loss": 0.6796,
+      "step": 6126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1388538123676108e-06,
+      "loss": 0.7286,
+      "step": 6127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1367693157458782e-06,
+      "loss": 0.7078,
+      "step": 6128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1346866135771995e-06,
+      "loss": 0.7259,
+      "step": 6129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1326057062832452e-06,
+      "loss": 0.6882,
+      "step": 6130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1305265942853139e-06,
+      "loss": 0.6838,
+      "step": 6131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1284492780043488e-06,
+      "loss": 0.7228,
+      "step": 6132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1263737578609157e-06,
+      "loss": 0.6727,
+      "step": 6133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124300034275232e-06,
+      "loss": 0.6603,
+      "step": 6134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1222281076671416e-06,
+      "loss": 0.8232,
+      "step": 6135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1201579784561288e-06,
+      "loss": 0.7108,
+      "step": 6136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1180896470613122e-06,
+      "loss": 0.7426,
+      "step": 6137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1160231139014443e-06,
+      "loss": 0.7169,
+      "step": 6138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1139583793949204e-06,
+      "loss": 0.7302,
+      "step": 6139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1118954439597662e-06,
+      "loss": 0.6584,
+      "step": 6140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1098343080136443e-06,
+      "loss": 0.6232,
+      "step": 6141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1077749719738517e-06,
+      "loss": 0.6514,
+      "step": 6142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.105717436257322e-06,
+      "loss": 0.2159,
+      "step": 6143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1036617012806284e-06,
+      "loss": 0.6919,
+      "step": 6144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1016077674599735e-06,
+      "loss": 0.6817,
+      "step": 6145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.099555635211197e-06,
+      "loss": 0.6908,
+      "step": 6146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097505304949773e-06,
+      "loss": 0.738,
+      "step": 6147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0954567770908154e-06,
+      "loss": 0.7371,
+      "step": 6148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093410052049071e-06,
+      "loss": 0.7891,
+      "step": 6149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0913651302389139e-06,
+      "loss": 0.6668,
+      "step": 6150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0893220120743654e-06,
+      "loss": 0.6854,
+      "step": 6151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0872806979690732e-06,
+      "loss": 0.6969,
+      "step": 6152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085241188336328e-06,
+      "loss": 0.6267,
+      "step": 6153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0832034835890404e-06,
+      "loss": 0.6864,
+      "step": 6154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0811675841397728e-06,
+      "loss": 0.7058,
+      "step": 6155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.079133490400711e-06,
+      "loss": 0.7496,
+      "step": 6156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.077101202783679e-06,
+      "loss": 0.6841,
+      "step": 6157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0750707217001344e-06,
+      "loss": 0.728,
+      "step": 6158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0730420475611668e-06,
+      "loss": 0.7342,
+      "step": 6159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.071015180777506e-06,
+      "loss": 0.7245,
+      "step": 6160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068990121759511e-06,
+      "loss": 0.6385,
+      "step": 6161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0669668709171766e-06,
+      "loss": 0.7174,
+      "step": 6162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064945428660129e-06,
+      "loss": 0.6929,
+      "step": 6163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0629257953976302e-06,
+      "loss": 0.7107,
+      "step": 6164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0609079715385794e-06,
+      "loss": 0.7072,
+      "step": 6165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0588919574915024e-06,
+      "loss": 0.6972,
+      "step": 6166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0568777536645648e-06,
+      "loss": 0.6663,
+      "step": 6167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0548653604655602e-06,
+      "loss": 0.6791,
+      "step": 6168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0528547783019194e-06,
+      "loss": 0.7181,
+      "step": 6169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.050846007580707e-06,
+      "loss": 0.6475,
+      "step": 6170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0488390487086187e-06,
+      "loss": 0.7143,
+      "step": 6171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0468339020919848e-06,
+      "loss": 0.7165,
+      "step": 6172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0448305681367654e-06,
+      "loss": 0.7287,
+      "step": 6173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042829047248559e-06,
+      "loss": 0.709,
+      "step": 6174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0408293398325975e-06,
+      "loss": 0.6775,
+      "step": 6175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0388314462937333e-06,
+      "loss": 0.6626,
+      "step": 6176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0368353670364674e-06,
+      "loss": 0.7167,
+      "step": 6177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348411024649242e-06,
+      "loss": 0.7862,
+      "step": 6178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0328486529828674e-06,
+      "loss": 0.6719,
+      "step": 6179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0308580189936822e-06,
+      "loss": 0.687,
+      "step": 6180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0288692009003987e-06,
+      "loss": 0.692,
+      "step": 6181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0268821991056722e-06,
+      "loss": 0.6169,
+      "step": 6182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.024897014011792e-06,
+      "loss": 0.2442,
+      "step": 6183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.022913646020678e-06,
+      "loss": 0.2119,
+      "step": 6184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.020932095533883e-06,
+      "loss": 0.7403,
+      "step": 6185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0189523629525966e-06,
+      "loss": 0.6639,
+      "step": 6186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0169744486776334e-06,
+      "loss": 0.7084,
+      "step": 6187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.014998353109443e-06,
+      "loss": 0.6236,
+      "step": 6188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013024076648106e-06,
+      "loss": 0.6775,
+      "step": 6189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011051619693334e-06,
+      "loss": 0.7673,
+      "step": 6190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0090809826444759e-06,
+      "loss": 0.7528,
+      "step": 6191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0071121659005034e-06,
+      "loss": 0.2229,
+      "step": 6192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0051451698600256e-06,
+      "loss": 0.7183,
+      "step": 6193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0031799949212805e-06,
+      "loss": 0.7092,
+      "step": 6194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0012166414821366e-06,
+      "loss": 0.721,
+      "step": 6195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992551099401015e-07,
+      "loss": 0.6464,
+      "step": 6196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.972954006922997e-07,
+      "loss": 0.7781,
+      "step": 6197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.953375141354994e-07,
+      "loss": 0.7401,
+      "step": 6198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.933814506660922e-07,
+      "loss": 0.7066,
+      "step": 6199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.914272106801092e-07,
+      "loss": 0.6885,
+      "step": 6200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.894747945732008e-07,
+      "loss": 0.7738,
+      "step": 6201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.875242027406528e-07,
+      "loss": 0.6964,
+      "step": 6202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.855754355773883e-07,
+      "loss": 0.6704,
+      "step": 6203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.836284934779526e-07,
+      "loss": 0.2163,
+      "step": 6204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.816833768365241e-07,
+      "loss": 0.7419,
+      "step": 6205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.797400860469109e-07,
+      "loss": 0.7381,
+      "step": 6206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.777986215025548e-07,
+      "loss": 0.6695,
+      "step": 6207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.758589835965249e-07,
+      "loss": 0.7626,
+      "step": 6208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739211727215203e-07,
+      "loss": 0.7063,
+      "step": 6209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.719851892698718e-07,
+      "loss": 0.637,
+      "step": 6210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700510336335355e-07,
+      "loss": 0.7381,
+      "step": 6211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.681187062041076e-07,
+      "loss": 0.707,
+      "step": 6212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.661882073728047e-07,
+      "loss": 0.5952,
+      "step": 6213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642595375304752e-07,
+      "loss": 0.6651,
+      "step": 6214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.623326970676006e-07,
+      "loss": 0.6739,
+      "step": 6215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.604076863742862e-07,
+      "loss": 0.7062,
+      "step": 6216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.584845058402758e-07,
+      "loss": 0.711,
+      "step": 6217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.565631558549315e-07,
+      "loss": 0.6707,
+      "step": 6218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.546436368072553e-07,
+      "loss": 0.7137,
+      "step": 6219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.527259490858698e-07,
+      "loss": 0.739,
+      "step": 6220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.508100930790365e-07,
+      "loss": 0.6606,
+      "step": 6221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.48896069174634e-07,
+      "loss": 0.7212,
+      "step": 6222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.46983877760178e-07,
+      "loss": 0.6464,
+      "step": 6223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.450735192228133e-07,
+      "loss": 0.7373,
+      "step": 6224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.431649939493115e-07,
+      "loss": 0.647,
+      "step": 6225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.412583023260735e-07,
+      "loss": 0.7426,
+      "step": 6226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.39353444739125e-07,
+      "loss": 0.7481,
+      "step": 6227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374504215741298e-07,
+      "loss": 0.6965,
+      "step": 6228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.355492332163718e-07,
+      "loss": 0.7387,
+      "step": 6229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336498800507676e-07,
+      "loss": 0.6854,
+      "step": 6230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.317523624618585e-07,
+      "loss": 0.7839,
+      "step": 6231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298566808338183e-07,
+      "loss": 0.6981,
+      "step": 6232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.279628355504478e-07,
+      "loss": 0.6496,
+      "step": 6233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.260708269951757e-07,
+      "loss": 0.5977,
+      "step": 6234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.241806555510579e-07,
+      "loss": 0.6846,
+      "step": 6235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.222923216007795e-07,
+      "loss": 0.6803,
+      "step": 6236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.204058255266513e-07,
+      "loss": 0.7238,
+      "step": 6237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.185211677106187e-07,
+      "loss": 0.6541,
+      "step": 6238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.166383485342478e-07,
+      "loss": 0.6245,
+      "step": 6239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.147573683787337e-07,
+      "loss": 0.7155,
+      "step": 6240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128782276249015e-07,
+      "loss": 0.7308,
+      "step": 6241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.110009266532016e-07,
+      "loss": 0.6999,
+      "step": 6242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.091254658437176e-07,
+      "loss": 0.2061,
+      "step": 6243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.072518455761481e-07,
+      "loss": 0.7601,
+      "step": 6244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.05380066229834e-07,
+      "loss": 0.716,
+      "step": 6245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.035101281837321e-07,
+      "loss": 0.6731,
+      "step": 6246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.016420318164354e-07,
+      "loss": 0.6437,
+      "step": 6247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.997757775061555e-07,
+      "loss": 0.7212,
+      "step": 6248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.979113656307337e-07,
+      "loss": 0.6941,
+      "step": 6249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.960487965676434e-07,
+      "loss": 0.6909,
+      "step": 6250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.941880706939798e-07,
+      "loss": 0.751,
+      "step": 6251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.923291883864648e-07,
+      "loss": 0.6426,
+      "step": 6252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.904721500214486e-07,
+      "loss": 0.6945,
+      "step": 6253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.886169559749103e-07,
+      "loss": 0.7376,
+      "step": 6254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.867636066224516e-07,
+      "loss": 0.7529,
+      "step": 6255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.849121023393026e-07,
+      "loss": 0.7206,
+      "step": 6256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.830624435003199e-07,
+      "loss": 0.2183,
+      "step": 6257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.812146304799829e-07,
+      "loss": 0.6983,
+      "step": 6258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.793686636524046e-07,
+      "loss": 0.6873,
+      "step": 6259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.775245433913193e-07,
+      "loss": 0.8077,
+      "step": 6260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.756822700700873e-07,
+      "loss": 0.6768,
+      "step": 6261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.738418440616969e-07,
+      "loss": 0.6525,
+      "step": 6262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.720032657387589e-07,
+      "loss": 0.6504,
+      "step": 6263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.701665354735178e-07,
+      "loss": 0.7585,
+      "step": 6264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.683316536378316e-07,
+      "loss": 0.2305,
+      "step": 6265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.664986206031977e-07,
+      "loss": 0.7047,
+      "step": 6266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.646674367407282e-07,
+      "loss": 0.6602,
+      "step": 6267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.628381024211696e-07,
+      "loss": 0.6186,
+      "step": 6268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.61010618014887e-07,
+      "loss": 0.7532,
+      "step": 6269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.591849838918709e-07,
+      "loss": 0.2437,
+      "step": 6270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.573612004217446e-07,
+      "loss": 0.6332,
+      "step": 6271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.555392679737495e-07,
+      "loss": 0.6453,
+      "step": 6272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.53719186916756e-07,
+      "loss": 0.6997,
+      "step": 6273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.519009576192583e-07,
+      "loss": 0.7118,
+      "step": 6274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.500845804493729e-07,
+      "loss": 0.6848,
+      "step": 6275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.48270055774848e-07,
+      "loss": 0.6831,
+      "step": 6276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.464573839630519e-07,
+      "loss": 0.2383,
+      "step": 6277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.446465653809777e-07,
+      "loss": 0.6849,
+      "step": 6278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.428376003952443e-07,
+      "loss": 0.7443,
+      "step": 6279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.410304893720978e-07,
+      "loss": 0.6542,
+      "step": 6280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.392252326774064e-07,
+      "loss": 0.6944,
+      "step": 6281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.374218306766579e-07,
+      "loss": 0.6811,
+      "step": 6282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.356202837349759e-07,
+      "loss": 0.6986,
+      "step": 6283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338205922170984e-07,
+      "loss": 0.7058,
+      "step": 6284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320227564873962e-07,
+      "loss": 0.6832,
+      "step": 6285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.302267769098537e-07,
+      "loss": 0.6903,
+      "step": 6286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.284326538480902e-07,
+      "loss": 0.7518,
+      "step": 6287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.266403876653428e-07,
+      "loss": 0.7044,
+      "step": 6288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.248499787244757e-07,
+      "loss": 0.7588,
+      "step": 6289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.230614273879733e-07,
+      "loss": 0.219,
+      "step": 6290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.212747340179473e-07,
+      "loss": 0.7334,
+      "step": 6291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194898989761335e-07,
+      "loss": 0.6794,
+      "step": 6292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.177069226238909e-07,
+      "loss": 0.712,
+      "step": 6293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.159258053221985e-07,
+      "loss": 0.7419,
+      "step": 6294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.141465474316656e-07,
+      "loss": 0.7508,
+      "step": 6295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.123691493125164e-07,
+      "loss": 0.6378,
+      "step": 6296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.105936113246093e-07,
+      "loss": 0.6669,
+      "step": 6297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.088199338274172e-07,
+      "loss": 0.8139,
+      "step": 6298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.070481171800415e-07,
+      "loss": 0.7425,
+      "step": 6299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.052781617412009e-07,
+      "loss": 0.6625,
+      "step": 6300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.035100678692453e-07,
+      "loss": 0.7059,
+      "step": 6301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.01743835922143e-07,
+      "loss": 0.7427,
+      "step": 6302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.999794662574845e-07,
+      "loss": 0.7423,
+      "step": 6303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.982169592324862e-07,
+      "loss": 0.7062,
+      "step": 6304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.964563152039839e-07,
+      "loss": 0.746,
+      "step": 6305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.946975345284402e-07,
+      "loss": 0.7231,
+      "step": 6306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.929406175619403e-07,
+      "loss": 0.6693,
+      "step": 6307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.911855646601852e-07,
+      "loss": 0.7599,
+      "step": 6308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.894323761785073e-07,
+      "loss": 0.6165,
+      "step": 6309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876810524718548e-07,
+      "loss": 0.5971,
+      "step": 6310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.859315938948065e-07,
+      "loss": 0.6712,
+      "step": 6311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.841840008015522e-07,
+      "loss": 0.7399,
+      "step": 6312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.824382735459157e-07,
+      "loss": 0.647,
+      "step": 6313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.806944124813354e-07,
+      "loss": 0.6726,
+      "step": 6314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.789524179608732e-07,
+      "loss": 0.6806,
+      "step": 6315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.772122903372148e-07,
+      "loss": 0.7755,
+      "step": 6316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.754740299626651e-07,
+      "loss": 0.694,
+      "step": 6317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737376371891581e-07,
+      "loss": 0.6433,
+      "step": 6318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.720031123682393e-07,
+      "loss": 0.693,
+      "step": 6319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.702704558510842e-07,
+      "loss": 0.7215,
+      "step": 6320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.685396679884872e-07,
+      "loss": 0.2342,
+      "step": 6321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668107491308596e-07,
+      "loss": 0.5947,
+      "step": 6322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.650836996282463e-07,
+      "loss": 0.771,
+      "step": 6323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.633585198303018e-07,
+      "loss": 0.694,
+      "step": 6324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.616352100863067e-07,
+      "loss": 0.618,
+      "step": 6325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.599137707451632e-07,
+      "loss": 0.656,
+      "step": 6326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.581942021553967e-07,
+      "loss": 0.6837,
+      "step": 6327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564765046651512e-07,
+      "loss": 0.7477,
+      "step": 6328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.547606786221895e-07,
+      "loss": 0.681,
+      "step": 6329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.530467243739004e-07,
+      "loss": 0.6142,
+      "step": 6330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.513346422672907e-07,
+      "loss": 0.2283,
+      "step": 6331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.496244326489943e-07,
+      "loss": 0.7313,
+      "step": 6332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47916095865252e-07,
+      "loss": 0.7084,
+      "step": 6333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.462096322619416e-07,
+      "loss": 0.6875,
+      "step": 6334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.445050421845512e-07,
+      "loss": 0.7122,
+      "step": 6335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.428023259781935e-07,
+      "loss": 0.6913,
+      "step": 6336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.411014839876008e-07,
+      "loss": 0.7093,
+      "step": 6337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.39402516557125e-07,
+      "loss": 0.7146,
+      "step": 6338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.377054240307425e-07,
+      "loss": 0.6964,
+      "step": 6339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.360102067520458e-07,
+      "loss": 0.6736,
+      "step": 6340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343168650642485e-07,
+      "loss": 0.6674,
+      "step": 6341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.326253993101873e-07,
+      "loss": 0.6816,
+      "step": 6342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.309358098323138e-07,
+      "loss": 0.7543,
+      "step": 6343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.292480969727056e-07,
+      "loss": 0.8008,
+      "step": 6344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.27562261073057e-07,
+      "loss": 0.7047,
+      "step": 6345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25878302474684e-07,
+      "loss": 0.6704,
+      "step": 6346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.241962215185194e-07,
+      "loss": 0.6771,
+      "step": 6347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.225160185451174e-07,
+      "loss": 0.7381,
+      "step": 6348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.20837693894657e-07,
+      "loss": 0.7161,
+      "step": 6349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.191612479069265e-07,
+      "loss": 0.7207,
+      "step": 6350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.174866809213443e-07,
+      "loss": 0.6818,
+      "step": 6351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.158139932769415e-07,
+      "loss": 0.7264,
+      "step": 6352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.141431853123748e-07,
+      "loss": 0.6896,
+      "step": 6353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.124742573659116e-07,
+      "loss": 0.7226,
+      "step": 6354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.108072097754448e-07,
+      "loss": 0.651,
+      "step": 6355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091420428784879e-07,
+      "loss": 0.6738,
+      "step": 6356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0747875701217e-07,
+      "loss": 0.7719,
+      "step": 6357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.058173525132394e-07,
+      "loss": 0.2536,
+      "step": 6358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.041578297180651e-07,
+      "loss": 0.6917,
+      "step": 6359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025001889626371e-07,
+      "loss": 0.6359,
+      "step": 6360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.008444305825601e-07,
+      "loss": 0.7016,
+      "step": 6361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991905549130595e-07,
+      "loss": 0.7052,
+      "step": 6362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.975385622889797e-07,
+      "loss": 0.6018,
+      "step": 6363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.95888453044783e-07,
+      "loss": 0.6963,
+      "step": 6364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.942402275145522e-07,
+      "loss": 0.6408,
+      "step": 6365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.925938860319881e-07,
+      "loss": 0.654,
+      "step": 6366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.909494289304098e-07,
+      "loss": 0.7543,
+      "step": 6367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.89306856542753e-07,
+      "loss": 0.6831,
+      "step": 6368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.876661692015729e-07,
+      "loss": 0.7146,
+      "step": 6369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860273672390471e-07,
+      "loss": 0.6565,
+      "step": 6370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.843904509869659e-07,
+      "loss": 0.7859,
+      "step": 6371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.827554207767407e-07,
+      "loss": 0.7246,
+      "step": 6372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.811222769393977e-07,
+      "loss": 0.6564,
+      "step": 6373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.794910198055882e-07,
+      "loss": 0.7402,
+      "step": 6374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.778616497055757e-07,
+      "loss": 0.2284,
+      "step": 6375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.762341669692396e-07,
+      "loss": 0.7594,
+      "step": 6376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.746085719260853e-07,
+      "loss": 0.6855,
+      "step": 6377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.729848649052273e-07,
+      "loss": 0.7251,
+      "step": 6378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.713630462354082e-07,
+      "loss": 0.6915,
+      "step": 6379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.697431162449741e-07,
+      "loss": 0.6866,
+      "step": 6380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.681250752619028e-07,
+      "loss": 0.6746,
+      "step": 6381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.665089236137812e-07,
+      "loss": 0.656,
+      "step": 6382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.648946616278151e-07,
+      "loss": 0.6323,
+      "step": 6383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6328228963083e-07,
+      "loss": 0.7055,
+      "step": 6384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.616718079492657e-07,
+      "loss": 0.7339,
+      "step": 6385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600632169091836e-07,
+      "loss": 0.6781,
+      "step": 6386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.584565168362578e-07,
+      "loss": 0.7621,
+      "step": 6387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.568517080557812e-07,
+      "loss": 0.6286,
+      "step": 6388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.552487908926652e-07,
+      "loss": 0.7159,
+      "step": 6389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.536477656714346e-07,
+      "loss": 0.6089,
+      "step": 6390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.520486327162367e-07,
+      "loss": 0.6925,
+      "step": 6391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504513923508315e-07,
+      "loss": 0.7357,
+      "step": 6392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48856044898597e-07,
+      "loss": 0.7775,
+      "step": 6393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.472625906825269e-07,
+      "loss": 0.6923,
+      "step": 6394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.456710300252322e-07,
+      "loss": 0.7694,
+      "step": 6395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44081363248944e-07,
+      "loss": 0.6694,
+      "step": 6396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.424935906755026e-07,
+      "loss": 0.6523,
+      "step": 6397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.409077126263707e-07,
+      "loss": 0.6908,
+      "step": 6398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.393237294226262e-07,
+      "loss": 0.2093,
+      "step": 6399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.377416413849658e-07,
+      "loss": 0.7173,
+      "step": 6400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.361614488336943e-07,
+      "loss": 0.6626,
+      "step": 6401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.345831520887403e-07,
+      "loss": 0.6807,
+      "step": 6402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.33006751469647e-07,
+      "loss": 0.691,
+      "step": 6403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.314322472955726e-07,
+      "loss": 0.597,
+      "step": 6404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.29859639885293e-07,
+      "loss": 0.5943,
+      "step": 6405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282889295571947e-07,
+      "loss": 0.7515,
+      "step": 6406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.267201166292902e-07,
+      "loss": 0.7327,
+      "step": 6407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251532014191985e-07,
+      "loss": 0.7123,
+      "step": 6408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.235881842441583e-07,
+      "loss": 0.6951,
+      "step": 6409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.220250654210236e-07,
+      "loss": 0.6797,
+      "step": 6410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204638452662625e-07,
+      "loss": 0.7464,
+      "step": 6411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.189045240959646e-07,
+      "loss": 0.7569,
+      "step": 6412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.17347102225827e-07,
+      "loss": 0.6612,
+      "step": 6413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.157915799711666e-07,
+      "loss": 0.2622,
+      "step": 6414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142379576469159e-07,
+      "loss": 0.6877,
+      "step": 6415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1268623556762e-07,
+      "loss": 0.6631,
+      "step": 6416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111364140474452e-07,
+      "loss": 0.5963,
+      "step": 6417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.095884934001628e-07,
+      "loss": 0.6662,
+      "step": 6418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08042473939171e-07,
+      "loss": 0.6474,
+      "step": 6419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.064983559774728e-07,
+      "loss": 0.6845,
+      "step": 6420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.04956139827697e-07,
+      "loss": 0.759,
+      "step": 6421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.034158258020762e-07,
+      "loss": 0.6533,
+      "step": 6422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.018774142124628e-07,
+      "loss": 0.7568,
+      "step": 6423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.003409053703269e-07,
+      "loss": 0.6829,
+      "step": 6424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.988062995867494e-07,
+      "loss": 0.6733,
+      "step": 6425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.972735971724286e-07,
+      "loss": 0.7331,
+      "step": 6426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.957427984376729e-07,
+      "loss": 0.6948,
+      "step": 6427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942139036924099e-07,
+      "loss": 0.711,
+      "step": 6428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.926869132461821e-07,
+      "loss": 0.815,
+      "step": 6429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911618274081432e-07,
+      "loss": 0.6699,
+      "step": 6430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896386464870618e-07,
+      "loss": 0.7606,
+      "step": 6431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.881173707913213e-07,
+      "loss": 0.7619,
+      "step": 6432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865980006289229e-07,
+      "loss": 0.7013,
+      "step": 6433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.850805363074752e-07,
+      "loss": 0.683,
+      "step": 6434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.835649781342078e-07,
+      "loss": 0.643,
+      "step": 6435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.820513264159589e-07,
+      "loss": 0.6262,
+      "step": 6436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.805395814591819e-07,
+      "loss": 0.6957,
+      "step": 6437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.790297435699488e-07,
+      "loss": 0.7247,
+      "step": 6438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775218130539407e-07,
+      "loss": 0.674,
+      "step": 6439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.76015790216452e-07,
+      "loss": 0.7587,
+      "step": 6440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.745116753623948e-07,
+      "loss": 0.2326,
+      "step": 6441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.730094687962906e-07,
+      "loss": 0.7453,
+      "step": 6442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.715091708222808e-07,
+      "loss": 0.6312,
+      "step": 6443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700107817441103e-07,
+      "loss": 0.6996,
+      "step": 6444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.685143018651485e-07,
+      "loss": 0.7375,
+      "step": 6445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.670197314883708e-07,
+      "loss": 0.6918,
+      "step": 6446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.655270709163707e-07,
+      "loss": 0.659,
+      "step": 6447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.64036320451351e-07,
+      "loss": 0.612,
+      "step": 6448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.62547480395128e-07,
+      "loss": 0.7476,
+      "step": 6449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.610605510491362e-07,
+      "loss": 0.7154,
+      "step": 6450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.595755327144182e-07,
+      "loss": 0.689,
+      "step": 6451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.58092425691632e-07,
+      "loss": 0.6583,
+      "step": 6452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.566112302810456e-07,
+      "loss": 0.6536,
+      "step": 6453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551319467825467e-07,
+      "loss": 0.7411,
+      "step": 6454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.536545754956291e-07,
+      "loss": 0.6754,
+      "step": 6455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521791167194024e-07,
+      "loss": 0.7547,
+      "step": 6456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.507055707525877e-07,
+      "loss": 0.7562,
+      "step": 6457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.492339378935186e-07,
+      "loss": 0.619,
+      "step": 6458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.477642184401466e-07,
+      "loss": 0.6744,
+      "step": 6459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.462964126900305e-07,
+      "loss": 0.6756,
+      "step": 6460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.448305209403382e-07,
+      "loss": 0.7062,
+      "step": 6461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.433665434878599e-07,
+      "loss": 0.6326,
+      "step": 6462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.419044806289897e-07,
+      "loss": 0.6791,
+      "step": 6463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404443326597409e-07,
+      "loss": 0.6819,
+      "step": 6464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389860998757313e-07,
+      "loss": 0.7223,
+      "step": 6465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375297825721992e-07,
+      "loss": 0.6919,
+      "step": 6466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36075381043989e-07,
+      "loss": 0.6949,
+      "step": 6467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346228955855603e-07,
+      "loss": 0.675,
+      "step": 6468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.331723264909838e-07,
+      "loss": 0.6704,
+      "step": 6469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317236740539399e-07,
+      "loss": 0.6244,
+      "step": 6470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.302769385677264e-07,
+      "loss": 0.74,
+      "step": 6471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2883212032525e-07,
+      "loss": 0.7417,
+      "step": 6472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.273892196190267e-07,
+      "loss": 0.7156,
+      "step": 6473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.259482367411895e-07,
+      "loss": 0.6699,
+      "step": 6474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24509171983476e-07,
+      "loss": 0.6634,
+      "step": 6475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230720256372446e-07,
+      "loss": 0.7462,
+      "step": 6476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.216367979934589e-07,
+      "loss": 0.6585,
+      "step": 6477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.202034893426955e-07,
+      "loss": 0.6972,
+      "step": 6478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1877209997514e-07,
+      "loss": 0.7216,
+      "step": 6479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.173426301805962e-07,
+      "loss": 0.7012,
+      "step": 6480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.159150802484736e-07,
+      "loss": 0.694,
+      "step": 6481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.144894504677944e-07,
+      "loss": 0.6771,
+      "step": 6482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.130657411271922e-07,
+      "loss": 0.7405,
+      "step": 6483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116439525149097e-07,
+      "loss": 0.7169,
+      "step": 6484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.102240849188067e-07,
+      "loss": 0.7885,
+      "step": 6485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.088061386263465e-07,
+      "loss": 0.7451,
+      "step": 6486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073901139246096e-07,
+      "loss": 0.7204,
+      "step": 6487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.059760111002843e-07,
+      "loss": 0.6355,
+      "step": 6488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045638304396694e-07,
+      "loss": 0.2188,
+      "step": 6489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.031535722286762e-07,
+      "loss": 0.6941,
+      "step": 6490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.017452367528253e-07,
+      "loss": 0.6966,
+      "step": 6491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.003388242972496e-07,
+      "loss": 0.7411,
+      "step": 6492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.989343351466924e-07,
+      "loss": 0.7171,
+      "step": 6493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975317695855075e-07,
+      "loss": 0.6928,
+      "step": 6494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.961311278976566e-07,
+      "loss": 0.6335,
+      "step": 6495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.94732410366715e-07,
+      "loss": 0.6641,
+      "step": 6496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933356172758697e-07,
+      "loss": 0.6754,
+      "step": 6497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.919407489079131e-07,
+      "loss": 0.7848,
+      "step": 6498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.905478055452529e-07,
+      "loss": 0.6713,
+      "step": 6499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891567874699044e-07,
+      "loss": 0.6567,
+      "step": 6500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.877676949634913e-07,
+      "loss": 0.7232,
+      "step": 6501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.863805283072532e-07,
+      "loss": 0.7339,
+      "step": 6502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849952877820352e-07,
+      "loss": 0.6709,
+      "step": 6503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.836119736682932e-07,
+      "loss": 0.793,
+      "step": 6504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.822305862460919e-07,
+      "loss": 0.6185,
+      "step": 6505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.80851125795111e-07,
+      "loss": 0.647,
+      "step": 6506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.79473592594637e-07,
+      "loss": 0.7048,
+      "step": 6507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.780979869235603e-07,
+      "loss": 0.6763,
+      "step": 6508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7672430906039124e-07,
+      "loss": 0.6978,
+      "step": 6509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.753525592832431e-07,
+      "loss": 0.7591,
+      "step": 6510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.739827378698436e-07,
+      "loss": 0.7663,
+      "step": 6511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.726148450975232e-07,
+      "loss": 0.6792,
+      "step": 6512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712488812432292e-07,
+      "loss": 0.6982,
+      "step": 6513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6988484658351483e-07,
+      "loss": 0.6967,
+      "step": 6514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6852274139454144e-07,
+      "loss": 0.7106,
+      "step": 6515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6716256595208287e-07,
+      "loss": 0.705,
+      "step": 6516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658043205315188e-07,
+      "loss": 0.6935,
+      "step": 6517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6444800540784263e-07,
+      "loss": 0.6227,
+      "step": 6518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630936208556536e-07,
+      "loss": 0.7211,
+      "step": 6519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6174116714916007e-07,
+      "loss": 0.697,
+      "step": 6520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6039064456218086e-07,
+      "loss": 0.7244,
+      "step": 6521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.590420533681428e-07,
+      "loss": 0.7192,
+      "step": 6522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.57695393840083e-07,
+      "loss": 0.683,
+      "step": 6523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.563506662506467e-07,
+      "loss": 0.6889,
+      "step": 6524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5500787087208733e-07,
+      "loss": 0.7312,
+      "step": 6525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5366700797626726e-07,
+      "loss": 0.6305,
+      "step": 6526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.523280778346606e-07,
+      "loss": 0.7172,
+      "step": 6527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.50991080718346e-07,
+      "loss": 0.6961,
+      "step": 6528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.496560168980102e-07,
+      "loss": 0.6607,
+      "step": 6529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4832288664395486e-07,
+      "loss": 0.7519,
+      "step": 6530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4699169022608177e-07,
+      "loss": 0.7715,
+      "step": 6531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.456624279139099e-07,
+      "loss": 0.6997,
+      "step": 6532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4433509997655834e-07,
+      "loss": 0.7044,
+      "step": 6533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.43009706682761e-07,
+      "loss": 0.7357,
+      "step": 6534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.416862483008577e-07,
+      "loss": 0.6756,
+      "step": 6535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4036472509879416e-07,
+      "loss": 0.745,
+      "step": 6536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3904513734412735e-07,
+      "loss": 0.7242,
+      "step": 6537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3772748530402143e-07,
+      "loss": 0.6529,
+      "step": 6538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3641176924524853e-07,
+      "loss": 0.766,
+      "step": 6539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3509798943419114e-07,
+      "loss": 0.7479,
+      "step": 6540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3378614613683533e-07,
+      "loss": 0.6787,
+      "step": 6541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.324762396187787e-07,
+      "loss": 0.677,
+      "step": 6542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3116827014522246e-07,
+      "loss": 0.6334,
+      "step": 6543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298622379809825e-07,
+      "loss": 0.6829,
+      "step": 6544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2855814339047727e-07,
+      "loss": 0.7649,
+      "step": 6545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.272559866377335e-07,
+      "loss": 0.7237,
+      "step": 6546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2595576798638795e-07,
+      "loss": 0.7559,
+      "step": 6547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2465748769968006e-07,
+      "loss": 0.7037,
+      "step": 6548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2336114604046406e-07,
+      "loss": 0.7108,
+      "step": 6549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2206674327119556e-07,
+      "loss": 0.7322,
+      "step": 6550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2077427965394044e-07,
+      "loss": 0.7073,
+      "step": 6551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1948375545037055e-07,
+      "loss": 0.6973,
+      "step": 6552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1819517092176796e-07,
+      "loss": 0.6825,
+      "step": 6553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1690852632901956e-07,
+      "loss": 0.7248,
+      "step": 6554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1562382193261584e-07,
+      "loss": 0.7451,
+      "step": 6555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1434105799266323e-07,
+      "loss": 0.7771,
+      "step": 6556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1306023476886946e-07,
+      "loss": 0.7205,
+      "step": 6557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1178135252055054e-07,
+      "loss": 0.6978,
+      "step": 6558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.10504411506627e-07,
+      "loss": 0.6719,
+      "step": 6559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0922941198563327e-07,
+      "loss": 0.7227,
+      "step": 6560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.079563542157028e-07,
+      "loss": 0.7502,
+      "step": 6561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0668523845458054e-07,
+      "loss": 0.7276,
+      "step": 6562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.054160649596173e-07,
+      "loss": 0.6691,
+      "step": 6563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.041488339877675e-07,
+      "loss": 0.7101,
+      "step": 6564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0288354579560043e-07,
+      "loss": 0.6362,
+      "step": 6565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.016202006392833e-07,
+      "loss": 0.6496,
+      "step": 6566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0035879877459495e-07,
+      "loss": 0.6309,
+      "step": 6567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.990993404569177e-07,
+      "loss": 0.6686,
+      "step": 6568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.97841825941242e-07,
+      "loss": 0.6776,
+      "step": 6569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.965862554821676e-07,
+      "loss": 0.5983,
+      "step": 6570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.953326293338944e-07,
+      "loss": 0.7773,
+      "step": 6571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9408094775023385e-07,
+      "loss": 0.7109,
+      "step": 6572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.928312109846022e-07,
+      "loss": 0.686,
+      "step": 6573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.915834192900181e-07,
+      "loss": 0.701,
+      "step": 6574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.903375729191161e-07,
+      "loss": 0.6815,
+      "step": 6575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.890936721241245e-07,
+      "loss": 0.6542,
+      "step": 6576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.878517171568885e-07,
+      "loss": 0.7006,
+      "step": 6577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8661170826885143e-07,
+      "loss": 0.6188,
+      "step": 6578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.853736457110702e-07,
+      "loss": 0.6828,
+      "step": 6579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8413752973419983e-07,
+      "loss": 0.7188,
+      "step": 6580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8290336058850354e-07,
+      "loss": 0.6927,
+      "step": 6581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8167113852385584e-07,
+      "loss": 0.6958,
+      "step": 6582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.804408637897306e-07,
+      "loss": 0.7616,
+      "step": 6583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.792125366352106e-07,
+      "loss": 0.708,
+      "step": 6584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7798615730898046e-07,
+      "loss": 0.6185,
+      "step": 6585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7676172605933816e-07,
+      "loss": 0.2795,
+      "step": 6586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7553924313417975e-07,
+      "loss": 0.6794,
+      "step": 6587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7431870878101076e-07,
+      "loss": 0.6214,
+      "step": 6588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.731001232469389e-07,
+      "loss": 0.6248,
+      "step": 6589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7188348677868026e-07,
+      "loss": 0.7005,
+      "step": 6590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7066879962255774e-07,
+      "loss": 0.7157,
+      "step": 6591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.694560620244958e-07,
+      "loss": 0.6144,
+      "step": 6592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6824527423002465e-07,
+      "loss": 0.7287,
+      "step": 6593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6703643648428264e-07,
+      "loss": 0.7103,
+      "step": 6594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6582954903200963e-07,
+      "loss": 0.7079,
+      "step": 6595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6462461211755673e-07,
+      "loss": 0.7053,
+      "step": 6596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6342162598486996e-07,
+      "loss": 0.6937,
+      "step": 6597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6222059087751003e-07,
+      "loss": 0.7311,
+      "step": 6598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6102150703863693e-07,
+      "loss": 0.7176,
+      "step": 6599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5982437471102194e-07,
+      "loss": 0.7232,
+      "step": 6600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.586291941370323e-07,
+      "loss": 0.7454,
+      "step": 6601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5743596555864457e-07,
+      "loss": 0.7031,
+      "step": 6602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5624468921744424e-07,
+      "loss": 0.25,
+      "step": 6603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.55055365354614e-07,
+      "loss": 0.6913,
+      "step": 6604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5386799421094684e-07,
+      "loss": 0.7229,
+      "step": 6605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.526825760268371e-07,
+      "loss": 0.6806,
+      "step": 6606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5149911104228605e-07,
+      "loss": 0.6929,
+      "step": 6607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.503175994968977e-07,
+      "loss": 0.7709,
+      "step": 6608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4913804162988176e-07,
+      "loss": 0.6853,
+      "step": 6609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4796043768005273e-07,
+      "loss": 0.6967,
+      "step": 6610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4678478788582545e-07,
+      "loss": 0.6904,
+      "step": 6611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.456110924852274e-07,
+      "loss": 0.6818,
+      "step": 6612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.444393517158817e-07,
+      "loss": 0.6814,
+      "step": 6613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.432695658150209e-07,
+      "loss": 0.6881,
+      "step": 6614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4210173501947887e-07,
+      "loss": 0.7053,
+      "step": 6615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.409358595656964e-07,
+      "loss": 0.6402,
+      "step": 6616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.39771939689717e-07,
+      "loss": 0.6804,
+      "step": 6617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.386099756271888e-07,
+      "loss": 0.7193,
+      "step": 6618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.374499676133625e-07,
+      "loss": 0.2467,
+      "step": 6619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3629191588309465e-07,
+      "loss": 0.7088,
+      "step": 6620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3513582067084215e-07,
+      "loss": 0.7328,
+      "step": 6621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.339816822106745e-07,
+      "loss": 0.703,
+      "step": 6622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328295007362514e-07,
+      "loss": 0.2484,
+      "step": 6623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3167927648084963e-07,
+      "loss": 0.231,
+      "step": 6624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3053100967734063e-07,
+      "loss": 0.7705,
+      "step": 6625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2938470055820626e-07,
+      "loss": 0.7558,
+      "step": 6626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282403493555264e-07,
+      "loss": 0.7308,
+      "step": 6627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2709795630098575e-07,
+      "loss": 0.6803,
+      "step": 6628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.259575216258759e-07,
+      "loss": 0.7748,
+      "step": 6629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.248190455610889e-07,
+      "loss": 0.7133,
+      "step": 6630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2368252833712256e-07,
+      "loss": 0.6976,
+      "step": 6631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.225479701840728e-07,
+      "loss": 0.7376,
+      "step": 6632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.214153713316459e-07,
+      "loss": 0.6429,
+      "step": 6633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2028473200914734e-07,
+      "loss": 0.7115,
+      "step": 6634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.191560524454873e-07,
+      "loss": 0.68,
+      "step": 6635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.180293328691786e-07,
+      "loss": 0.7599,
+      "step": 6636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.169045735083354e-07,
+      "loss": 0.7485,
+      "step": 6637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.157817745906799e-07,
+      "loss": 0.6422,
+      "step": 6638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1466093634353157e-07,
+      "loss": 0.7756,
+      "step": 6639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.135420589938187e-07,
+      "loss": 0.7502,
+      "step": 6640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.124251427680669e-07,
+      "loss": 0.6931,
+      "step": 6641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1131018789240743e-07,
+      "loss": 0.7278,
+      "step": 6642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.101971945925786e-07,
+      "loss": 0.6943,
+      "step": 6643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0908616309391237e-07,
+      "loss": 0.563,
+      "step": 6644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.079770936213511e-07,
+      "loss": 0.7682,
+      "step": 6645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.068699863994362e-07,
+      "loss": 0.7335,
+      "step": 6646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.057648416523162e-07,
+      "loss": 0.7575,
+      "step": 6647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.046616596037355e-07,
+      "loss": 0.6926,
+      "step": 6648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0356044047704425e-07,
+      "loss": 0.7081,
+      "step": 6649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.024611844951997e-07,
+      "loss": 0.793,
+      "step": 6650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.013638918807549e-07,
+      "loss": 0.7035,
+      "step": 6651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.002685628558688e-07,
+      "loss": 0.2322,
+      "step": 6652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9917519764230186e-07,
+      "loss": 0.7122,
+      "step": 6653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9808379646141584e-07,
+      "loss": 0.7598,
+      "step": 6654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.969943595341795e-07,
+      "loss": 0.7433,
+      "step": 6655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9590688708115766e-07,
+      "loss": 0.734,
+      "step": 6656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9482137932252295e-07,
+      "loss": 0.6235,
+      "step": 6657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.937378364780452e-07,
+      "loss": 0.7137,
+      "step": 6658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9265625876710004e-07,
+      "loss": 0.666,
+      "step": 6659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.915766464086656e-07,
+      "loss": 0.654,
+      "step": 6660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9049899962131813e-07,
+      "loss": 0.7543,
+      "step": 6661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8942331862323983e-07,
+      "loss": 0.6916,
+      "step": 6662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.883496036322109e-07,
+      "loss": 0.6674,
+      "step": 6663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.872778548656219e-07,
+      "loss": 0.7562,
+      "step": 6664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8620807254045256e-07,
+      "loss": 0.782,
+      "step": 6665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8514025687329637e-07,
+      "loss": 0.6818,
+      "step": 6666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.840744080803426e-07,
+      "loss": 0.5703,
+      "step": 6667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.830105263773819e-07,
+      "loss": 0.2099,
+      "step": 6668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.819486119798098e-07,
+      "loss": 0.6663,
+      "step": 6669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8088866510261883e-07,
+      "loss": 0.7125,
+      "step": 6670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.798306859604105e-07,
+      "loss": 0.7597,
+      "step": 6671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7877467476738254e-07,
+      "loss": 0.6801,
+      "step": 6672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7772063173733375e-07,
+      "loss": 0.7395,
+      "step": 6673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.766685570836669e-07,
+      "loss": 0.7249,
+      "step": 6674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.756184510193849e-07,
+      "loss": 0.6333,
+      "step": 6675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.745703137570932e-07,
+      "loss": 0.7921,
+      "step": 6676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7352414550899986e-07,
+      "loss": 0.6746,
+      "step": 6677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7247994648690877e-07,
+      "loss": 0.7101,
+      "step": 6678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.714377169022309e-07,
+      "loss": 0.7388,
+      "step": 6679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.703974569659773e-07,
+      "loss": 0.6823,
+      "step": 6680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.693591668887585e-07,
+      "loss": 0.7174,
+      "step": 6681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.683228468807864e-07,
+      "loss": 0.6559,
+      "step": 6682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6728849715187524e-07,
+      "loss": 0.662,
+      "step": 6683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.662561179114398e-07,
+      "loss": 0.7367,
+      "step": 6684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6522570936849734e-07,
+      "loss": 0.7185,
+      "step": 6685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641972717316643e-07,
+      "loss": 0.7063,
+      "step": 6686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6317080520915525e-07,
+      "loss": 0.7139,
+      "step": 6687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6214631000879285e-07,
+      "loss": 0.7802,
+      "step": 6688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.611237863379945e-07,
+      "loss": 0.229,
+      "step": 6689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601032344037846e-07,
+      "loss": 0.6833,
+      "step": 6690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5908465441277896e-07,
+      "loss": 0.7681,
+      "step": 6691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5806804657120264e-07,
+      "loss": 0.6811,
+      "step": 6692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5705341108487993e-07,
+      "loss": 0.7272,
+      "step": 6693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5604074815923083e-07,
+      "loss": 0.7306,
+      "step": 6694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550300579992826e-07,
+      "loss": 0.6505,
+      "step": 6695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.54021340809657e-07,
+      "loss": 0.6742,
+      "step": 6696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5301459679458187e-07,
+      "loss": 0.7235,
+      "step": 6697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5200982615788185e-07,
+      "loss": 0.6987,
+      "step": 6698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5100702910298316e-07,
+      "loss": 0.7566,
+      "step": 6699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.500062058329122e-07,
+      "loss": 0.6295,
+      "step": 6700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4900735655029594e-07,
+      "loss": 0.6788,
+      "step": 6701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480104814573636e-07,
+      "loss": 0.7234,
+      "step": 6702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4701558075594155e-07,
+      "loss": 0.6839,
+      "step": 6703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4602265464745755e-07,
+      "loss": 0.6463,
+      "step": 6704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.450317033329386e-07,
+      "loss": 0.7516,
+      "step": 6705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4404272701301635e-07,
+      "loss": 0.738,
+      "step": 6706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.430557258879174e-07,
+      "loss": 0.7381,
+      "step": 6707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4207070015746956e-07,
+      "loss": 0.6728,
+      "step": 6708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4108765002110234e-07,
+      "loss": 0.7184,
+      "step": 6709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.401065756778442e-07,
+      "loss": 0.6782,
+      "step": 6710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3912747732632633e-07,
+      "loss": 0.2175,
+      "step": 6711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3815035516477347e-07,
+      "loss": 0.6702,
+      "step": 6712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3717520939101624e-07,
+      "loss": 0.6502,
+      "step": 6713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3620204020248227e-07,
+      "loss": 0.7545,
+      "step": 6714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3523084779620065e-07,
+      "loss": 0.2337,
+      "step": 6715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3426163236879963e-07,
+      "loss": 0.7098,
+      "step": 6716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3329439411650334e-07,
+      "loss": 0.736,
+      "step": 6717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3232913323514404e-07,
+      "loss": 0.7094,
+      "step": 6718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3136584992014766e-07,
+      "loss": 0.68,
+      "step": 6719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3040454436653815e-07,
+      "loss": 0.6737,
+      "step": 6720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2944521676894537e-07,
+      "loss": 0.7278,
+      "step": 6721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2848786732159178e-07,
+      "loss": 0.7021,
+      "step": 6722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2753249621830453e-07,
+      "loss": 0.7475,
+      "step": 6723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2657910365251001e-07,
+      "loss": 0.6968,
+      "step": 6724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2562768981722938e-07,
+      "loss": 0.6784,
+      "step": 6725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2467825490508855e-07,
+      "loss": 0.7169,
+      "step": 6726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2373079910830708e-07,
+      "loss": 0.6732,
+      "step": 6727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2278532261871267e-07,
+      "loss": 0.7144,
+      "step": 6728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2184182562772215e-07,
+      "loss": 0.7249,
+      "step": 6729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2090030832635833e-07,
+      "loss": 0.654,
+      "step": 6730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1996077090523981e-07,
+      "loss": 0.7367,
+      "step": 6731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1902321355458889e-07,
+      "loss": 0.2349,
+      "step": 6732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1808763646422037e-07,
+      "loss": 0.6917,
+      "step": 6733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1715403982355277e-07,
+      "loss": 0.6509,
+      "step": 6734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.162224238216015e-07,
+      "loss": 0.7323,
+      "step": 6735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1529278864698467e-07,
+      "loss": 0.7105,
+      "step": 6736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1436513448791386e-07,
+      "loss": 0.1925,
+      "step": 6737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.134394615322022e-07,
+      "loss": 0.2193,
+      "step": 6738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1251576996726308e-07,
+      "loss": 0.6144,
+      "step": 6739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1159405998010808e-07,
+      "loss": 0.7723,
+      "step": 6740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1067433175734564e-07,
+      "loss": 0.6525,
+      "step": 6741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0975658548518462e-07,
+      "loss": 0.7238,
+      "step": 6742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0884082134943084e-07,
+      "loss": 0.7192,
+      "step": 6743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0792703953549266e-07,
+      "loss": 0.6943,
+      "step": 6744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0701524022837427e-07,
+      "loss": 0.6672,
+      "step": 6745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0610542361267805e-07,
+      "loss": 0.681,
+      "step": 6746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.051975898726055e-07,
+      "loss": 0.7452,
+      "step": 6747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0429173919195743e-07,
+      "loss": 0.6412,
+      "step": 6748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0338787175413376e-07,
+      "loss": 0.7485,
+      "step": 6749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0248598774213147e-07,
+      "loss": 0.7561,
+      "step": 6750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0158608733854445e-07,
+      "loss": 0.6625,
+      "step": 6751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0068817072556812e-07,
+      "loss": 0.7488,
+      "step": 6752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9979223808499593e-07,
+      "loss": 0.6445,
+      "step": 6753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9889828959821943e-07,
+      "loss": 0.689,
+      "step": 6754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9800632544622389e-07,
+      "loss": 0.2414,
+      "step": 6755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9711634580960037e-07,
+      "loss": 0.678,
+      "step": 6756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9622835086853252e-07,
+      "loss": 0.7899,
+      "step": 6757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9534234080280546e-07,
+      "loss": 0.6953,
+      "step": 6758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9445831579180008e-07,
+      "loss": 0.6768,
+      "step": 6759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935762760144977e-07,
+      "loss": 0.7589,
+      "step": 6760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9269622164947432e-07,
+      "loss": 0.6998,
+      "step": 6761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9181815287490856e-07,
+      "loss": 0.7232,
+      "step": 6762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9094206986857265e-07,
+      "loss": 0.6878,
+      "step": 6763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9006797280783918e-07,
+      "loss": 0.6756,
+      "step": 6764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8919586186967876e-07,
+      "loss": 0.6868,
+      "step": 6765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8832573723065905e-07,
+      "loss": 0.7238,
+      "step": 6766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8745759906694582e-07,
+      "loss": 0.7209,
+      "step": 6767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8659144755430181e-07,
+      "loss": 0.6358,
+      "step": 6768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8572728286808894e-07,
+      "loss": 0.6141,
+      "step": 6769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8486510518326727e-07,
+      "loss": 0.2284,
+      "step": 6770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8400491467439275e-07,
+      "loss": 0.6804,
+      "step": 6771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8314671151562046e-07,
+      "loss": 0.6509,
+      "step": 6772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8229049588070147e-07,
+      "loss": 0.763,
+      "step": 6773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8143626794298487e-07,
+      "loss": 0.6989,
+      "step": 6774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8058402787542228e-07,
+      "loss": 0.6761,
+      "step": 6775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.797337758505524e-07,
+      "loss": 0.6815,
+      "step": 6776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7888551204052196e-07,
+      "loss": 0.6774,
+      "step": 6777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7803923661706912e-07,
+      "loss": 0.6786,
+      "step": 6778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7719494975153238e-07,
+      "loss": 0.6188,
+      "step": 6779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7635265161484504e-07,
+      "loss": 0.7023,
+      "step": 6780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7551234237753845e-07,
+      "loss": 0.7106,
+      "step": 6781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7467402220974428e-07,
+      "loss": 0.6958,
+      "step": 6782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7383769128118676e-07,
+      "loss": 0.7049,
+      "step": 6783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7300334976119158e-07,
+      "loss": 0.6926,
+      "step": 6784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7217099781867918e-07,
+      "loss": 0.6799,
+      "step": 6785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7134063562216695e-07,
+      "loss": 0.661,
+      "step": 6786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7051226333977267e-07,
+      "loss": 0.6776,
+      "step": 6787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6968588113920658e-07,
+      "loss": 0.7035,
+      "step": 6788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6886148918777934e-07,
+      "loss": 0.6705,
+      "step": 6789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6803908765239741e-07,
+      "loss": 0.6939,
+      "step": 6790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6721867669956538e-07,
+      "loss": 0.6995,
+      "step": 6791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.664002564953826e-07,
+      "loss": 0.6908,
+      "step": 6792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6558382720554768e-07,
+      "loss": 0.7385,
+      "step": 6793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6476938899535388e-07,
+      "loss": 0.777,
+      "step": 6794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6395694202969382e-07,
+      "loss": 0.7211,
+      "step": 6795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6314648647305698e-07,
+      "loss": 0.6362,
+      "step": 6796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6233802248952658e-07,
+      "loss": 0.6228,
+      "step": 6797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61531550242785e-07,
+      "loss": 0.7573,
+      "step": 6798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6072706989611053e-07,
+      "loss": 0.6995,
+      "step": 6799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5992458161238067e-07,
+      "loss": 0.7502,
+      "step": 6800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5912408555406432e-07,
+      "loss": 0.2446,
+      "step": 6801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5832558188323188e-07,
+      "loss": 0.2297,
+      "step": 6802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5752907076154956e-07,
+      "loss": 0.676,
+      "step": 6803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5673455235027835e-07,
+      "loss": 0.6543,
+      "step": 6804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5594202681027627e-07,
+      "loss": 0.7288,
+      "step": 6805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5515149430199938e-07,
+      "loss": 0.7136,
+      "step": 6806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5436295498549746e-07,
+      "loss": 0.6682,
+      "step": 6807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.535764090204217e-07,
+      "loss": 0.6574,
+      "step": 6808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5279185656601358e-07,
+      "loss": 0.6641,
+      "step": 6809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5200929778111495e-07,
+      "loss": 0.7218,
+      "step": 6810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5122873282416128e-07,
+      "loss": 0.7466,
+      "step": 6811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.504501618531895e-07,
+      "loss": 0.7502,
+      "step": 6812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4967358502582685e-07,
+      "loss": 0.6735,
+      "step": 6813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.488990024992998e-07,
+      "loss": 0.7209,
+      "step": 6814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4812641443043174e-07,
+      "loss": 0.6647,
+      "step": 6815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4735582097563872e-07,
+      "loss": 0.7542,
+      "step": 6816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4658722229093813e-07,
+      "loss": 0.7066,
+      "step": 6817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4582061853193884e-07,
+      "loss": 0.6764,
+      "step": 6818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4505600985384895e-07,
+      "loss": 0.7534,
+      "step": 6819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4429339641147123e-07,
+      "loss": 0.2568,
+      "step": 6820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.435327783592033e-07,
+      "loss": 0.7094,
+      "step": 6821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4277415585104205e-07,
+      "loss": 0.7163,
+      "step": 6822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4201752904057675e-07,
+      "loss": 0.7419,
+      "step": 6823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4126289808099603e-07,
+      "loss": 0.6815,
+      "step": 6824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40510263125081e-07,
+      "loss": 0.7612,
+      "step": 6825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3975962432521307e-07,
+      "loss": 0.7001,
+      "step": 6826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3901098183336515e-07,
+      "loss": 0.6878,
+      "step": 6827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.38264335801106e-07,
+      "loss": 0.7169,
+      "step": 6828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3751968637960467e-07,
+      "loss": 0.7277,
+      "step": 6829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367770337196217e-07,
+      "loss": 0.7307,
+      "step": 6830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.360363779715157e-07,
+      "loss": 0.6205,
+      "step": 6831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3529771928523894e-07,
+      "loss": 0.6641,
+      "step": 6832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.345610578103418e-07,
+      "loss": 0.6895,
+      "step": 6833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.338263936959683e-07,
+      "loss": 0.2295,
+      "step": 6834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3309372709085943e-07,
+      "loss": 0.2138,
+      "step": 6835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3236305814334993e-07,
+      "loss": 0.2237,
+      "step": 6836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3163438700137254e-07,
+      "loss": 0.6431,
+      "step": 6837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3090771381245372e-07,
+      "loss": 0.6705,
+      "step": 6838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3018303872371684e-07,
+      "loss": 0.2151,
+      "step": 6839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2946036188187793e-07,
+      "loss": 0.711,
+      "step": 6840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2873968343325215e-07,
+      "loss": 0.7554,
+      "step": 6841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2802100352374836e-07,
+      "loss": 0.6478,
+      "step": 6842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2730432229887014e-07,
+      "loss": 0.6469,
+      "step": 6843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2658963990371699e-07,
+      "loss": 0.694,
+      "step": 6844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2587695648298314e-07,
+      "loss": 0.6339,
+      "step": 6845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.25166272180961e-07,
+      "loss": 0.7203,
+      "step": 6846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2445758714153433e-07,
+      "loss": 0.6877,
+      "step": 6847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.237509015081828e-07,
+      "loss": 0.7263,
+      "step": 6848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.230462154239842e-07,
+      "loss": 0.5667,
+      "step": 6849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2234352903160885e-07,
+      "loss": 0.6792,
+      "step": 6850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2164284247332292e-07,
+      "loss": 0.6919,
+      "step": 6851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.209441558909874e-07,
+      "loss": 0.7156,
+      "step": 6852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2024746942605915e-07,
+      "loss": 0.6763,
+      "step": 6853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1955278321958863e-07,
+      "loss": 0.7308,
+      "step": 6854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1886009741222337e-07,
+      "loss": 0.718,
+      "step": 6855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1816941214420452e-07,
+      "loss": 0.7451,
+      "step": 6856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1748072755536799e-07,
+      "loss": 0.7096,
+      "step": 6857
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1679404378514448e-07,
+      "loss": 0.7064,
+      "step": 6858
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1610936097256276e-07,
+      "loss": 0.7099,
+      "step": 6859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1542667925624306e-07,
+      "loss": 0.6999,
+      "step": 6860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1474599877439818e-07,
+      "loss": 0.6859,
+      "step": 6861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1406731966484342e-07,
+      "loss": 0.693,
+      "step": 6862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1339064206498107e-07,
+      "loss": 0.7146,
+      "step": 6863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.127159661118149e-07,
+      "loss": 0.6947,
+      "step": 6864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1204329194193675e-07,
+      "loss": 0.7161,
+      "step": 6865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1137261969153879e-07,
+      "loss": 0.7475,
+      "step": 6866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1070394949640351e-07,
+      "loss": 0.686,
+      "step": 6867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1003728149191262e-07,
+      "loss": 0.7313,
+      "step": 6868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0937261581303704e-07,
+      "loss": 0.74,
+      "step": 6869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0870995259434802e-07,
+      "loss": 0.6797,
+      "step": 6870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0804929197000714e-07,
+      "loss": 0.6758,
+      "step": 6871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0739063407377181e-07,
+      "loss": 0.7314,
+      "step": 6872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0673397903899541e-07,
+      "loss": 0.7412,
+      "step": 6873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0607932699862377e-07,
+      "loss": 0.6684,
+      "step": 6874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0542667808519647e-07,
+      "loss": 0.2399,
+      "step": 6875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0477603243085221e-07,
+      "loss": 0.6639,
+      "step": 6876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0412739016731899e-07,
+      "loss": 0.6738,
+      "step": 6877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0348075142592174e-07,
+      "loss": 0.7952,
+      "step": 6878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0283611633757795e-07,
+      "loss": 0.6794,
+      "step": 6879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0219348503280212e-07,
+      "loss": 0.6523,
+      "step": 6880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0155285764170241e-07,
+      "loss": 0.7161,
+      "step": 6881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.009142342939784e-07,
+      "loss": 0.809,
+      "step": 6882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0027761511892664e-07,
+      "loss": 0.6786,
+      "step": 6883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.964300024543738e-08,
+      "loss": 0.6975,
+      "step": 6884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.901038980199562e-08,
+      "loss": 0.6949,
+      "step": 6885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.837978391668001e-08,
+      "loss": 0.6789,
+      "step": 6886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.775118271716178e-08,
+      "loss": 0.6875,
+      "step": 6887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.71245863307091e-08,
+      "loss": 0.6593,
+      "step": 6888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.64999948841816e-08,
+      "loss": 0.711,
+      "step": 6889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.587740850403704e-08,
+      "loss": 0.646,
+      "step": 6890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.525682731632125e-08,
+      "loss": 0.6869,
+      "step": 6891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.463825144667927e-08,
+      "loss": 0.6999,
+      "step": 6892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.402168102034758e-08,
+      "loss": 0.6779,
+      "step": 6893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.340711616215748e-08,
+      "loss": 0.6442,
+      "step": 6894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.279455699653383e-08,
+      "loss": 0.7136,
+      "step": 6895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.218400364749414e-08,
+      "loss": 0.6882,
+      "step": 6896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.157545623865505e-08,
+      "loss": 0.792,
+      "step": 6897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.096891489322024e-08,
+      "loss": 0.7382,
+      "step": 6898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.036437973399036e-08,
+      "loss": 0.6348,
+      "step": 6899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.976185088336198e-08,
+      "loss": 0.8033,
+      "step": 6900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.916132846332081e-08,
+      "loss": 0.6272,
+      "step": 6901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.856281259544963e-08,
+      "loss": 0.72,
+      "step": 6902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.796630340092593e-08,
+      "loss": 0.6512,
+      "step": 6903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.737180100051645e-08,
+      "loss": 0.7675,
+      "step": 6904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.677930551458602e-08,
+      "loss": 0.6305,
+      "step": 6905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.6188817063092e-08,
+      "loss": 0.6751,
+      "step": 6906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.560033576558325e-08,
+      "loss": 0.6709,
+      "step": 6907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.501386174120441e-08,
+      "loss": 0.6701,
+      "step": 6908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.442939510869497e-08,
+      "loss": 0.7736,
+      "step": 6909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.38469359863836e-08,
+      "loss": 0.7443,
+      "step": 6910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.32664844921971e-08,
+      "loss": 0.6848,
+      "step": 6911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.268804074365255e-08,
+      "loss": 0.2296,
+      "step": 6912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.211160485786296e-08,
+      "loss": 0.7133,
+      "step": 6913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.153717695153385e-08,
+      "loss": 0.6006,
+      "step": 6914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.09647571409633e-08,
+      "loss": 0.6707,
+      "step": 6915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.039434554204417e-08,
+      "loss": 0.6229,
+      "step": 6916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.982594227026074e-08,
+      "loss": 0.807,
+      "step": 6917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.925954744069542e-08,
+      "loss": 0.2239,
+      "step": 6918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.869516116801757e-08,
+      "loss": 0.7109,
+      "step": 6919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.813278356649468e-08,
+      "loss": 0.6533,
+      "step": 6920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.757241474998566e-08,
+      "loss": 0.7098,
+      "step": 6921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.701405483194202e-08,
+      "loss": 0.6841,
+      "step": 6922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.645770392541108e-08,
+      "loss": 0.6534,
+      "step": 6923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.590336214303052e-08,
+      "loss": 0.7203,
+      "step": 6924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.535102959703388e-08,
+      "loss": 0.6931,
+      "step": 6925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.480070639924508e-08,
+      "loss": 0.7121,
+      "step": 6926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.425239266108386e-08,
+      "loss": 0.7241,
+      "step": 6927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.370608849356254e-08,
+      "loss": 0.7386,
+      "step": 6928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.316179400728485e-08,
+      "loss": 0.7335,
+      "step": 6929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.261950931244932e-08,
+      "loss": 0.6211,
+      "step": 6930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.207923451884591e-08,
+      "loss": 0.6202,
+      "step": 6931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.154096973586155e-08,
+      "loss": 0.7154,
+      "step": 6932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.100471507247242e-08,
+      "loss": 0.658,
+      "step": 6933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.04704706372461e-08,
+      "loss": 0.7393,
+      "step": 6934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.993823653835053e-08,
+      "loss": 0.7,
+      "step": 6935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.940801288353837e-08,
+      "loss": 0.6869,
+      "step": 6936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.887979978016157e-08,
+      "loss": 0.2666,
+      "step": 6937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.835359733516011e-08,
+      "loss": 0.6878,
+      "step": 6938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.78294056550699e-08,
+      "loss": 0.7195,
+      "step": 6939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.730722484602048e-08,
+      "loss": 0.7466,
+      "step": 6940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.678705501373062e-08,
+      "loss": 0.6928,
+      "step": 6941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.626889626351496e-08,
+      "loss": 0.6744,
+      "step": 6942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.57527487002807e-08,
+      "loss": 0.6534,
+      "step": 6943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.523861242852536e-08,
+      "loss": 0.6744,
+      "step": 6944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.472648755234456e-08,
+      "loss": 0.685,
+      "step": 6945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.421637417541981e-08,
+      "loss": 0.6565,
+      "step": 6946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.370827240103072e-08,
+      "loss": 0.2342,
+      "step": 6947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.320218233204722e-08,
+      "loss": 0.698,
+      "step": 6948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.269810407093291e-08,
+      "loss": 0.6294,
+      "step": 6949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.21960377197428e-08,
+      "loss": 0.6666,
+      "step": 6950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.169598338012672e-08,
+      "loss": 0.744,
+      "step": 6951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.119794115332478e-08,
+      "loss": 0.6575,
+      "step": 6952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.070191114017076e-08,
+      "loss": 0.7549,
+      "step": 6953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.020789344109213e-08,
+      "loss": 0.7079,
+      "step": 6954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.971588815610663e-08,
+      "loss": 0.7002,
+      "step": 6955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9225895384827926e-08,
+      "loss": 0.6615,
+      "step": 6956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.873791522645666e-08,
+      "loss": 0.6897,
+      "step": 6957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8251947779793815e-08,
+      "loss": 0.6862,
+      "step": 6958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.776799314322512e-08,
+      "loss": 0.735,
+      "step": 6959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.72860514147322e-08,
+      "loss": 0.2015,
+      "step": 6960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.680612269189145e-08,
+      "loss": 0.7285,
+      "step": 6961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.632820707186848e-08,
+      "loss": 0.6758,
+      "step": 6962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5852304651422554e-08,
+      "loss": 0.6966,
+      "step": 6963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.537841552690326e-08,
+      "loss": 0.6876,
+      "step": 6964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.490653979425608e-08,
+      "loss": 0.6307,
+      "step": 6965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4436677549016825e-08,
+      "loss": 0.693,
+      "step": 6966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.396882888631383e-08,
+      "loss": 0.6728,
+      "step": 6967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3502993900869105e-08,
+      "loss": 0.6462,
+      "step": 6968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3039172686992767e-08,
+      "loss": 0.6536,
+      "step": 6969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.257736533859303e-08,
+      "loss": 0.6905,
+      "step": 6970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2117571949167336e-08,
+      "loss": 0.7055,
+      "step": 6971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165979261180454e-08,
+      "loss": 0.6977,
+      "step": 6972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1204027419187174e-08,
+      "loss": 0.7295,
+      "step": 6973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.075027646358921e-08,
+      "loss": 0.7,
+      "step": 6974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.029853983687827e-08,
+      "loss": 0.7372,
+      "step": 6975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9848817630512304e-08,
+      "loss": 0.6944,
+      "step": 6976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9401109935542926e-08,
+      "loss": 0.7144,
+      "step": 6977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.895541684261318e-08,
+      "loss": 0.7297,
+      "step": 6978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.851173844195756e-08,
+      "loss": 0.7843,
+      "step": 6979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.80700748234042e-08,
+      "loss": 0.7727,
+      "step": 6980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.763042607637047e-08,
+      "loss": 0.6625,
+      "step": 6981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7192792289870726e-08,
+      "loss": 0.6774,
+      "step": 6982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.675717355250631e-08,
+      "loss": 0.6627,
+      "step": 6983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6323569952474444e-08,
+      "loss": 0.6374,
+      "step": 6984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5891981577560475e-08,
+      "loss": 0.6216,
+      "step": 6985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5462408515146716e-08,
+      "loss": 0.624,
+      "step": 6986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.503485085220249e-08,
+      "loss": 0.2569,
+      "step": 6987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4609308675291894e-08,
+      "loss": 0.6372,
+      "step": 6988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.418578207057045e-08,
+      "loss": 0.215,
+      "step": 6989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.376427112378512e-08,
+      "loss": 0.7404,
+      "step": 6990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.334477592027653e-08,
+      "loss": 0.7104,
+      "step": 6991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2927296544975624e-08,
+      "loss": 0.6806,
+      "step": 6992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.251183308240259e-08,
+      "loss": 0.7107,
+      "step": 6993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.209838561667567e-08,
+      "loss": 0.6666,
+      "step": 6994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.168695423150015e-08,
+      "loss": 0.6574,
+      "step": 6995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.127753901017606e-08,
+      "loss": 0.674,
+      "step": 6996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0870140035590425e-08,
+      "loss": 0.725,
+      "step": 6997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0464757390228374e-08,
+      "loss": 0.7254,
+      "step": 6998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.006139115616314e-08,
+      "loss": 0.7001,
+      "step": 6999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9660041415060526e-08,
+      "loss": 0.2426,
+      "step": 7000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.926070824817774e-08,
+      "loss": 0.7345,
+      "step": 7001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.886339173636233e-08,
+      "loss": 0.7845,
+      "step": 7002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.846809196005774e-08,
+      "loss": 0.7535,
+      "step": 7003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.80748089992955e-08,
+      "loss": 0.7105,
+      "step": 7004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.768354293369969e-08,
+      "loss": 0.7485,
+      "step": 7005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.729429384248584e-08,
+      "loss": 0.6707,
+      "step": 7006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.690706180446202e-08,
+      "loss": 0.6925,
+      "step": 7007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6521846898026625e-08,
+      "loss": 0.6779,
+      "step": 7008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6138649201171715e-08,
+      "loss": 0.7414,
+      "step": 7009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.575746879147968e-08,
+      "loss": 0.688,
+      "step": 7010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.537830574612211e-08,
+      "loss": 0.6144,
+      "step": 7011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.50011601418665e-08,
+      "loss": 0.7085,
+      "step": 7012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4626032055070647e-08,
+      "loss": 0.726,
+      "step": 7013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.425292156168158e-08,
+      "loss": 0.6997,
+      "step": 7014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3881828737238886e-08,
+      "loss": 0.7316,
+      "step": 7015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3512753656875785e-08,
+      "loss": 0.2008,
+      "step": 7016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.314569639531362e-08,
+      "loss": 0.7415,
+      "step": 7017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.278065702686961e-08,
+      "loss": 0.6742,
+      "step": 7018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2417635625446864e-08,
+      "loss": 0.6942,
+      "step": 7019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.205663226454436e-08,
+      "loss": 0.7261,
+      "step": 7020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.169764701725031e-08,
+      "loss": 0.7051,
+      "step": 7021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1340679956246566e-08,
+      "loss": 0.7583,
+      "step": 7022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.098573115380199e-08,
+      "loss": 0.6949,
+      "step": 7023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.063280068178243e-08,
+      "loss": 0.6615,
+      "step": 7024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.028188861164072e-08,
+      "loss": 0.6776,
+      "step": 7025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9932995014424484e-08,
+      "loss": 0.6813,
+      "step": 7026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9586119960768322e-08,
+      "loss": 0.7076,
+      "step": 7027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.924126352090162e-08,
+      "loss": 0.6963,
+      "step": 7028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.889842576464408e-08,
+      "loss": 0.705,
+      "step": 7029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8557606761407952e-08,
+      "loss": 0.6988,
+      "step": 7030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8218806580194713e-08,
+      "loss": 0.6489,
+      "step": 7031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7882025289597268e-08,
+      "loss": 0.6628,
+      "step": 7032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.754726295780108e-08,
+      "loss": 0.7325,
+      "step": 7033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.721451965258304e-08,
+      "loss": 0.7034,
+      "step": 7034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6883795441309256e-08,
+      "loss": 0.6876,
+      "step": 7035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6555090390939508e-08,
+      "loss": 0.6971,
+      "step": 7036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.622840456802167e-08,
+      "loss": 0.7186,
+      "step": 7037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.59037380386995e-08,
+      "loss": 0.7788,
+      "step": 7038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.558109086870264e-08,
+      "loss": 0.7421,
+      "step": 7039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5260463123354395e-08,
+      "loss": 0.22,
+      "step": 7040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4941854867571725e-08,
+      "loss": 0.6842,
+      "step": 7041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4625266165856366e-08,
+      "loss": 0.6926,
+      "step": 7042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4310697082309263e-08,
+      "loss": 0.7299,
+      "step": 7043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3998147680613927e-08,
+      "loss": 0.2506,
+      "step": 7044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.368761802405195e-08,
+      "loss": 0.7235,
+      "step": 7045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.337910817549305e-08,
+      "loss": 0.7404,
+      "step": 7046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3072618197398366e-08,
+      "loss": 0.7296,
+      "step": 7047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.276814815181827e-08,
+      "loss": 0.6445,
+      "step": 7048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.246569810039678e-08,
+      "loss": 0.6509,
+      "step": 7049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2165268104369365e-08,
+      "loss": 0.7279,
+      "step": 7050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1866858224559582e-08,
+      "loss": 0.7088,
+      "step": 7051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.157046852138467e-08,
+      "loss": 0.6867,
+      "step": 7052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1276099054851064e-08,
+      "loss": 0.6784,
+      "step": 7053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0983749884556646e-08,
+      "loss": 0.7266,
+      "step": 7054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0693421069691853e-08,
+      "loss": 0.7135,
+      "step": 7055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.040511266903522e-08,
+      "loss": 0.7174,
+      "step": 7056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011882474095894e-08,
+      "loss": 0.2256,
+      "step": 7057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9834557343424433e-08,
+      "loss": 0.6518,
+      "step": 7058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9552310533984543e-08,
+      "loss": 0.6212,
+      "step": 7059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.927208436978356e-08,
+      "loss": 0.2215,
+      "step": 7060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8993878907554995e-08,
+      "loss": 0.6973,
+      "step": 7061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8717694203626013e-08,
+      "loss": 0.6593,
+      "step": 7062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8443530313912993e-08,
+      "loss": 0.6735,
+      "step": 7063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.817138729392154e-08,
+      "loss": 0.6666,
+      "step": 7064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7901265198750905e-08,
+      "loss": 0.7684,
+      "step": 7065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7633164083090682e-08,
+      "loss": 0.6335,
+      "step": 7066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.736708400121967e-08,
+      "loss": 0.753,
+      "step": 7067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7103025007009223e-08,
+      "loss": 0.745,
+      "step": 7068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6840987153921018e-08,
+      "loss": 0.6589,
+      "step": 7069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6580970495007065e-08,
+      "loss": 0.7167,
+      "step": 7070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6322975082909698e-08,
+      "loss": 0.6856,
+      "step": 7071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.60670009698638e-08,
+      "loss": 0.5703,
+      "step": 7072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5813048207692362e-08,
+      "loss": 0.7616,
+      "step": 7073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.556111684781314e-08,
+      "loss": 0.6753,
+      "step": 7074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5311206941230892e-08,
+      "loss": 0.789,
+      "step": 7075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5063318538542925e-08,
+      "loss": 0.606,
+      "step": 7076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4817451689935758e-08,
+      "loss": 0.7406,
+      "step": 7077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4573606445189569e-08,
+      "loss": 0.6669,
+      "step": 7078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4331782853671539e-08,
+      "loss": 0.6922,
+      "step": 7079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.40919809643425e-08,
+      "loss": 0.6893,
+      "step": 7080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3854200825751395e-08,
+      "loss": 0.7173,
+      "step": 7081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3618442486041938e-08,
+      "loss": 0.6555,
+      "step": 7082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3384705992942615e-08,
+      "loss": 0.6886,
+      "step": 7083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3152991393778903e-08,
+      "loss": 0.656,
+      "step": 7084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2923298735461054e-08,
+      "loss": 0.7536,
+      "step": 7085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2695628064495202e-08,
+      "loss": 0.7535,
+      "step": 7086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2469979426973366e-08,
+      "loss": 0.7348,
+      "step": 7087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2246352868582335e-08,
+      "loss": 0.2448,
+      "step": 7088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2024748434597e-08,
+      "loss": 0.6383,
+      "step": 7089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1805166169882588e-08,
+      "loss": 0.6898,
+      "step": 7090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1587606118896866e-08,
+      "loss": 0.6841,
+      "step": 7091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1372068325686824e-08,
+      "loss": 0.726,
+      "step": 7092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.115855283388978e-08,
+      "loss": 0.7602,
+      "step": 7093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0947059686735594e-08,
+      "loss": 0.6818,
+      "step": 7094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0737588927042241e-08,
+      "loss": 0.7084,
+      "step": 7095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0530140597219129e-08,
+      "loss": 0.2085,
+      "step": 7096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0324714739265996e-08,
+      "loss": 0.6376,
+      "step": 7097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0121311394775124e-08,
+      "loss": 0.5834,
+      "step": 7098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.919930604925798e-09,
+      "loss": 0.7409,
+      "step": 7099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.720572410489849e-09,
+      "loss": 0.7146,
+      "step": 7100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.523236851830541e-09,
+      "loss": 0.6607,
+      "step": 7101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.327923968900365e-09,
+      "loss": 0.7877,
+      "step": 7102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.134633801241022e-09,
+      "loss": 0.6745,
+      "step": 7103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.943366387987872e-09,
+      "loss": 0.725,
+      "step": 7104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.754121767863276e-09,
+      "loss": 0.701,
+      "step": 7105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.566899979183252e-09,
+      "loss": 0.6781,
+      "step": 7106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.381701059851921e-09,
+      "loss": 0.6567,
+      "step": 7107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.198525047364847e-09,
+      "loss": 0.6953,
+      "step": 7108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.01737197880792e-09,
+      "loss": 0.7825,
+      "step": 7109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.838241890858466e-09,
+      "loss": 0.743,
+      "step": 7110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.661134819781923e-09,
+      "loss": 0.6603,
+      "step": 7111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.486050801436273e-09,
+      "loss": 0.7419,
+      "step": 7112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.312989871268716e-09,
+      "loss": 0.7489,
+      "step": 7113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.141952064316782e-09,
+      "loss": 0.7121,
+      "step": 7114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.972937415209435e-09,
+      "loss": 0.2274,
+      "step": 7115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.805945958164862e-09,
+      "loss": 0.7488,
+      "step": 7116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.640977726993791e-09,
+      "loss": 0.6778,
+      "step": 7117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.478032755093955e-09,
+      "loss": 0.7335,
+      "step": 7118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.317111075455628e-09,
+      "loss": 0.7418,
+      "step": 7119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.158212720659418e-09,
+      "loss": 0.6821,
+      "step": 7120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.001337722877365e-09,
+      "loss": 0.7011,
+      "step": 7121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8464861138674e-09,
+      "loss": 0.676,
+      "step": 7122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.693657924983332e-09,
+      "loss": 0.7737,
+      "step": 7123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.542853187164854e-09,
+      "loss": 0.668,
+      "step": 7124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.394071930946431e-09,
+      "loss": 0.2219,
+      "step": 7125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.247314186447306e-09,
+      "loss": 0.6902,
+      "step": 7126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.102579983382594e-09,
+      "loss": 0.714,
+      "step": 7127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.959869351055524e-09,
+      "loss": 0.6996,
+      "step": 7128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.819182318356319e-09,
+      "loss": 0.6799,
+      "step": 7129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.680518913772192e-09,
+      "loss": 0.7264,
+      "step": 7130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5438791653740214e-09,
+      "loss": 0.615,
+      "step": 7131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.409263100827455e-09,
+      "loss": 0.7332,
+      "step": 7132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.276670747386247e-09,
+      "loss": 0.6199,
+      "step": 7133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1461021318955904e-09,
+      "loss": 0.6908,
+      "step": 7134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.017557280791007e-09,
+      "loss": 0.6956,
+      "step": 7135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.891036220096123e-09,
+      "loss": 0.644,
+      "step": 7136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7665389754282246e-09,
+      "loss": 0.7139,
+      "step": 7137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.644065571991595e-09,
+      "loss": 0.7249,
+      "step": 7138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.523616034583066e-09,
+      "loss": 0.6315,
+      "step": 7139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.405190387588686e-09,
+      "loss": 0.7315,
+      "step": 7140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2887886549859417e-09,
+      "loss": 0.7506,
+      "step": 7141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1744108603393166e-09,
+      "loss": 0.6603,
+      "step": 7142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.062057026806953e-09,
+      "loss": 0.7337,
+      "step": 7143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.95172717713732e-09,
+      "loss": 0.6876,
+      "step": 7144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8434213336658854e-09,
+      "loss": 0.6959,
+      "step": 7145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.737139518320664e-09,
+      "loss": 0.7434,
+      "step": 7146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.632881752621108e-09,
+      "loss": 0.6549,
+      "step": 7147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.530648057673668e-09,
+      "loss": 0.6532,
+      "step": 7148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4304384541762315e-09,
+      "loss": 0.7025,
+      "step": 7149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.332252962419235e-09,
+      "loss": 0.2712,
+      "step": 7150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2360916022790004e-09,
+      "loss": 0.7137,
+      "step": 7151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1419543932266198e-09,
+      "loss": 0.6621,
+      "step": 7152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.049841354320181e-09,
+      "loss": 0.756,
+      "step": 7153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.959752504208101e-09,
+      "loss": 0.6778,
+      "step": 7154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8716878611313438e-09,
+      "loss": 0.7279,
+      "step": 7155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7856474429189807e-09,
+      "loss": 0.7236,
+      "step": 7156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.701631266989301e-09,
+      "loss": 0.6634,
+      "step": 7157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6196393503542519e-09,
+      "loss": 0.7454,
+      "step": 7158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.539671709613888e-09,
+      "loss": 0.7119,
+      "step": 7159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4617283609563716e-09,
+      "loss": 0.702,
+      "step": 7160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3858093201646328e-09,
+      "loss": 0.6759,
+      "step": 7161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3119146026085994e-09,
+      "loss": 0.7293,
+      "step": 7162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2400442232474163e-09,
+      "loss": 0.2224,
+      "step": 7163
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1701981966338871e-09,
+      "loss": 0.6987,
+      "step": 7164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.102376536907812e-09,
+      "loss": 0.7329,
+      "step": 7165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0365792578015398e-09,
+      "loss": 0.6453,
+      "step": 7166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.728063726366366e-10,
+      "loss": 0.6992,
+      "step": 7167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.110578943227754e-10,
+      "loss": 0.6501,
+      "step": 7168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.51333835362178e-10,
+      "loss": 0.6835,
+      "step": 7169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.936342078473936e-10,
+      "loss": 0.6891,
+      "step": 7170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.379590234601885e-10,
+      "loss": 0.815,
+      "step": 7171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.843082934726575e-10,
+      "loss": 0.6511,
+      "step": 7172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.326820287461121e-10,
+      "loss": 0.7025,
+      "step": 7173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.830802397333024e-10,
+      "loss": 0.7622,
+      "step": 7174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.355029364761954e-10,
+      "loss": 0.7061,
+      "step": 7175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.899501286081965e-10,
+      "loss": 0.7188,
+      "step": 7176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4642182535192814e-10,
+      "loss": 0.2377,
+      "step": 7177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.049180355192306e-10,
+      "loss": 0.6241,
+      "step": 7178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6543876751338194e-10,
+      "loss": 0.7108,
+      "step": 7179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2798402932798787e-10,
+      "loss": 0.7463,
+      "step": 7180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.925538285458718e-10,
+      "loss": 0.77,
+      "step": 7181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.591481723390743e-10,
+      "loss": 0.6406,
+      "step": 7182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2776706747218437e-10,
+      "loss": 0.7461,
+      "step": 7183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9841052029900832e-10,
+      "loss": 0.6899,
+      "step": 7184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.710785367614598e-10,
+      "loss": 0.7282,
+      "step": 7185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.457711223940006e-10,
+      "loss": 0.7171,
+      "step": 7186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.224882823214202e-10,
+      "loss": 0.6932,
+      "step": 7187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0123002125550507e-10,
+      "loss": 0.7874,
+      "step": 7188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.199634350170015e-11,
+      "loss": 0.6574,
+      "step": 7189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.478725295466781e-11,
+      "loss": 0.6462,
+      "step": 7190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.960275309606743e-11,
+      "loss": 0.7086,
+      "step": 7191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.644284700232703e-11,
+      "loss": 0.6461,
+      "step": 7192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5307537337981857e-11,
+      "loss": 0.7814,
+      "step": 7193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6196826355674434e-11,
+      "loss": 0.665,
+      "step": 7194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.110715900595424e-12,
+      "loss": 0.7519,
+      "step": 7195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0492074082632e-12,
+      "loss": 0.6682,
+      "step": 7196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.012301903413615e-12,
+      "loss": 0.7206,
+      "step": 7197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.666,
+      "step": 7198
+    },
+    {
+      "epoch": 1.0,
+      "step": 7198,
+      "total_flos": 2.5355341370308428e+19,
+      "train_loss": 0.7314568244606033,
+      "train_runtime": 143874.6268,
+      "train_samples_per_second": 6.404,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "max_steps": 7198,
+  "num_train_epochs": 1,
+  "total_flos": 2.5355341370308428e+19,
+  "trial_name": null,
+  "trial_params": null
+}