diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26935 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9987463435018804,
+  "global_step": 4485,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4814814814814817e-07,
+      "loss": 1.2797,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 1.2151,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.2671,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.925925925925927e-07,
+      "loss": 1.34,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 1.3541,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.2493,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0370370370370371e-06,
+      "loss": 1.2993,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1851851851851854e-06,
+      "loss": 1.2836,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.2388,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.5114,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62962962962963e-06,
+      "loss": 1.3228,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 1.3813,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.925925925925926e-06,
+      "loss": 1.4845,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0740740740740742e-06,
+      "loss": 1.2389,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.2721,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3703703703703707e-06,
+      "loss": 1.3042,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5185185185185186e-06,
+      "loss": 1.1457,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.3571,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.814814814814815e-06,
+      "loss": 1.4283,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 1.3648,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 1.2484,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.25925925925926e-06,
+      "loss": 1.2263,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4074074074074077e-06,
+      "loss": 1.5211,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 1.1479,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 1.4861,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.851851851851852e-06,
+      "loss": 1.4382,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3721,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1481481481481485e-06,
+      "loss": 1.2091,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.296296296296296e-06,
+      "loss": 1.1798,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.4912,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.592592592592593e-06,
+      "loss": 1.3095,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7407407407407415e-06,
+      "loss": 1.35,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 1.2305,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.037037037037037e-06,
+      "loss": 1.406,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 1.3868,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.3133,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.481481481481482e-06,
+      "loss": 1.165,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.481481481481482e-06,
+      "loss": 1.1171,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.62962962962963e-06,
+      "loss": 1.511,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 1.2642,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 1.1439,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0740740740740745e-06,
+      "loss": 1.3145,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 1.1727,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 1.251,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 1.1578,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.238,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.814814814814815e-06,
+      "loss": 1.3158,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.962962962962964e-06,
+      "loss": 1.3345,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 1.0969,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2592592592592605e-06,
+      "loss": 1.0856,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 1.1338,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 1.0331,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 1.1464,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 1.234,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 1.1518,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1681,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 1.0643,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.296296296296297e-06,
+      "loss": 1.1124,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 1.0414,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.592592592592593e-06,
+      "loss": 1.2404,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.740740740740741e-06,
+      "loss": 0.9904,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.1016,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.037037037037037e-06,
+      "loss": 1.0413,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.9822,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.9915,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 1.0623,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 1.0228,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.9984,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.925925925925927e-06,
+      "loss": 1.0333,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0074074074074074e-05,
+      "loss": 0.9983,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.9669,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.9729,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0518518518518519e-05,
+      "loss": 0.9986,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.9706,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0814814814814816e-05,
+      "loss": 0.9252,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0962962962962965e-05,
+      "loss": 1.0086,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.0158,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.125925925925926e-05,
+      "loss": 0.9692,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1407407407407409e-05,
+      "loss": 0.9979,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.9429,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1703703703703703e-05,
+      "loss": 1.0394,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 1.0078,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2e-05,
+      "loss": 0.9951,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.9519,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2296296296296298e-05,
+      "loss": 0.9088,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.9586,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.9398,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2740740740740742e-05,
+      "loss": 0.9954,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 0.9984,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.303703703703704e-05,
+      "loss": 0.9599,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3185185185185185e-05,
+      "loss": 0.9514,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9632,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3481481481481482e-05,
+      "loss": 0.9531,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.362962962962963e-05,
+      "loss": 0.8256,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 0.9765,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3925925925925928e-05,
+      "loss": 0.9251,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.9957,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.9672,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4370370370370372e-05,
+      "loss": 0.8556,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4518518518518521e-05,
+      "loss": 0.9792,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.8455,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.8489,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.9314,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.9739,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.525925925925926e-05,
+      "loss": 0.9031,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.9013,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.9302,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5703703703703705e-05,
+      "loss": 0.8036,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5851851851851852e-05,
+      "loss": 0.9055,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.7863,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.614814814814815e-05,
+      "loss": 0.8655,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.8673,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 0.8374,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.9161,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.674074074074074e-05,
+      "loss": 0.9239,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.7975,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.8612,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7185185185185185e-05,
+      "loss": 0.9015,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.7949,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7481481481481483e-05,
+      "loss": 0.8967,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7629629629629633e-05,
+      "loss": 0.8585,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.8709,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7925925925925927e-05,
+      "loss": 0.9206,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8074074074074074e-05,
+      "loss": 0.8526,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 0.848,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.837037037037037e-05,
+      "loss": 0.8084,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.8315,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.8068,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8814814814814816e-05,
+      "loss": 0.7527,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8962962962962966e-05,
+      "loss": 0.7445,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.8543,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.8776,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9407407407407407e-05,
+      "loss": 0.7545,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.7918,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9703703703703704e-05,
+      "loss": 0.7893,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9851851851851855e-05,
+      "loss": 0.8248,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.8405,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997392098304e-05,
+      "loss": 0.8477,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999989568394573e-05,
+      "loss": 0.8111,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997652889289e-05,
+      "loss": 0.812,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999958273600055e-05,
+      "loss": 0.8115,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993480252559e-05,
+      "loss": 0.7606,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999906115681737e-05,
+      "loss": 0.8605,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999872213083455e-05,
+      "loss": 0.8635,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999833094748436e-05,
+      "loss": 0.8002,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999788760697073e-05,
+      "loss": 0.7852,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999739210952493e-05,
+      "loss": 0.835,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999684445540547e-05,
+      "loss": 0.7361,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999962446448979e-05,
+      "loss": 0.7799,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999955926783151e-05,
+      "loss": 0.715,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999488855599718e-05,
+      "loss": 0.7817,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999413227831133e-05,
+      "loss": 0.7471,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999332384565204e-05,
+      "loss": 0.8037,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999246325844094e-05,
+      "loss": 0.7626,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999915505171269e-05,
+      "loss": 0.7982,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999058562218605e-05,
+      "loss": 0.7731,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999895685741216e-05,
+      "loss": 0.7945,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998849937346402e-05,
+      "loss": 0.8014,
+      "step": 158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99987378020771e-05,
+      "loss": 0.9183,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998620451662748e-05,
+      "loss": 0.7546,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999849788616454e-05,
+      "loss": 0.767,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998370105646414e-05,
+      "loss": 0.7927,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998237110175018e-05,
+      "loss": 0.7555,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999809889981971e-05,
+      "loss": 0.76,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999795547465259e-05,
+      "loss": 0.8231,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 0.7582,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999765298018484e-05,
+      "loss": 0.728,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997493911041994e-05,
+      "loss": 0.7741,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999732962740288e-05,
+      "loss": 0.7482,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997160129353183e-05,
+      "loss": 0.7596,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996985416981314e-05,
+      "loss": 0.732,
+      "step": 171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99968054903784e-05,
+      "loss": 0.8493,
+      "step": 172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996620349638286e-05,
+      "loss": 0.7532,
+      "step": 173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996429994857533e-05,
+      "loss": 0.7612,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996234426135435e-05,
+      "loss": 0.7859,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999603364357399e-05,
+      "loss": 0.8629,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995827647277926e-05,
+      "loss": 0.7485,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995616437354684e-05,
+      "loss": 0.7328,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999540001391443e-05,
+      "loss": 0.796,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999517837707004e-05,
+      "loss": 0.7298,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994951526937126e-05,
+      "loss": 0.7346,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994719463633996e-05,
+      "loss": 0.7478,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994482187281702e-05,
+      "loss": 0.7023,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999423969800399e-05,
+      "loss": 0.7973,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999399199592735e-05,
+      "loss": 0.7769,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999373908118097e-05,
+      "loss": 0.7453,
+      "step": 186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999348095389677e-05,
+      "loss": 0.754,
+      "step": 187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9993217614209384e-05,
+      "loss": 0.7688,
+      "step": 188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999294906225616e-05,
+      "loss": 0.7374,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992675298177174e-05,
+      "loss": 0.7312,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992396322115213e-05,
+      "loss": 0.6732,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999211213421579e-05,
+      "loss": 0.6939,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991822734627126e-05,
+      "loss": 0.705,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999152812350017e-05,
+      "loss": 0.7735,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.8321,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990923267248753e-05,
+      "loss": 0.8343,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999061302243977e-05,
+      "loss": 0.7735,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990297566723456e-05,
+      "loss": 0.8102,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989976900264348e-05,
+      "loss": 0.7987,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989651023229694e-05,
+      "loss": 0.7699,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989319935789475e-05,
+      "loss": 0.7656,
+      "step": 201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988983638116372e-05,
+      "loss": 0.7321,
+      "step": 202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998864213038579e-05,
+      "loss": 0.775,
+      "step": 203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988295412775858e-05,
+      "loss": 0.8108,
+      "step": 204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987943485467413e-05,
+      "loss": 0.7295,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987586348644016e-05,
+      "loss": 0.7731,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998722400249194e-05,
+      "loss": 0.7262,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986856447200182e-05,
+      "loss": 0.7423,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986483682960445e-05,
+      "loss": 0.7451,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986105709967162e-05,
+      "loss": 0.7024,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985722528417477e-05,
+      "loss": 0.735,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998533413851124e-05,
+      "loss": 0.7603,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984940540451037e-05,
+      "loss": 0.782,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984541734442157e-05,
+      "loss": 0.7207,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984137720692615e-05,
+      "loss": 0.6792,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998372849941313e-05,
+      "loss": 0.726,
+      "step": 216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9983314070817142e-05,
+      "loss": 0.7219,
+      "step": 217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982894435120817e-05,
+      "loss": 0.6455,
+      "step": 218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982469592543022e-05,
+      "loss": 0.7213,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998203954330535e-05,
+      "loss": 0.7169,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981604287632104e-05,
+      "loss": 0.7848,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981163825750306e-05,
+      "loss": 0.6868,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980718157889695e-05,
+      "loss": 0.7377,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 0.6529,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979811205164544e-05,
+      "loss": 0.7009,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997934992077305e-05,
+      "loss": 0.7234,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978883431348845e-05,
+      "loss": 0.6719,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978411737135232e-05,
+      "loss": 0.8173,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977934838378236e-05,
+      "loss": 0.7113,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99774527353266e-05,
+      "loss": 0.7794,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997696542823178e-05,
+      "loss": 0.6808,
+      "step": 231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997647291734795e-05,
+      "loss": 0.6931,
+      "step": 232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975975202931982e-05,
+      "loss": 0.7381,
+      "step": 233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997547228524349e-05,
+      "loss": 0.7985,
+      "step": 234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974964164544775e-05,
+      "loss": 0.7121,
+      "step": 235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974450841100864e-05,
+      "loss": 0.7304,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973932315179502e-05,
+      "loss": 0.7554,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973408587051132e-05,
+      "loss": 0.6628,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997287965698893e-05,
+      "loss": 0.7695,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997234552526877e-05,
+      "loss": 0.7003,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971806192169252e-05,
+      "loss": 0.7189,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997126165797167e-05,
+      "loss": 0.731,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997071192296005e-05,
+      "loss": 0.7034,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997015698742112e-05,
+      "loss": 0.7008,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996959685164433e-05,
+      "loss": 0.7381,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969031515921825e-05,
+      "loss": 0.7312,
+      "step": 246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996846098054848e-05,
+      "loss": 0.7025,
+      "step": 247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996788524582187e-05,
+      "loss": 0.7071,
+      "step": 248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967304312042298e-05,
+      "loss": 0.7597,
+      "step": 249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966718179512755e-05,
+      "loss": 0.7052,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996612684853896e-05,
+      "loss": 0.6402,
+      "step": 251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965530319429348e-05,
+      "loss": 0.7668,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.7014,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964321668049904e-05,
+      "loss": 0.7093,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996370954641049e-05,
+      "loss": 0.6369,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963092227896065e-05,
+      "loss": 0.646,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962469712828613e-05,
+      "loss": 0.6903,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996184200153283e-05,
+      "loss": 0.6798,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961209094336115e-05,
+      "loss": 0.6576,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996057099156858e-05,
+      "loss": 0.7395,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959927693563044e-05,
+      "loss": 0.706,
+      "step": 261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9959279200655044e-05,
+      "loss": 0.7418,
+      "step": 262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958625513182816e-05,
+      "loss": 0.7403,
+      "step": 263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957966631487316e-05,
+      "loss": 0.668,
+      "step": 264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99573025559122e-05,
+      "loss": 0.7065,
+      "step": 265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956633286803836e-05,
+      "loss": 0.6676,
+      "step": 266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955958824511302e-05,
+      "loss": 0.6778,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995527916938639e-05,
+      "loss": 0.6709,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954594321783585e-05,
+      "loss": 0.7032,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953904282060098e-05,
+      "loss": 0.6349,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953209050575838e-05,
+      "loss": 0.6434,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995250862769342e-05,
+      "loss": 0.6202,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995180301377818e-05,
+      "loss": 0.7281,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951092209198138e-05,
+      "loss": 0.7481,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995037621432405e-05,
+      "loss": 0.6566,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994965502952936e-05,
+      "loss": 0.7479,
+      "step": 276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994892865519022e-05,
+      "loss": 0.7607,
+      "step": 277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948197091685497e-05,
+      "loss": 0.6987,
+      "step": 278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947460339396756e-05,
+      "loss": 0.6059,
+      "step": 279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994671839870828e-05,
+      "loss": 0.6763,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994597127000704e-05,
+      "loss": 0.6604,
+      "step": 281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.6989,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994446145012775e-05,
+      "loss": 0.6357,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943698759737188e-05,
+      "loss": 0.7193,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942930882908852e-05,
+      "loss": 0.7149,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942157820043248e-05,
+      "loss": 0.7234,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941379571543597e-05,
+      "loss": 0.7063,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940596137815814e-05,
+      "loss": 0.676,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939807519268525e-05,
+      "loss": 0.7272,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939013716313055e-05,
+      "loss": 0.6595,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938214729363442e-05,
+      "loss": 0.7142,
+      "step": 291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9937410558836415e-05,
+      "loss": 0.6702,
+      "step": 292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936601205151415e-05,
+      "loss": 0.6794,
+      "step": 293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935786668730586e-05,
+      "loss": 0.6781,
+      "step": 294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934966949998772e-05,
+      "loss": 0.6048,
+      "step": 295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934142049383526e-05,
+      "loss": 0.6869,
+      "step": 296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99333119673151e-05,
+      "loss": 0.7044,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932476704226445e-05,
+      "loss": 0.7389,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931636260553226e-05,
+      "loss": 0.6903,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930790636733784e-05,
+      "loss": 0.6667,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99299398332092e-05,
+      "loss": 0.6226,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929083850423226e-05,
+      "loss": 0.7338,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928222688822324e-05,
+      "loss": 0.7133,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927356348855667e-05,
+      "loss": 0.6819,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926484830975116e-05,
+      "loss": 0.6674,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925608135635236e-05,
+      "loss": 0.5736,
+      "step": 306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9924726263293297e-05,
+      "loss": 0.6411,
+      "step": 307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992383921440926e-05,
+      "loss": 0.6811,
+      "step": 308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922946989445804e-05,
+      "loss": 0.7242,
+      "step": 309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992204958886829e-05,
+      "loss": 0.6827,
+      "step": 310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.6628,
+      "step": 311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920239262746045e-05,
+      "loss": 0.6613,
+      "step": 312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991932633814555e-05,
+      "loss": 0.7242,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918408239819453e-05,
+      "loss": 0.6441,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917484968246625e-05,
+      "loss": 0.6524,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991655652390862e-05,
+      "loss": 0.6662,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915622907289695e-05,
+      "loss": 0.6825,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914684118876812e-05,
+      "loss": 0.6802,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991374015915962e-05,
+      "loss": 0.6379,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991279102863047e-05,
+      "loss": 0.6893,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911836727784413e-05,
+      "loss": 0.6615,
+      "step": 321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.991087725711919e-05,
+      "loss": 0.7623,
+      "step": 322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909912617135246e-05,
+      "loss": 0.621,
+      "step": 323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908942808335712e-05,
+      "loss": 0.6315,
+      "step": 324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907967831226428e-05,
+      "loss": 0.7532,
+      "step": 325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990698768631592e-05,
+      "loss": 0.7045,
+      "step": 326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990600237411541e-05,
+      "loss": 0.6851,
+      "step": 327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990501189513882e-05,
+      "loss": 0.6764,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904016249902765e-05,
+      "loss": 0.607,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903015438926554e-05,
+      "loss": 0.6184,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990200946273219e-05,
+      "loss": 0.6733,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900998321844367e-05,
+      "loss": 0.6945,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989998201679048e-05,
+      "loss": 0.6595,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898960548100615e-05,
+      "loss": 0.6883,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897933916307542e-05,
+      "loss": 0.6332,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896902121946745e-05,
+      "loss": 0.6242,
+      "step": 336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9895865165556375e-05,
+      "loss": 0.724,
+      "step": 337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894823047677297e-05,
+      "loss": 0.6787,
+      "step": 338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989377576885305e-05,
+      "loss": 0.6722,
+      "step": 339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 0.6778,
+      "step": 340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891665730556727e-05,
+      "loss": 0.6498,
+      "step": 341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98906029721852e-05,
+      "loss": 0.6346,
+      "step": 342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988953505506962e-05,
+      "loss": 0.6725,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888461979766985e-05,
+      "loss": 0.7346,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887383746836995e-05,
+      "loss": 0.6821,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886300356842038e-05,
+      "loss": 0.716,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885211810347185e-05,
+      "loss": 0.6883,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884118107920198e-05,
+      "loss": 0.6845,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883019250131538e-05,
+      "loss": 0.676,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988191523755434e-05,
+      "loss": 0.5914,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988080607076444e-05,
+      "loss": 0.6595,
+      "step": 351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9879691750340356e-05,
+      "loss": 0.7262,
+      "step": 352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878572276863296e-05,
+      "loss": 0.6605,
+      "step": 353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987744765091715e-05,
+      "loss": 0.6747,
+      "step": 354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876317873088512e-05,
+      "loss": 0.6224,
+      "step": 355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875182943966645e-05,
+      "loss": 0.6686,
+      "step": 356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987404286414351e-05,
+      "loss": 0.602,
+      "step": 357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987289763421374e-05,
+      "loss": 0.6667,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871747254774673e-05,
+      "loss": 0.6461,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870591726426322e-05,
+      "loss": 0.6183,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869431049771393e-05,
+      "loss": 0.6315,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868265225415263e-05,
+      "loss": 0.6539,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986709425396601e-05,
+      "loss": 0.6032,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865918136034384e-05,
+      "loss": 0.6304,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986473687223383e-05,
+      "loss": 0.5862,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863550463180472e-05,
+      "loss": 0.6756,
+      "step": 366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9862358909493113e-05,
+      "loss": 0.6429,
+      "step": 367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9861162211793248e-05,
+      "loss": 0.6363,
+      "step": 368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.5656,
+      "step": 369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858753386855375e-05,
+      "loss": 0.6833,
+      "step": 370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857541260873764e-05,
+      "loss": 0.5847,
+      "step": 371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856323993392436e-05,
+      "loss": 0.6575,
+      "step": 372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855101585046292e-05,
+      "loss": 0.622,
+      "step": 373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853874036472922e-05,
+      "loss": 0.6262,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852641348312588e-05,
+      "loss": 0.6957,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851403521208232e-05,
+      "loss": 0.6536,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850160555805485e-05,
+      "loss": 0.6665,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848912452752655e-05,
+      "loss": 0.6638,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984765921270072e-05,
+      "loss": 0.6285,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846400836303354e-05,
+      "loss": 0.6431,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98451373242169e-05,
+      "loss": 0.6605,
+      "step": 381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9843868677100375e-05,
+      "loss": 0.7103,
+      "step": 382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842594895615487e-05,
+      "loss": 0.7113,
+      "step": 383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841315980426613e-05,
+      "loss": 0.6411,
+      "step": 384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984003193220081e-05,
+      "loss": 0.7547,
+      "step": 385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983874275160781e-05,
+      "loss": 0.6625,
+      "step": 386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837448439320027e-05,
+      "loss": 0.6256,
+      "step": 387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983614899601255e-05,
+      "loss": 0.6702,
+      "step": 388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834844422363142e-05,
+      "loss": 0.685,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983353471905224e-05,
+      "loss": 0.5611,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832219886762965e-05,
+      "loss": 0.5957,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830899926181103e-05,
+      "loss": 0.6221,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982957483799512e-05,
+      "loss": 0.6634,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828244622896155e-05,
+      "loss": 0.6607,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826909281578027e-05,
+      "loss": 0.7068,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982556881473722e-05,
+      "loss": 0.7104,
+      "step": 396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9824223223072895e-05,
+      "loss": 0.5494,
+      "step": 397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 0.7206,
+      "step": 398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821516668083705e-05,
+      "loss": 0.591,
+      "step": 399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820155706170525e-05,
+      "loss": 0.6321,
+      "step": 400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818789622257197e-05,
+      "loss": 0.5947,
+      "step": 401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981741841705625e-05,
+      "loss": 0.6405,
+      "step": 402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816042091282867e-05,
+      "loss": 0.613,
+      "step": 403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814660645654926e-05,
+      "loss": 0.6506,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981327408089295e-05,
+      "loss": 0.688,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811882397720146e-05,
+      "loss": 0.6535,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981048559686239e-05,
+      "loss": 0.5725,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809083679048235e-05,
+      "loss": 0.6654,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980767664500888e-05,
+      "loss": 0.655,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806264495478213e-05,
+      "loss": 0.5733,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980484723119278e-05,
+      "loss": 0.6527,
+      "step": 411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9803424852891803e-05,
+      "loss": 0.6133,
+      "step": 412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801997361317163e-05,
+      "loss": 0.5926,
+      "step": 413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800564757213414e-05,
+      "loss": 0.6688,
+      "step": 414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799127041327773e-05,
+      "loss": 0.6479,
+      "step": 415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797684214410122e-05,
+      "loss": 0.6025,
+      "step": 416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796236277213013e-05,
+      "loss": 0.6286,
+      "step": 417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794783230491664e-05,
+      "loss": 0.5776,
+      "step": 418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793325075003952e-05,
+      "loss": 0.6789,
+      "step": 419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791861811510424e-05,
+      "loss": 0.6103,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979039344077429e-05,
+      "loss": 0.6291,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788919963561423e-05,
+      "loss": 0.6079,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787441380640355e-05,
+      "loss": 0.6356,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978595769278229e-05,
+      "loss": 0.6718,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784468900761097e-05,
+      "loss": 0.5779,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782975005353287e-05,
+      "loss": 0.5929,
+      "step": 426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.6528,
+      "step": 427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779971907497253e-05,
+      "loss": 0.5635,
+      "step": 428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977846270661538e-05,
+      "loss": 0.601,
+      "step": 429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776948405479607e-05,
+      "loss": 0.6043,
+      "step": 430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977542900487977e-05,
+      "loss": 0.6107,
+      "step": 431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773904505608354e-05,
+      "loss": 0.5807,
+      "step": 432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772374908460508e-05,
+      "loss": 0.5601,
+      "step": 433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770840214234043e-05,
+      "loss": 0.5944,
+      "step": 434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769300423729422e-05,
+      "loss": 0.6523,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767755537749767e-05,
+      "loss": 0.6142,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 0.5985,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764650482591157e-05,
+      "loss": 0.6647,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763090315031736e-05,
+      "loss": 0.5986,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761525055236356e-05,
+      "loss": 0.6195,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975995470402142e-05,
+      "loss": 0.6403,
+      "step": 441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9758379262206e-05,
+      "loss": 0.5889,
+      "step": 442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756798730611814e-05,
+      "loss": 0.6561,
+      "step": 443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755213110063233e-05,
+      "loss": 0.6678,
+      "step": 444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975362240138728e-05,
+      "loss": 0.5903,
+      "step": 445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752026605413656e-05,
+      "loss": 0.5822,
+      "step": 446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975042572297468e-05,
+      "loss": 0.6389,
+      "step": 447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748819754905346e-05,
+      "loss": 0.5551,
+      "step": 448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747208702043297e-05,
+      "loss": 0.636,
+      "step": 449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745592565228827e-05,
+      "loss": 0.5824,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743971345304877e-05,
+      "loss": 0.6351,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742345043117045e-05,
+      "loss": 0.6119,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974071365951358e-05,
+      "loss": 0.6637,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739077195345377e-05,
+      "loss": 0.6622,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737435651465987e-05,
+      "loss": 0.6235,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735789028731603e-05,
+      "loss": 0.6258,
+      "step": 456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9734137328001077e-05,
+      "loss": 0.5809,
+      "step": 457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97324805501359e-05,
+      "loss": 0.654,
+      "step": 458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730818696000207e-05,
+      "loss": 0.6326,
+      "step": 459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97291517664608e-05,
+      "loss": 0.5767,
+      "step": 460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727479762387115e-05,
+      "loss": 0.559,
+      "step": 461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725802684651235e-05,
+      "loss": 0.5797,
+      "step": 462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724120534127885e-05,
+      "loss": 0.6592,
+      "step": 463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972243331169445e-05,
+      "loss": 0.5684,
+      "step": 464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972074101823095e-05,
+      "loss": 0.6084,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719043654620048e-05,
+      "loss": 0.6253,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717341221747056e-05,
+      "loss": 0.6285,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715633720499935e-05,
+      "loss": 0.5733,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713921151769278e-05,
+      "loss": 0.6508,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712203516448333e-05,
+      "loss": 0.557,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710480815432983e-05,
+      "loss": 0.6576,
+      "step": 471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970875304962175e-05,
+      "loss": 0.6627,
+      "step": 472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9707020219915805e-05,
+      "loss": 0.5723,
+      "step": 473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705282327218965e-05,
+      "loss": 0.6319,
+      "step": 474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970353937243767e-05,
+      "loss": 0.6166,
+      "step": 475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701791356481016e-05,
+      "loss": 0.5944,
+      "step": 476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700038280260734e-05,
+      "loss": 0.5799,
+      "step": 477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698280144691194e-05,
+      "loss": 0.5614,
+      "step": 478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696516950689404e-05,
+      "loss": 0.5919,
+      "step": 479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694748699175016e-05,
+      "loss": 0.5941,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969297539107031e-05,
+      "loss": 0.7233,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691197027300208e-05,
+      "loss": 0.6093,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968941360879227e-05,
+      "loss": 0.6374,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687625136476692e-05,
+      "loss": 0.5824,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.5863,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684033034156594e-05,
+      "loss": 0.6092,
+      "step": 486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9682229406025635e-05,
+      "loss": 0.6233,
+      "step": 487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968042072783418e-05,
+      "loss": 0.5683,
+      "step": 488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967860700052559e-05,
+      "loss": 0.6174,
+      "step": 489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967678822504588e-05,
+      "loss": 0.5989,
+      "step": 490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674964402343684e-05,
+      "loss": 0.6299,
+      "step": 491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673135533370273e-05,
+      "loss": 0.5889,
+      "step": 492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671301619079547e-05,
+      "loss": 0.6445,
+      "step": 493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669462660428036e-05,
+      "loss": 0.5847,
+      "step": 494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667618658374912e-05,
+      "loss": 0.5803,
+      "step": 495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665769613881968e-05,
+      "loss": 0.6007,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663915527913628e-05,
+      "loss": 0.6459,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662056401436943e-05,
+      "loss": 0.5982,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660192235421607e-05,
+      "loss": 0.5748,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965832303083992e-05,
+      "loss": 0.6548,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656448788666833e-05,
+      "loss": 0.5967,
+      "step": 501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9654569509879907e-05,
+      "loss": 0.6696,
+      "step": 502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652685195459346e-05,
+      "loss": 0.561,
+      "step": 503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650795846387957e-05,
+      "loss": 0.5707,
+      "step": 504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96489014636512e-05,
+      "loss": 0.6022,
+      "step": 505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964700204823714e-05,
+      "loss": 0.6478,
+      "step": 506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645097601136475e-05,
+      "loss": 0.5631,
+      "step": 507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643188123342535e-05,
+      "loss": 0.6042,
+      "step": 508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641273615851256e-05,
+      "loss": 0.5444,
+      "step": 509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639354079661217e-05,
+      "loss": 0.6007,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637429515773603e-05,
+      "loss": 0.6184,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963549992519223e-05,
+      "loss": 0.6258,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963356530892354e-05,
+      "loss": 0.6244,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 0.5522,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629681003363043e-05,
+      "loss": 0.6317,
+      "step": 515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9627731316097215e-05,
+      "loss": 0.5273,
+      "step": 516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9625776607196024e-05,
+      "loss": 0.5826,
+      "step": 517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623816877679e-05,
+      "loss": 0.5606,
+      "step": 518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621852128568303e-05,
+      "loss": 0.5791,
+      "step": 519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619882360888706e-05,
+      "loss": 0.687,
+      "step": 520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617907575667602e-05,
+      "loss": 0.5663,
+      "step": 521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615927773935004e-05,
+      "loss": 0.5866,
+      "step": 522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961394295672353e-05,
+      "loss": 0.616,
+      "step": 523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611953125068432e-05,
+      "loss": 0.5947,
+      "step": 524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609958280007556e-05,
+      "loss": 0.6241,
+      "step": 525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960795842258138e-05,
+      "loss": 0.605,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605953553832987e-05,
+      "loss": 0.6688,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960394367480808e-05,
+      "loss": 0.592,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601928786554977e-05,
+      "loss": 0.585,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599908890124593e-05,
+      "loss": 0.5798,
+      "step": 530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9597883986570473e-05,
+      "loss": 0.6298,
+      "step": 531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9595854076948766e-05,
+      "loss": 0.5702,
+      "step": 532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593819162318232e-05,
+      "loss": 0.5944,
+      "step": 533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591779243740246e-05,
+      "loss": 0.5403,
+      "step": 534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589734322278784e-05,
+      "loss": 0.6301,
+      "step": 535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958768439900044e-05,
+      "loss": 0.549,
+      "step": 536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585629474974413e-05,
+      "loss": 0.5665,
+      "step": 537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583569551272514e-05,
+      "loss": 0.5198,
+      "step": 538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581504628969153e-05,
+      "loss": 0.5526,
+      "step": 539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579434709141358e-05,
+      "loss": 0.5552,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577359792868752e-05,
+      "loss": 0.5169,
+      "step": 541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575279881233578e-05,
+      "loss": 0.6,
+      "step": 542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.5262,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571105076217485e-05,
+      "loss": 0.5818,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569010185014062e-05,
+      "loss": 0.5599,
+      "step": 545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956691030280306e-05,
+      "loss": 0.5974,
+      "step": 546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9564805430679733e-05,
+      "loss": 0.5695,
+      "step": 547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562695569741948e-05,
+      "loss": 0.6435,
+      "step": 548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560580721090155e-05,
+      "loss": 0.567,
+      "step": 549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558460885827432e-05,
+      "loss": 0.5995,
+      "step": 550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955633606505943e-05,
+      "loss": 0.5779,
+      "step": 551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554206259894424e-05,
+      "loss": 0.623,
+      "step": 552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955207147144327e-05,
+      "loss": 0.5988,
+      "step": 553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549931700819443e-05,
+      "loss": 0.6213,
+      "step": 554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547786949138992e-05,
+      "loss": 0.6467,
+      "step": 555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545637217520588e-05,
+      "loss": 0.5766,
+      "step": 556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543482507085484e-05,
+      "loss": 0.5583,
+      "step": 557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954132281895753e-05,
+      "loss": 0.5981,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539158154263188e-05,
+      "loss": 0.6528,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536988514131496e-05,
+      "loss": 0.6399,
+      "step": 560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9534813899694097e-05,
+      "loss": 0.5909,
+      "step": 561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953263431208523e-05,
+      "loss": 0.5916,
+      "step": 562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953044975244172e-05,
+      "loss": 0.5582,
+      "step": 563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528260221902996e-05,
+      "loss": 0.5578,
+      "step": 564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526065721611075e-05,
+      "loss": 0.5932,
+      "step": 565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523866252710553e-05,
+      "loss": 0.5355,
+      "step": 566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521661816348647e-05,
+      "loss": 0.6225,
+      "step": 567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519452413675134e-05,
+      "loss": 0.568,
+      "step": 568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517238045842403e-05,
+      "loss": 0.6232,
+      "step": 569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951501871400542e-05,
+      "loss": 0.5902,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951279441932175e-05,
+      "loss": 0.5958,
+      "step": 571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.5861,
+      "step": 572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950833094605752e-05,
+      "loss": 0.6114,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506091769805018e-05,
+      "loss": 0.5312,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503847635361944e-05,
+      "loss": 0.5622,
+      "step": 575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9501598543898796e-05,
+      "loss": 0.6198,
+      "step": 576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9499344496588656e-05,
+      "loss": 0.6067,
+      "step": 577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497085494607192e-05,
+      "loss": 0.5419,
+      "step": 578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949482153913265e-05,
+      "loss": 0.6177,
+      "step": 579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492552631345862e-05,
+      "loss": 0.6144,
+      "step": 580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490278772430255e-05,
+      "loss": 0.6059,
+      "step": 581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487999963571827e-05,
+      "loss": 0.599,
+      "step": 582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485716205959154e-05,
+      "loss": 0.5071,
+      "step": 583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483427500783406e-05,
+      "loss": 0.6076,
+      "step": 584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481133849238322e-05,
+      "loss": 0.6125,
+      "step": 585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947883525252023e-05,
+      "loss": 0.5285,
+      "step": 586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476531711828027e-05,
+      "loss": 0.5776,
+      "step": 587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474223228363197e-05,
+      "loss": 0.6339,
+      "step": 588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94719098033298e-05,
+      "loss": 0.6039,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469591437934475e-05,
+      "loss": 0.577,
+      "step": 590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9467268133386434e-05,
+      "loss": 0.5916,
+      "step": 591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9464939890897464e-05,
+      "loss": 0.604,
+      "step": 592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462606711681936e-05,
+      "loss": 0.5388,
+      "step": 593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946026859695679e-05,
+      "loss": 0.6358,
+      "step": 594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457925547941535e-05,
+      "loss": 0.5625,
+      "step": 595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455577565858264e-05,
+      "loss": 0.606,
+      "step": 596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453224651931638e-05,
+      "loss": 0.5401,
+      "step": 597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450866807388885e-05,
+      "loss": 0.5308,
+      "step": 598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448504033459817e-05,
+      "loss": 0.6029,
+      "step": 599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944613633137681e-05,
+      "loss": 0.5599,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 0.5401,
+      "step": 601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441386147691336e-05,
+      "loss": 0.5445,
+      "step": 602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439003668566466e-05,
+      "loss": 0.6255,
+      "step": 603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436616266242862e-05,
+      "loss": 0.5342,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434223941965738e-05,
+      "loss": 0.615,
+      "step": 605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.943182669698289e-05,
+      "loss": 0.626,
+      "step": 606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9429424532544677e-05,
+      "loss": 0.5664,
+      "step": 607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427017449904012e-05,
+      "loss": 0.6242,
+      "step": 608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424605450316387e-05,
+      "loss": 0.5938,
+      "step": 609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422188535039853e-05,
+      "loss": 0.6452,
+      "step": 610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419766705335025e-05,
+      "loss": 0.6432,
+      "step": 611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417339962465084e-05,
+      "loss": 0.6148,
+      "step": 612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414908307695765e-05,
+      "loss": 0.5681,
+      "step": 613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412471742295377e-05,
+      "loss": 0.6064,
+      "step": 614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410030267534785e-05,
+      "loss": 0.5596,
+      "step": 615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940758388468741e-05,
+      "loss": 0.6049,
+      "step": 616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940513259502924e-05,
+      "loss": 0.5048,
+      "step": 617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940267639983882e-05,
+      "loss": 0.6766,
+      "step": 618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400215300397255e-05,
+      "loss": 0.6525,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397749297988196e-05,
+      "loss": 0.6234,
+      "step": 620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9395278393897876e-05,
+      "loss": 0.6219,
+      "step": 621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939280258941506e-05,
+      "loss": 0.5893,
+      "step": 622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390321885831082e-05,
+      "loss": 0.5978,
+      "step": 623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387836284439826e-05,
+      "loss": 0.5902,
+      "step": 624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385345786537734e-05,
+      "loss": 0.5896,
+      "step": 625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382850393423802e-05,
+      "loss": 0.6136,
+      "step": 626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380350106399578e-05,
+      "loss": 0.5789,
+      "step": 627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377844926769166e-05,
+      "loss": 0.6054,
+      "step": 628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937533485583921e-05,
+      "loss": 0.5421,
+      "step": 629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937281989491892e-05,
+      "loss": 0.6113,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370300045320044e-05,
+      "loss": 0.6541,
+      "step": 631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367775308356892e-05,
+      "loss": 0.6324,
+      "step": 632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365245685346316e-05,
+      "loss": 0.5553,
+      "step": 633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362711177607715e-05,
+      "loss": 0.6213,
+      "step": 634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360171786463043e-05,
+      "loss": 0.6051,
+      "step": 635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9357627513236795e-05,
+      "loss": 0.5442,
+      "step": 636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935507835925601e-05,
+      "loss": 0.5613,
+      "step": 637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352524325850282e-05,
+      "loss": 0.548,
+      "step": 638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934996541435174e-05,
+      "loss": 0.609,
+      "step": 639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347401626095065e-05,
+      "loss": 0.5609,
+      "step": 640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344832962417475e-05,
+      "loss": 0.6008,
+      "step": 641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934225942465874e-05,
+      "loss": 0.6595,
+      "step": 642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339681014161163e-05,
+      "loss": 0.5983,
+      "step": 643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933709773226959e-05,
+      "loss": 0.5523,
+      "step": 644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334509580331417e-05,
+      "loss": 0.5478,
+      "step": 645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933191655969657e-05,
+      "loss": 0.5272,
+      "step": 646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932931867171751e-05,
+      "loss": 0.5542,
+      "step": 647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326715917749258e-05,
+      "loss": 0.5455,
+      "step": 648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324108299149345e-05,
+      "loss": 0.609,
+      "step": 649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321495817277867e-05,
+      "loss": 0.5859,
+      "step": 650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9318878473497437e-05,
+      "loss": 0.522,
+      "step": 651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9316256269173203e-05,
+      "loss": 0.5909,
+      "step": 652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931362920567287e-05,
+      "loss": 0.6404,
+      "step": 653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931099728436665e-05,
+      "loss": 0.6099,
+      "step": 654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930836050662731e-05,
+      "loss": 0.6027,
+      "step": 655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305718873830135e-05,
+      "loss": 0.5922,
+      "step": 656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303072387352952e-05,
+      "loss": 0.569,
+      "step": 657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930042104857612e-05,
+      "loss": 0.5497,
+      "step": 658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.6086,
+      "step": 659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929510381965756e-05,
+      "loss": 0.5708,
+      "step": 660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292437932289205e-05,
+      "loss": 0.5324,
+      "step": 661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289767198167918e-05,
+      "loss": 0.5994,
+      "step": 662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92870916186867e-05,
+      "loss": 0.572,
+      "step": 663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284411195241086e-05,
+      "loss": 0.5789,
+      "step": 664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281725929229127e-05,
+      "loss": 0.523,
+      "step": 665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927903582205141e-05,
+      "loss": 0.5693,
+      "step": 666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927634087511104e-05,
+      "loss": 0.5736,
+      "step": 667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273641089813644e-05,
+      "loss": 0.5449,
+      "step": 668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270936467567388e-05,
+      "loss": 0.5466,
+      "step": 669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926822700978294e-05,
+      "loss": 0.6406,
+      "step": 670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92655127178735e-05,
+      "loss": 0.5474,
+      "step": 671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262793593254794e-05,
+      "loss": 0.6549,
+      "step": 672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926006963734506e-05,
+      "loss": 0.6117,
+      "step": 673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257340851565064e-05,
+      "loss": 0.5845,
+      "step": 674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254607237338082e-05,
+      "loss": 0.5648,
+      "step": 675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925186879608992e-05,
+      "loss": 0.5757,
+      "step": 676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924912552924889e-05,
+      "loss": 0.5996,
+      "step": 677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246377438245827e-05,
+      "loss": 0.64,
+      "step": 678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243624524514082e-05,
+      "loss": 0.5727,
+      "step": 679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924086678948952e-05,
+      "loss": 0.5202,
+      "step": 680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923810423461052e-05,
+      "loss": 0.576,
+      "step": 681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9235336861317982e-05,
+      "loss": 0.5999,
+      "step": 682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232564671055308e-05,
+      "loss": 0.5506,
+      "step": 683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922978766526842e-05,
+      "loss": 0.5816,
+      "step": 684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227005845405747e-05,
+      "loss": 0.5413,
+      "step": 685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224219212918235e-05,
+      "loss": 0.5881,
+      "step": 686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221427769259333e-05,
+      "loss": 0.5524,
+      "step": 687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218631515885007e-05,
+      "loss": 0.5933,
+      "step": 688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215830454253726e-05,
+      "loss": 0.5059,
+      "step": 689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213024585826468e-05,
+      "loss": 0.6101,
+      "step": 690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921021391206672e-05,
+      "loss": 0.6029,
+      "step": 691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207398434440478e-05,
+      "loss": 0.5278,
+      "step": 692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204578154416233e-05,
+      "loss": 0.582,
+      "step": 693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920175307346499e-05,
+      "loss": 0.6084,
+      "step": 694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198923193060255e-05,
+      "loss": 0.5245,
+      "step": 695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919608851467804e-05,
+      "loss": 0.5585,
+      "step": 696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9193249039796856e-05,
+      "loss": 0.6117,
+      "step": 697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919040476989772e-05,
+      "loss": 0.4741,
+      "step": 698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918755570646414e-05,
+      "loss": 0.5724,
+      "step": 699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918470185098214e-05,
+      "loss": 0.5711,
+      "step": 700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181843204940232e-05,
+      "loss": 0.6131,
+      "step": 701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178979769829428e-05,
+      "loss": 0.5036,
+      "step": 702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176111547143237e-05,
+      "loss": 0.5747,
+      "step": 703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173238538377674e-05,
+      "loss": 0.5104,
+      "step": 704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170360745031243e-05,
+      "loss": 0.5714,
+      "step": 705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916747816860494e-05,
+      "loss": 0.6763,
+      "step": 706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164590810602262e-05,
+      "loss": 0.4811,
+      "step": 707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91616986725292e-05,
+      "loss": 0.6092,
+      "step": 708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915880175589423e-05,
+      "loss": 0.4965,
+      "step": 709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915590006220834e-05,
+      "loss": 0.574,
+      "step": 710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9152993592984983e-05,
+      "loss": 0.5603,
+      "step": 711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9150082349740123e-05,
+      "loss": 0.6511,
+      "step": 712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147166333992205e-05,
+      "loss": 0.5892,
+      "step": 713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144245547262165e-05,
+      "loss": 0.5803,
+      "step": 714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9141319991073432e-05,
+      "loss": 0.5583,
+      "step": 715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138389666951913e-05,
+      "loss": 0.5451,
+      "step": 716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.621,
+      "step": 717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913251472102661e-05,
+      "loss": 0.5119,
+      "step": 718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129570102287084e-05,
+      "loss": 0.6377,
+      "step": 719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912662072174328e-05,
+      "loss": 0.4837,
+      "step": 720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123666580933547e-05,
+      "loss": 0.496,
+      "step": 721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91207076813987e-05,
+      "loss": 0.5659,
+      "step": 722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117744024682044e-05,
+      "loss": 0.5809,
+      "step": 723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114775612329367e-05,
+      "loss": 0.6379,
+      "step": 724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111802445888936e-05,
+      "loss": 0.4747,
+      "step": 725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9108824526911486e-05,
+      "loss": 0.5322,
+      "step": 726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9105841856950254e-05,
+      "loss": 0.5537,
+      "step": 727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102854437560934e-05,
+      "loss": 0.5265,
+      "step": 728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099862270301704e-05,
+      "loss": 0.5472,
+      "step": 729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096865356733226e-05,
+      "loss": 0.5752,
+      "step": 730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093863698418627e-05,
+      "loss": 0.5147,
+      "step": 731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090857296923513e-05,
+      "loss": 0.5501,
+      "step": 732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087846153815965e-05,
+      "loss": 0.5562,
+      "step": 733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908483027066654e-05,
+      "loss": 0.6322,
+      "step": 734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081809649048254e-05,
+      "loss": 0.554,
+      "step": 735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078784290536608e-05,
+      "loss": 0.5837,
+      "step": 736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 0.5571,
+      "step": 737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072719369147583e-05,
+      "loss": 0.534,
+      "step": 738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069679809433545e-05,
+      "loss": 0.648,
+      "step": 739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066635519152833e-05,
+      "loss": 0.6422,
+      "step": 740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9063586499893292e-05,
+      "loss": 0.5471,
+      "step": 741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.906053275324523e-05,
+      "loss": 0.551,
+      "step": 742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057474280801414e-05,
+      "loss": 0.6386,
+      "step": 743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054411084157092e-05,
+      "loss": 0.5634,
+      "step": 744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051343164909965e-05,
+      "loss": 0.4709,
+      "step": 745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048270524660197e-05,
+      "loss": 0.555,
+      "step": 746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045193165010417e-05,
+      "loss": 0.575,
+      "step": 747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042111087565716e-05,
+      "loss": 0.5876,
+      "step": 748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039024293933647e-05,
+      "loss": 0.5012,
+      "step": 749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035932785724217e-05,
+      "loss": 0.6286,
+      "step": 750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90328365645499e-05,
+      "loss": 0.4772,
+      "step": 751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029735632025618e-05,
+      "loss": 0.5343,
+      "step": 752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026629989768764e-05,
+      "loss": 0.5987,
+      "step": 753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902351963939918e-05,
+      "loss": 0.571,
+      "step": 754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020404582539153e-05,
+      "loss": 0.5727,
+      "step": 755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9017284820813443e-05,
+      "loss": 0.548,
+      "step": 756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901416035584926e-05,
+      "loss": 0.5263,
+      "step": 757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011031189276257e-05,
+      "loss": 0.544,
+      "step": 758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007897322726546e-05,
+      "loss": 0.6046,
+      "step": 759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004758757834696e-05,
+      "loss": 0.594,
+      "step": 760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001615496237714e-05,
+      "loss": 0.5283,
+      "step": 761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899846753957507e-05,
+      "loss": 0.5184,
+      "step": 762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899531488948867e-05,
+      "loss": 0.627,
+      "step": 763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992157547622877e-05,
+      "loss": 0.5208,
+      "step": 764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988995515624497e-05,
+      "loss": 0.552,
+      "step": 765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985828795142788e-05,
+      "loss": 0.4764,
+      "step": 766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982657387829447e-05,
+      "loss": 0.5035,
+      "step": 767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979481295338615e-05,
+      "loss": 0.4862,
+      "step": 768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897630051932688e-05,
+      "loss": 0.5612,
+      "step": 769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973115061453274e-05,
+      "loss": 0.6124,
+      "step": 770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896992492337927e-05,
+      "loss": 0.5942,
+      "step": 771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8966730106768782e-05,
+      "loss": 0.5641,
+      "step": 772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896353061328816e-05,
+      "loss": 0.5422,
+      "step": 773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89603264446062e-05,
+      "loss": 0.5793,
+      "step": 774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.573,
+      "step": 775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895390408832562e-05,
+      "loss": 0.5789,
+      "step": 776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895068590407678e-05,
+      "loss": 0.5944,
+      "step": 777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947463051326148e-05,
+      "loss": 0.5607,
+      "step": 778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944235531754697e-05,
+      "loss": 0.5379,
+      "step": 779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941003347045846e-05,
+      "loss": 0.5716,
+      "step": 780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937766498885435e-05,
+      "loss": 0.4987,
+      "step": 781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893452498896174e-05,
+      "loss": 0.551,
+      "step": 782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931278818965463e-05,
+      "loss": 0.5254,
+      "step": 783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928027990589758e-05,
+      "loss": 0.541,
+      "step": 784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924772505530177e-05,
+      "loss": 0.5103,
+      "step": 785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8921512365484723e-05,
+      "loss": 0.5345,
+      "step": 786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8918247572153822e-05,
+      "loss": 0.4673,
+      "step": 787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914978127240328e-05,
+      "loss": 0.6107,
+      "step": 788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911704032449515e-05,
+      "loss": 0.5359,
+      "step": 789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890842528948909e-05,
+      "loss": 0.5766,
+      "step": 790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890514190006918e-05,
+      "loss": 0.5605,
+      "step": 791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890185386590233e-05,
+      "loss": 0.5203,
+      "step": 792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898561188703527e-05,
+      "loss": 0.5466,
+      "step": 793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895263870190154e-05,
+      "loss": 0.4669,
+      "step": 794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891961912082033e-05,
+      "loss": 0.5236,
+      "step": 795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888655316101404e-05,
+      "loss": 0.5283,
+      "step": 796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885344083972912e-05,
+      "loss": 0.5531,
+      "step": 797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882028217423644e-05,
+      "loss": 0.5077,
+      "step": 798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887870771818308e-05,
+      "loss": 0.4715,
+      "step": 799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887538258798313e-05,
+      "loss": 0.5379,
+      "step": 800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.887205282855812e-05,
+      "loss": 0.5888,
+      "step": 801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886871844164478e-05,
+      "loss": 0.4808,
+      "step": 802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865379428982272e-05,
+      "loss": 0.4934,
+      "step": 803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862035792312148e-05,
+      "loss": 0.53,
+      "step": 804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885868753337839e-05,
+      "loss": 0.4676,
+      "step": 805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885533465392738e-05,
+      "loss": 0.5289,
+      "step": 806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851977155707917e-05,
+      "loss": 0.51,
+      "step": 807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848615040471206e-05,
+      "loss": 0.5956,
+      "step": 808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845248309970855e-05,
+      "loss": 0.5374,
+      "step": 809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841876965962893e-05,
+      "loss": 0.5809,
+      "step": 810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883850101020574e-05,
+      "loss": 0.4818,
+      "step": 811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883512044446023e-05,
+      "loss": 0.5596,
+      "step": 812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88317352704896e-05,
+      "loss": 0.5077,
+      "step": 813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882834549005949e-05,
+      "loss": 0.5324,
+      "step": 814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882495110493795e-05,
+      "loss": 0.5525,
+      "step": 815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8821552116895407e-05,
+      "loss": 0.5393,
+      "step": 816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881814852770472e-05,
+      "loss": 0.5838,
+      "step": 817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814740339141134e-05,
+      "loss": 0.4872,
+      "step": 818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881132755298229e-05,
+      "loss": 0.4798,
+      "step": 819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880791017100823e-05,
+      "loss": 0.4853,
+      "step": 820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804488195001394e-05,
+      "loss": 0.5677,
+      "step": 821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801061626746615e-05,
+      "loss": 0.5595,
+      "step": 822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797630468031128e-05,
+      "loss": 0.5429,
+      "step": 823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794194720644553e-05,
+      "loss": 0.5775,
+      "step": 824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790754386378914e-05,
+      "loss": 0.4769,
+      "step": 825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787309467028617e-05,
+      "loss": 0.5477,
+      "step": 826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783859964390466e-05,
+      "loss": 0.5355,
+      "step": 827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780405880263653e-05,
+      "loss": 0.4691,
+      "step": 828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877694721644976e-05,
+      "loss": 0.5518,
+      "step": 829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773483974752765e-05,
+      "loss": 0.5204,
+      "step": 830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8770016156979016e-05,
+      "loss": 0.5027,
+      "step": 831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766543764937262e-05,
+      "loss": 0.5545,
+      "step": 832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.5772,
+      "step": 833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759585265296657e-05,
+      "loss": 0.4955,
+      "step": 834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756099161327222e-05,
+      "loss": 0.5036,
+      "step": 835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752608490348612e-05,
+      "loss": 0.6134,
+      "step": 836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749113254181498e-05,
+      "loss": 0.5226,
+      "step": 837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745613454648923e-05,
+      "loss": 0.5865,
+      "step": 838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8742109093576313e-05,
+      "loss": 0.5406,
+      "step": 839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738600172791476e-05,
+      "loss": 0.5082,
+      "step": 840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735086694124596e-05,
+      "loss": 0.5321,
+      "step": 841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873156865940823e-05,
+      "loss": 0.526,
+      "step": 842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728046070477322e-05,
+      "loss": 0.5619,
+      "step": 843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724518929169183e-05,
+      "loss": 0.4384,
+      "step": 844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720987237323497e-05,
+      "loss": 0.5492,
+      "step": 845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8717450996782333e-05,
+      "loss": 0.5013,
+      "step": 846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713910209390113e-05,
+      "loss": 0.5067,
+      "step": 847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710364876993652e-05,
+      "loss": 0.5201,
+      "step": 848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706815001442124e-05,
+      "loss": 0.5636,
+      "step": 849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703260584587068e-05,
+      "loss": 0.5838,
+      "step": 850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869970162828241e-05,
+      "loss": 0.5239,
+      "step": 851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696138134384412e-05,
+      "loss": 0.5928,
+      "step": 852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692570104751744e-05,
+      "loss": 0.5425,
+      "step": 853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688997541245412e-05,
+      "loss": 0.5367,
+      "step": 854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685420445728785e-05,
+      "loss": 0.5331,
+      "step": 855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868183882006762e-05,
+      "loss": 0.4774,
+      "step": 856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678252666130016e-05,
+      "loss": 0.4975,
+      "step": 857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674661985786444e-05,
+      "loss": 0.5251,
+      "step": 858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671066780909722e-05,
+      "loss": 0.4946,
+      "step": 859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866746705337505e-05,
+      "loss": 0.5596,
+      "step": 860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866386280505997e-05,
+      "loss": 0.4831,
+      "step": 861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.5278,
+      "step": 862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656640753610563e-05,
+      "loss": 0.5354,
+      "step": 863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865302295424312e-05,
+      "loss": 0.5805,
+      "step": 864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649400641629028e-05,
+      "loss": 0.499,
+      "step": 865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864577381765761e-05,
+      "loss": 0.5085,
+      "step": 866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864214248422055e-05,
+      "loss": 0.4881,
+      "step": 867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638506643211884e-05,
+      "loss": 0.5318,
+      "step": 868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863486629652799e-05,
+      "loss": 0.5429,
+      "step": 869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86312214460676e-05,
+      "loss": 0.4311,
+      "step": 870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86275720937318e-05,
+      "loss": 0.5389,
+      "step": 871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623918241424023e-05,
+      "loss": 0.5126,
+      "step": 872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620259891050037e-05,
+      "loss": 0.5428,
+      "step": 873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616597044517974e-05,
+      "loss": 0.5052,
+      "step": 874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8612929703738302e-05,
+      "loss": 0.4786,
+      "step": 875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860925787062383e-05,
+      "loss": 0.5752,
+      "step": 876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605581547089715e-05,
+      "loss": 0.4574,
+      "step": 877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860190073505346e-05,
+      "loss": 0.5135,
+      "step": 878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598215436434894e-05,
+      "loss": 0.5314,
+      "step": 879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594525653156204e-05,
+      "loss": 0.5415,
+      "step": 880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859083138714191e-05,
+      "loss": 0.48,
+      "step": 881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587132640318864e-05,
+      "loss": 0.5344,
+      "step": 882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858342941461626e-05,
+      "loss": 0.5887,
+      "step": 883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857972171196563e-05,
+      "loss": 0.5399,
+      "step": 884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857600953430084e-05,
+      "loss": 0.4859,
+      "step": 885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572292883558082e-05,
+      "loss": 0.4738,
+      "step": 886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 0.5105,
+      "step": 887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564846170595137e-05,
+      "loss": 0.4666,
+      "step": 888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856111611225901e-05,
+      "loss": 0.5907,
+      "step": 889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8557381588613035e-05,
+      "loss": 0.5379,
+      "step": 890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 0.4892,
+      "step": 891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854989915318529e-05,
+      "loss": 0.5666,
+      "step": 892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546151245306205e-05,
+      "loss": 0.4822,
+      "step": 893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854239887992266e-05,
+      "loss": 0.518,
+      "step": 894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538642058991803e-05,
+      "loss": 0.4892,
+      "step": 895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534880784473123e-05,
+      "loss": 0.5099,
+      "step": 896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853111505832843e-05,
+      "loss": 0.5785,
+      "step": 897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527344882521847e-05,
+      "loss": 0.4763,
+      "step": 898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523570259019827e-05,
+      "loss": 0.4441,
+      "step": 899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851979118979114e-05,
+      "loss": 0.5831,
+      "step": 900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851600767680687e-05,
+      "loss": 0.4924,
+      "step": 901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512219722040426e-05,
+      "loss": 0.6355,
+      "step": 902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850842732746753e-05,
+      "loss": 0.573,
+      "step": 903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504630495066222e-05,
+      "loss": 0.5318,
+      "step": 904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8500829226816853e-05,
+      "loss": 0.5177,
+      "step": 905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8497023524702093e-05,
+      "loss": 0.5328,
+      "step": 906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849321339070692e-05,
+      "loss": 0.4467,
+      "step": 907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489398826818622e-05,
+      "loss": 0.5933,
+      "step": 908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84855798350268e-05,
+      "loss": 0.5105,
+      "step": 909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848175641732337e-05,
+      "loss": 0.5797,
+      "step": 910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847792857570255e-05,
+      "loss": 0.489,
+      "step": 911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474096312160866e-05,
+      "loss": 0.5017,
+      "step": 912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847025962869715e-05,
+      "loss": 0.5762,
+      "step": 913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466418527312547e-05,
+      "loss": 0.4809,
+      "step": 914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846257301001049e-05,
+      "loss": 0.5448,
+      "step": 915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458723078796734e-05,
+      "loss": 0.5628,
+      "step": 916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845486873567932e-05,
+      "loss": 0.5831,
+      "step": 917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8451009982668605e-05,
+      "loss": 0.4836,
+      "step": 918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844714682177723e-05,
+      "loss": 0.5369,
+      "step": 919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8443279255020153e-05,
+      "loss": 0.47,
+      "step": 920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8439407284414613e-05,
+      "loss": 0.603,
+      "step": 921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435530911980163e-05,
+      "loss": 0.5682,
+      "step": 922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431650139738632e-05,
+      "loss": 0.5293,
+      "step": 923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427764969714162e-05,
+      "loss": 0.5993,
+      "step": 924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423875403933177e-05,
+      "loss": 0.5784,
+      "step": 925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419981444424402e-05,
+      "loss": 0.5147,
+      "step": 926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416083093218847e-05,
+      "loss": 0.4697,
+      "step": 927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412180352349814e-05,
+      "loss": 0.5449,
+      "step": 928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84082732238529e-05,
+      "loss": 0.4791,
+      "step": 929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404361709765982e-05,
+      "loss": 0.5255,
+      "step": 930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400445812129234e-05,
+      "loss": 0.543,
+      "step": 931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839652553298511e-05,
+      "loss": 0.582,
+      "step": 932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392600874378343e-05,
+      "loss": 0.4913,
+      "step": 933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388671838355965e-05,
+      "loss": 0.5663,
+      "step": 934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8384738426967283e-05,
+      "loss": 0.494,
+      "step": 935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.838080064226389e-05,
+      "loss": 0.5378,
+      "step": 936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837685848629965e-05,
+      "loss": 0.4525,
+      "step": 937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372911961130716e-05,
+      "loss": 0.5288,
+      "step": 938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8368961068815525e-05,
+      "loss": 0.4976,
+      "step": 939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365005811414776e-05,
+      "loss": 0.5301,
+      "step": 940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361046190991457e-05,
+      "loss": 0.4359,
+      "step": 941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357082209610827e-05,
+      "loss": 0.5087,
+      "step": 942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353113869340425e-05,
+      "loss": 0.5016,
+      "step": 943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349141172250054e-05,
+      "loss": 0.589,
+      "step": 944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8345164120411798e-05,
+      "loss": 0.517,
+      "step": 945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341182715900006e-05,
+      "loss": 0.5551,
+      "step": 946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337196960791305e-05,
+      "loss": 0.4849,
+      "step": 947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333206857164582e-05,
+      "loss": 0.4946,
+      "step": 948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 0.4975,
+      "step": 949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.832521361268398e-05,
+      "loss": 0.4496,
+      "step": 950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.832121047599922e-05,
+      "loss": 0.5539,
+      "step": 951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317202999134674e-05,
+      "loss": 0.4884,
+      "step": 952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313191184180567e-05,
+      "loss": 0.4888,
+      "step": 953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830917503322938e-05,
+      "loss": 0.5406,
+      "step": 954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305154548375862e-05,
+      "loss": 0.5424,
+      "step": 955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301129731717014e-05,
+      "loss": 0.5055,
+      "step": 956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82971005853521e-05,
+      "loss": 0.4923,
+      "step": 957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293067111382653e-05,
+      "loss": 0.4749,
+      "step": 958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289029311912442e-05,
+      "loss": 0.5265,
+      "step": 959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828498718904751e-05,
+      "loss": 0.5736,
+      "step": 960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280940744896152e-05,
+      "loss": 0.4949,
+      "step": 961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827688998156891e-05,
+      "loss": 0.5602,
+      "step": 962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827283490117858e-05,
+      "loss": 0.5218,
+      "step": 963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268775505840214e-05,
+      "loss": 0.4965,
+      "step": 964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.826471179767111e-05,
+      "loss": 0.5572,
+      "step": 965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8260643778790827e-05,
+      "loss": 0.5081,
+      "step": 966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8256571451321155e-05,
+      "loss": 0.4599,
+      "step": 967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252494817386142e-05,
+      "loss": 0.5147,
+      "step": 968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824841387911209e-05,
+      "loss": 0.5401,
+      "step": 969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244328638627518e-05,
+      "loss": 0.5232,
+      "step": 970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824023909806322e-05,
+      "loss": 0.4589,
+      "step": 971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823614525955222e-05,
+      "loss": 0.5183,
+      "step": 972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232047125229775e-05,
+      "loss": 0.4819,
+      "step": 973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227944697233398e-05,
+      "loss": 0.487,
+      "step": 974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223837977702834e-05,
+      "loss": 0.5816,
+      "step": 975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219726968780068e-05,
+      "loss": 0.6032,
+      "step": 976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215611672609316e-05,
+      "loss": 0.4628,
+      "step": 977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821149209133704e-05,
+      "loss": 0.565,
+      "step": 978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207368227111936e-05,
+      "loss": 0.5232,
+      "step": 979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8203240082084924e-05,
+      "loss": 0.5298,
+      "step": 980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8199107658409168e-05,
+      "loss": 0.594,
+      "step": 981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8194970958240054e-05,
+      "loss": 0.4451,
+      "step": 982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190829983735206e-05,
+      "loss": 0.488,
+      "step": 983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818668473705448e-05,
+      "loss": 0.4423,
+      "step": 984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182535220359944e-05,
+      "loss": 0.5409,
+      "step": 985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178381435815914e-05,
+      "loss": 0.5455,
+      "step": 986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817422338558892e-05,
+      "loss": 0.5475,
+      "step": 987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170061071847717e-05,
+      "loss": 0.5129,
+      "step": 988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165894496763287e-05,
+      "loss": 0.4907,
+      "step": 989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161723662508834e-05,
+      "loss": 0.5083,
+      "step": 990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157548571259784e-05,
+      "loss": 0.4862,
+      "step": 991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153369225193782e-05,
+      "loss": 0.5378,
+      "step": 992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149185626490692e-05,
+      "loss": 0.5127,
+      "step": 993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144997777332597e-05,
+      "loss": 0.4794,
+      "step": 994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8140805679903793e-05,
+      "loss": 0.5302,
+      "step": 995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8136609336390803e-05,
+      "loss": 0.5701,
+      "step": 996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132408748982356e-05,
+      "loss": 0.5899,
+      "step": 997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812820391986939e-05,
+      "loss": 0.5559,
+      "step": 998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123994851245067e-05,
+      "loss": 0.4968,
+      "step": 999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119781545304755e-05,
+      "loss": 0.5482,
+      "step": 1000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115564004246025e-05,
+      "loss": 0.5385,
+      "step": 1001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111342230268667e-05,
+      "loss": 0.5396,
+      "step": 1002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107116225574674e-05,
+      "loss": 0.495,
+      "step": 1003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102885992368254e-05,
+      "loss": 0.4995,
+      "step": 1004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098651532855806e-05,
+      "loss": 0.5289,
+      "step": 1005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809441284924594e-05,
+      "loss": 0.4703,
+      "step": 1006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.5482,
+      "step": 1007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8085922818579426e-05,
+      "loss": 0.5258,
+      "step": 1008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808167147595101e-05,
+      "loss": 0.5249,
+      "step": 1009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807741591808164e-05,
+      "loss": 0.5192,
+      "step": 1010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8073156147190933e-05,
+      "loss": 0.5093,
+      "step": 1011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068892165500704e-05,
+      "loss": 0.4825,
+      "step": 1012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064623975234964e-05,
+      "loss": 0.4828,
+      "step": 1013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806035157861991e-05,
+      "loss": 0.5801,
+      "step": 1014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056074977883946e-05,
+      "loss": 0.4513,
+      "step": 1015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805179417525766e-05,
+      "loss": 0.5511,
+      "step": 1016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047509172973832e-05,
+      "loss": 0.5377,
+      "step": 1017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043219973267445e-05,
+      "loss": 0.4514,
+      "step": 1018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038926578375654e-05,
+      "loss": 0.5464,
+      "step": 1019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034628990537806e-05,
+      "loss": 0.5067,
+      "step": 1020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030327211995443e-05,
+      "loss": 0.602,
+      "step": 1021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802602124499229e-05,
+      "loss": 0.5017,
+      "step": 1022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802171109177425e-05,
+      "loss": 0.5552,
+      "step": 1023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017396754589415e-05,
+      "loss": 0.4527,
+      "step": 1024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.801307823568806e-05,
+      "loss": 0.4952,
+      "step": 1025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800875553732264e-05,
+      "loss": 0.453,
+      "step": 1026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800442866174779e-05,
+      "loss": 0.5035,
+      "step": 1027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8000097611220322e-05,
+      "loss": 0.5752,
+      "step": 1028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7995762387999228e-05,
+      "loss": 0.4687,
+      "step": 1029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991422994345667e-05,
+      "loss": 0.44,
+      "step": 1030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987079432522997e-05,
+      "loss": 0.5755,
+      "step": 1031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798273170479672e-05,
+      "loss": 0.4968,
+      "step": 1032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978379813434535e-05,
+      "loss": 0.572,
+      "step": 1033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974023760706298e-05,
+      "loss": 0.4948,
+      "step": 1034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796966354888404e-05,
+      "loss": 0.4812,
+      "step": 1035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965299180241963e-05,
+      "loss": 0.52,
+      "step": 1036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 0.516,
+      "step": 1037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956557981606e-05,
+      "loss": 0.5043,
+      "step": 1038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795218115617135e-05,
+      "loss": 0.5791,
+      "step": 1039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794780018303535e-05,
+      "loss": 0.4724,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7943415064483035e-05,
+      "loss": 0.4725,
+      "step": 1041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939025802801595e-05,
+      "loss": 0.4732,
+      "step": 1042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793463240028038e-05,
+      "loss": 0.4883,
+      "step": 1043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793023485921091e-05,
+      "loss": 0.507,
+      "step": 1044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925833181886845e-05,
+      "loss": 0.5688,
+      "step": 1045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921427370604024e-05,
+      "loss": 0.4832,
+      "step": 1046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917017427660424e-05,
+      "loss": 0.557,
+      "step": 1047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912603355356186e-05,
+      "loss": 0.4542,
+      "step": 1048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7908185155993604e-05,
+      "loss": 0.5032,
+      "step": 1049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903762831877128e-05,
+      "loss": 0.4891,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789933638531335e-05,
+      "loss": 0.5098,
+      "step": 1051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789490581861102e-05,
+      "loss": 0.5562,
+      "step": 1052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789047113408103e-05,
+      "loss": 0.539,
+      "step": 1053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788603233403643e-05,
+      "loss": 0.5815,
+      "step": 1054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.788158942079241e-05,
+      "loss": 0.4661,
+      "step": 1055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7877142396666305e-05,
+      "loss": 0.4993,
+      "step": 1056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78726912639776e-05,
+      "loss": 0.5271,
+      "step": 1057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868236025047904e-05,
+      "loss": 0.5323,
+      "step": 1058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863776682200997e-05,
+      "loss": 0.5447,
+      "step": 1059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7859313237762778e-05,
+      "loss": 0.5273,
+      "step": 1060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7854845694061294e-05,
+      "loss": 0.5633,
+      "step": 1061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850374053426725e-05,
+      "loss": 0.5156,
+      "step": 1062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845898318191395e-05,
+      "loss": 0.5104,
+      "step": 1063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784141849068976e-05,
+      "loss": 0.5087,
+      "step": 1064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 0.4957,
+      "step": 1065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783244656823605e-05,
+      "loss": 0.4738,
+      "step": 1066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782795447796356e-05,
+      "loss": 0.5276,
+      "step": 1067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823458304783916e-05,
+      "loss": 0.5332,
+      "step": 1068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818958051042234e-05,
+      "loss": 0.5563,
+      "step": 1069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.781445371908576e-05,
+      "loss": 0.5153,
+      "step": 1070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7809945311263863e-05,
+      "loss": 0.6044,
+      "step": 1071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7805432829928037e-05,
+      "loss": 0.5372,
+      "step": 1072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780091627743191e-05,
+      "loss": 0.5145,
+      "step": 1073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779639565613122e-05,
+      "loss": 0.6014,
+      "step": 1074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779187096838384e-05,
+      "loss": 0.5851,
+      "step": 1075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787342216549753e-05,
+      "loss": 0.5134,
+      "step": 1076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782809402991076e-05,
+      "loss": 0.4954,
+      "step": 1077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777827253007202e-05,
+      "loss": 0.4809,
+      "step": 1078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773731600158947e-05,
+      "loss": 0.5664,
+      "step": 1079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769186615620305e-05,
+      "loss": 0.5543,
+      "step": 1080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776463757882667e-05,
+      "loss": 0.4486,
+      "step": 1081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760084492150733e-05,
+      "loss": 0.507,
+      "step": 1082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755527357967288e-05,
+      "loss": 0.5023,
+      "step": 1083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775096617865326e-05,
+      "loss": 0.4958,
+      "step": 1084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7746400956587653e-05,
+      "loss": 0.4907,
+      "step": 1085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774183169415161e-05,
+      "loss": 0.4904,
+      "step": 1086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737258393728363e-05,
+      "loss": 0.5015,
+      "step": 1087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732681057703257e-05,
+      "loss": 0.5116,
+      "step": 1088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728099688463746e-05,
+      "loss": 0.4623,
+      "step": 1089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772351428839937e-05,
+      "loss": 0.5261,
+      "step": 1090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718924859901793e-05,
+      "loss": 0.4756,
+      "step": 1091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771433140536477e-05,
+      "loss": 0.4625,
+      "step": 1092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770973392718415e-05,
+      "loss": 0.421,
+      "step": 1093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705132427757895e-05,
+      "loss": 0.4686,
+      "step": 1094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770052690948605e-05,
+      "loss": 0.4793,
+      "step": 1095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769591737477077e-05,
+      "loss": 0.5465,
+      "step": 1096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769130382601629e-05,
+      "loss": 0.5453,
+      "step": 1097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686686265628953e-05,
+      "loss": 0.5507,
+      "step": 1098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7682064696017184e-05,
+      "loss": 0.5226,
+      "step": 1099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7677439119591502e-05,
+      "loss": 0.5619,
+      "step": 1100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767280953876452e-05,
+      "loss": 0.5333,
+      "step": 1101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668175955950935e-05,
+      "loss": 0.5239,
+      "step": 1102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766353837356753e-05,
+      "loss": 0.5071,
+      "step": 1103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765889679403318e-05,
+      "loss": 0.4705,
+      "step": 1104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765425121976884e-05,
+      "loss": 0.5275,
+      "step": 1105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649601653197556e-05,
+      "loss": 0.5305,
+      "step": 1106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764494809674444e-05,
+      "loss": 0.5309,
+      "step": 1107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640290552836704e-05,
+      "loss": 0.4912,
+      "step": 1108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763562902390363e-05,
+      "loss": 0.4965,
+      "step": 1109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630963512376576e-05,
+      "loss": 0.4926,
+      "step": 1110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626294020688984e-05,
+      "loss": 0.4869,
+      "step": 1111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621620551276366e-05,
+      "loss": 0.4874,
+      "step": 1112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616943106576316e-05,
+      "loss": 0.5927,
+      "step": 1113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612261689028493e-05,
+      "loss": 0.5705,
+      "step": 1114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760757630107464e-05,
+      "loss": 0.4857,
+      "step": 1115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7602886945158555e-05,
+      "loss": 0.529,
+      "step": 1116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598193623726115e-05,
+      "loss": 0.5422,
+      "step": 1117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759349633922527e-05,
+      "loss": 0.4837,
+      "step": 1118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588795094106024e-05,
+      "loss": 0.5126,
+      "step": 1119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758408989082046e-05,
+      "loss": 0.5522,
+      "step": 1120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579380731822712e-05,
+      "loss": 0.4736,
+      "step": 1121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574667619568994e-05,
+      "loss": 0.4475,
+      "step": 1122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 0.5818,
+      "step": 1123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756522954512876e-05,
+      "loss": 0.4641,
+      "step": 1124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756050458786496e-05,
+      "loss": 0.5013,
+      "step": 1125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555775687190607e-05,
+      "loss": 0.4848,
+      "step": 1126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755104284557221e-05,
+      "loss": 0.5384,
+      "step": 1127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754630606547832e-05,
+      "loss": 0.4388,
+      "step": 1128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541565349379555e-05,
+      "loss": 0.4363,
+      "step": 1129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7536820699748574e-05,
+      "loss": 0.432,
+      "step": 1130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7532072119060094e-05,
+      "loss": 0.4892,
+      "step": 1131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527319609790883e-05,
+      "loss": 0.4786,
+      "step": 1132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7522563174419753e-05,
+      "loss": 0.4477,
+      "step": 1133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517802815427572e-05,
+      "loss": 0.409,
+      "step": 1134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513038535297247e-05,
+      "loss": 0.4885,
+      "step": 1135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750827033651373e-05,
+      "loss": 0.4673,
+      "step": 1136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503498221564026e-05,
+      "loss": 0.4661,
+      "step": 1137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749872219293717e-05,
+      "loss": 0.5595,
+      "step": 1138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749394225312425e-05,
+      "loss": 0.562,
+      "step": 1139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489158404618386e-05,
+      "loss": 0.4492,
+      "step": 1140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748437064991474e-05,
+      "loss": 0.5103,
+      "step": 1141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7479578991510508e-05,
+      "loss": 0.4596,
+      "step": 1142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747478343190493e-05,
+      "loss": 0.4679,
+      "step": 1143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746998397359927e-05,
+      "loss": 0.4652,
+      "step": 1144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7465180619096834e-05,
+      "loss": 0.5571,
+      "step": 1145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.746037337090296e-05,
+      "loss": 0.4656,
+      "step": 1146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455562231525006e-05,
+      "loss": 0.4851,
+      "step": 1147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450747203472378e-05,
+      "loss": 0.495,
+      "step": 1148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744592828925649e-05,
+      "loss": 0.4883,
+      "step": 1149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441105491390804e-05,
+      "loss": 0.4551,
+      "step": 1150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7436278812390788e-05,
+      "loss": 0.5164,
+      "step": 1151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.4505,
+      "step": 1152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426613821059796e-05,
+      "loss": 0.4514,
+      "step": 1153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421775513769896e-05,
+      "loss": 0.494,
+      "step": 1154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7416933335427798e-05,
+      "loss": 0.4954,
+      "step": 1155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7412087288559097e-05,
+      "loss": 0.4725,
+      "step": 1156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7407237375691394e-05,
+      "loss": 0.3937,
+      "step": 1157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74023835993543e-05,
+      "loss": 0.5008,
+      "step": 1158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397525962079465e-05,
+      "loss": 0.4776,
+      "step": 1159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7392664466400524e-05,
+      "loss": 0.5201,
+      "step": 1160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738779911485314e-05,
+      "loss": 0.4919,
+      "step": 1161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382929909974988e-05,
+      "loss": 0.5068,
+      "step": 1162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737805685430575e-05,
+      "loss": 0.4713,
+      "step": 1163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737317995038711e-05,
+      "loss": 0.4333,
+      "step": 1164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368299200762768e-05,
+      "loss": 0.4823,
+      "step": 1165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363414607978432e-05,
+      "loss": 0.4945,
+      "step": 1166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358526174581806e-05,
+      "loss": 0.4794,
+      "step": 1167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353633903122598e-05,
+      "loss": 0.5393,
+      "step": 1168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348737796152525e-05,
+      "loss": 0.5022,
+      "step": 1169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343837856225297e-05,
+      "loss": 0.4961,
+      "step": 1170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338934085896626e-05,
+      "loss": 0.4676,
+      "step": 1171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7334026487724225e-05,
+      "loss": 0.5092,
+      "step": 1172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329115064267802e-05,
+      "loss": 0.4892,
+      "step": 1173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324199818089052e-05,
+      "loss": 0.4794,
+      "step": 1174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731928075175168e-05,
+      "loss": 0.4594,
+      "step": 1175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7314357867821366e-05,
+      "loss": 0.442,
+      "step": 1176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730943116886579e-05,
+      "loss": 0.4544,
+      "step": 1177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730450065745463e-05,
+      "loss": 0.5563,
+      "step": 1178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7299566336159536e-05,
+      "loss": 0.4971,
+      "step": 1179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294628207554157e-05,
+      "loss": 0.5229,
+      "step": 1180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 0.4798,
+      "step": 1181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728474053871704e-05,
+      "loss": 0.4422,
+      "step": 1182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279791003642515e-05,
+      "loss": 0.4898,
+      "step": 1183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274837671572133e-05,
+      "loss": 0.5142,
+      "step": 1184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726988054508945e-05,
+      "loss": 0.5103,
+      "step": 1185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726491962678e-05,
+      "loss": 0.5655,
+      "step": 1186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725995491923131e-05,
+      "loss": 0.5653,
+      "step": 1187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254986425032868e-05,
+      "loss": 0.5376,
+      "step": 1188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250014146776143e-05,
+      "loss": 0.5336,
+      "step": 1189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.724503808705458e-05,
+      "loss": 0.5169,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7240058248463593e-05,
+      "loss": 0.4611,
+      "step": 1191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235074633600566e-05,
+      "loss": 0.5032,
+      "step": 1192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230087245064857e-05,
+      "loss": 0.4684,
+      "step": 1193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722509608545779e-05,
+      "loss": 0.5057,
+      "step": 1194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220101157382653e-05,
+      "loss": 0.4763,
+      "step": 1195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72151024634447e-05,
+      "loss": 0.4663,
+      "step": 1196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210100006251164e-05,
+      "loss": 0.4267,
+      "step": 1197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7205093788411213e-05,
+      "loss": 0.4467,
+      "step": 1198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200083812536e-05,
+      "loss": 0.5328,
+      "step": 1199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719507008123863e-05,
+      "loss": 0.5226,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190052597134164e-05,
+      "loss": 0.4544,
+      "step": 1201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7185031362839625e-05,
+      "loss": 0.4771,
+      "step": 1202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718000638097399e-05,
+      "loss": 0.4802,
+      "step": 1203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7174977654158186e-05,
+      "loss": 0.4822,
+      "step": 1204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7169945185015106e-05,
+      "loss": 0.4622,
+      "step": 1205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164908976169582e-05,
+      "loss": 0.4422,
+      "step": 1206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159869030248404e-05,
+      "loss": 0.516,
+      "step": 1207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154825349880304e-05,
+      "loss": 0.5411,
+      "step": 1208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714977793769597e-05,
+      "loss": 0.4936,
+      "step": 1209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144726796328034e-05,
+      "loss": 0.4686,
+      "step": 1210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139671928411074e-05,
+      "loss": 0.5296,
+      "step": 1211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134613336581602e-05,
+      "loss": 0.5104,
+      "step": 1212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129551023478085e-05,
+      "loss": 0.5814,
+      "step": 1213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124484991740925e-05,
+      "loss": 0.5473,
+      "step": 1214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7119415244012464e-05,
+      "loss": 0.4504,
+      "step": 1215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114341782936987e-05,
+      "loss": 0.4838,
+      "step": 1216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710926461116071e-05,
+      "loss": 0.5227,
+      "step": 1217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104183731331782e-05,
+      "loss": 0.4606,
+      "step": 1218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709909914610029e-05,
+      "loss": 0.498,
+      "step": 1219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7094010858118257e-05,
+      "loss": 0.5273,
+      "step": 1220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088918870039634e-05,
+      "loss": 0.5061,
+      "step": 1221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70838231845203e-05,
+      "loss": 0.496,
+      "step": 1222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078723804218067e-05,
+      "loss": 0.4565,
+      "step": 1223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073620731792668e-05,
+      "loss": 0.3949,
+      "step": 1224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706851396990577e-05,
+      "loss": 0.4531,
+      "step": 1225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063403521220953e-05,
+      "loss": 0.533,
+      "step": 1226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705828938840373e-05,
+      "loss": 0.5627,
+      "step": 1227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7053171574121536e-05,
+      "loss": 0.4801,
+      "step": 1228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048050081043713e-05,
+      "loss": 0.463,
+      "step": 1229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704292491184154e-05,
+      "loss": 0.5265,
+      "step": 1230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037796069188203e-05,
+      "loss": 0.4942,
+      "step": 1231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70326635557588e-05,
+      "loss": 0.5058,
+      "step": 1232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7027527374230357e-05,
+      "loss": 0.4671,
+      "step": 1233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7022387527281797e-05,
+      "loss": 0.4747,
+      "step": 1234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701724401759397e-05,
+      "loss": 0.4675,
+      "step": 1235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7012096847849628e-05,
+      "loss": 0.5058,
+      "step": 1236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006946020733426e-05,
+      "loss": 0.461,
+      "step": 1237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001791538931945e-05,
+      "loss": 0.5577,
+      "step": 1238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 0.4798,
+      "step": 1239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699147162202894e-05,
+      "loss": 0.4564,
+      "step": 1240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986306192310086e-05,
+      "loss": 0.4864,
+      "step": 1241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981137118671275e-05,
+      "loss": 0.4936,
+      "step": 1242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6975964403808595e-05,
+      "loss": 0.5489,
+      "step": 1243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6970788050420033e-05,
+      "loss": 0.4942,
+      "step": 1244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965608061205477e-05,
+      "loss": 0.5447,
+      "step": 1245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69604244388667e-05,
+      "loss": 0.5414,
+      "step": 1246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6955237186107387e-05,
+      "loss": 0.5049,
+      "step": 1247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69500463056331e-05,
+      "loss": 0.5138,
+      "step": 1248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69448518001513e-05,
+      "loss": 0.4733,
+      "step": 1249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6939653672371346e-05,
+      "loss": 0.4649,
+      "step": 1250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693445192500447e-05,
+      "loss": 0.4932,
+      "step": 1251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6929246560763807e-05,
+      "loss": 0.4998,
+      "step": 1252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692403758236437e-05,
+      "loss": 0.5063,
+      "step": 1253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691882499252306e-05,
+      "loss": 0.5738,
+      "step": 1254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6913608793958664e-05,
+      "loss": 0.3795,
+      "step": 1255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690838898939185e-05,
+      "loss": 0.4858,
+      "step": 1256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690316558154516e-05,
+      "loss": 0.51,
+      "step": 1257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897938573143018e-05,
+      "loss": 0.4521,
+      "step": 1258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689270796691174e-05,
+      "loss": 0.4718,
+      "step": 1259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887473765579498e-05,
+      "loss": 0.4869,
+      "step": 1260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882235971876352e-05,
+      "loss": 0.44,
+      "step": 1261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876994588534234e-05,
+      "loss": 0.4438,
+      "step": 1262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871749618286943e-05,
+      "loss": 0.4756,
+      "step": 1263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6866501063870155e-05,
+      "loss": 0.5253,
+      "step": 1264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.686124892802141e-05,
+      "loss": 0.4684,
+      "step": 1265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855993213480123e-05,
+      "loss": 0.5418,
+      "step": 1266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685073392298757e-05,
+      "loss": 0.551,
+      "step": 1267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684547105928689e-05,
+      "loss": 0.4726,
+      "step": 1268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840204625123088e-05,
+      "loss": 0.4861,
+      "step": 1269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683493462324304e-05,
+      "loss": 0.5117,
+      "step": 1270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6829661056395473e-05,
+      "loss": 0.475,
+      "step": 1271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682438392733097e-05,
+      "loss": 0.4452,
+      "step": 1272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819103238801987e-05,
+      "loss": 0.4938,
+      "step": 1273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681381899356282e-05,
+      "loss": 0.4912,
+      "step": 1274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808531194369628e-05,
+      "loss": 0.4462,
+      "step": 1275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680323984398043e-05,
+      "loss": 0.5029,
+      "step": 1276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679794494515508e-05,
+      "loss": 0.4818,
+      "step": 1277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67926465006553e-05,
+      "loss": 0.4582,
+      "step": 1278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6787344513244654e-05,
+      "loss": 0.4917,
+      "step": 1279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.678203898568855e-05,
+      "loss": 0.5175,
+      "step": 1280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776729920754254e-05,
+      "loss": 0.4664,
+      "step": 1281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6771417321210862e-05,
+      "loss": 0.4938,
+      "step": 1282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6766101189829328e-05,
+      "loss": 0.4941,
+      "step": 1283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760781529382438e-05,
+      "loss": 0.5497,
+      "step": 1284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6755458342644825e-05,
+      "loss": 0.4476,
+      "step": 1285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750131632392958e-05,
+      "loss": 0.4248,
+      "step": 1286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750131632392958e-05,
+      "loss": 0.5358,
+      "step": 1287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6744801401405138e-05,
+      "loss": 0.4616,
+      "step": 1288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673946765246152e-05,
+      "loss": 0.4435,
+      "step": 1289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734130388344073e-05,
+      "loss": 0.3913,
+      "step": 1290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6728789611836616e-05,
+      "loss": 0.4379,
+      "step": 1291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6723445325724785e-05,
+      "loss": 0.409,
+      "step": 1292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718097532796064e-05,
+      "loss": 0.5386,
+      "step": 1293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.671274623583975e-05,
+      "loss": 0.4705,
+      "step": 1294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6707391437646973e-05,
+      "loss": 0.4455,
+      "step": 1295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6702033141010695e-05,
+      "loss": 0.4882,
+      "step": 1296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6696671348725697e-05,
+      "loss": 0.4753,
+      "step": 1297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.4927,
+      "step": 1298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685937288397785e-05,
+      "loss": 0.5007,
+      "step": 1299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680565025953542e-05,
+      "loss": 0.5028,
+      "step": 1300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667518927905793e-05,
+      "loss": 0.4751,
+      "step": 1301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669810050514827e-05,
+      "loss": 0.5105,
+      "step": 1302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664427343129933e-05,
+      "loss": 0.442,
+      "step": 1303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6659041159710762e-05,
+      "loss": 0.4681,
+      "step": 1304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653651503066646e-05,
+      "loss": 0.5385,
+      "step": 1305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648258376008723e-05,
+      "loss": 0.5203,
+      "step": 1306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6642861781349933e-05,
+      "loss": 0.4706,
+      "step": 1307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6637461721905047e-05,
+      "loss": 0.5145,
+      "step": 1308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.663205820049062e-05,
+      "loss": 0.3761,
+      "step": 1309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6626651219925027e-05,
+      "loss": 0.491,
+      "step": 1310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.662124078302844e-05,
+      "loss": 0.49,
+      "step": 1311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661582689262284e-05,
+      "loss": 0.4895,
+      "step": 1312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610409551532006e-05,
+      "loss": 0.5008,
+      "step": 1313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660498876258151e-05,
+      "loss": 0.6077,
+      "step": 1314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599564528598735e-05,
+      "loss": 0.566,
+      "step": 1315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594136852412853e-05,
+      "loss": 0.5415,
+      "step": 1316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588705736854838e-05,
+      "loss": 0.4487,
+      "step": 1317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583271184757444e-05,
+      "loss": 0.5025,
+      "step": 1318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577833198955235e-05,
+      "loss": 0.4478,
+      "step": 1319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657239178228455e-05,
+      "loss": 0.4308,
+      "step": 1320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656694693758353e-05,
+      "loss": 0.4139,
+      "step": 1321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561498667692096e-05,
+      "loss": 0.4607,
+      "step": 1322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6556046975451963e-05,
+      "loss": 0.5008,
+      "step": 1323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.655059186370662e-05,
+      "loss": 0.4598,
+      "step": 1324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6545133335301352e-05,
+      "loss": 0.5075,
+      "step": 1325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6539671393083218e-05,
+      "loss": 0.4904,
+      "step": 1326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534206039901057e-05,
+      "loss": 0.4905,
+      "step": 1327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528737278605492e-05,
+      "loss": 0.4767,
+      "step": 1328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6523265112048922e-05,
+      "loss": 0.4538,
+      "step": 1329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651778954308552e-05,
+      "loss": 0.4389,
+      "step": 1330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651231057457124e-05,
+      "loss": 0.4399,
+      "step": 1331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506828209363796e-05,
+      "loss": 0.4028,
+      "step": 1332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650134245032269e-05,
+      "loss": 0.4809,
+      "step": 1333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649585330030918e-05,
+      "loss": 0.4389,
+      "step": 1334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6490360762186305e-05,
+      "loss": 0.4605,
+      "step": 1335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484864838818856e-05,
+      "loss": 0.4593,
+      "step": 1336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64793655330734e-05,
+      "loss": 0.4838,
+      "step": 1337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 0.5321,
+      "step": 1338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.646835678592357e-05,
+      "loss": 0.4742,
+      "step": 1339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6462847350261137e-05,
+      "loss": 0.4449,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645733454370459e-05,
+      "loss": 0.4377,
+      "step": 1341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6451818369129298e-05,
+      "loss": 0.4888,
+      "step": 1342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6446298829412392e-05,
+      "loss": 0.4172,
+      "step": 1343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6440775927432754e-05,
+      "loss": 0.4912,
+      "step": 1344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643524966607102e-05,
+      "loss": 0.4399,
+      "step": 1345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429720048209586e-05,
+      "loss": 0.4926,
+      "step": 1346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424187076732585e-05,
+      "loss": 0.4174,
+      "step": 1347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641865075452591e-05,
+      "loss": 0.5201,
+      "step": 1348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413111084477195e-05,
+      "loss": 0.5529,
+      "step": 1349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407568069475822e-05,
+      "loss": 0.4965,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640202171241292e-05,
+      "loss": 0.4885,
+      "step": 1351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6396472016181363e-05,
+      "loss": 0.4804,
+      "step": 1352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6390918983675762e-05,
+      "loss": 0.4809,
+      "step": 1353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6385362617792465e-05,
+      "loss": 0.4294,
+      "step": 1354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637980292142957e-05,
+      "loss": 0.5489,
+      "step": 1355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.5159,
+      "step": 1356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636867354886602e-05,
+      "loss": 0.5891,
+      "step": 1357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636310387847023e-05,
+      "loss": 0.4068,
+      "step": 1358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635753088920456e-05,
+      "loss": 0.4386,
+      "step": 1359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635195458397577e-05,
+      "loss": 0.5066,
+      "step": 1360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634637496569235e-05,
+      "loss": 0.5088,
+      "step": 1361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634079203726453e-05,
+      "loss": 0.4423,
+      "step": 1362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6335205801604242e-05,
+      "loss": 0.5075,
+      "step": 1363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6329616261625158e-05,
+      "loss": 0.446,
+      "step": 1364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324023420242682e-05,
+      "loss": 0.4657,
+      "step": 1365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631842728037392e-05,
+      "loss": 0.4886,
+      "step": 1366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312827844937714e-05,
+      "loss": 0.4852,
+      "step": 1367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307225116854623e-05,
+      "loss": 0.3732,
+      "step": 1368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.630161909904691e-05,
+      "loss": 0.4293,
+      "step": 1369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6296009794438567e-05,
+      "loss": 0.5464,
+      "step": 1370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6290397205955303e-05,
+      "loss": 0.4834,
+      "step": 1371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6284781336524528e-05,
+      "loss": 0.4291,
+      "step": 1372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627916218907537e-05,
+      "loss": 0.4332,
+      "step": 1373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627353976653866e-05,
+      "loss": 0.43,
+      "step": 1374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267914071846956e-05,
+      "loss": 0.496,
+      "step": 1375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262285107934506e-05,
+      "loss": 0.537,
+      "step": 1376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625665287773726e-05,
+      "loss": 0.5132,
+      "step": 1377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251017384192885e-05,
+      "loss": 0.4832,
+      "step": 1378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245378630240738e-05,
+      "loss": 0.414,
+      "step": 1379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239736618821886e-05,
+      "loss": 0.5353,
+      "step": 1380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234091352879092e-05,
+      "loss": 0.5012,
+      "step": 1381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228442835356815e-05,
+      "loss": 0.489,
+      "step": 1382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222791069201208e-05,
+      "loss": 0.4859,
+      "step": 1383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6217136057360122e-05,
+      "loss": 0.4232,
+      "step": 1384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6211477802783105e-05,
+      "loss": 0.5301,
+      "step": 1385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6205816308421386e-05,
+      "loss": 0.4457,
+      "step": 1386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6200151577227892e-05,
+      "loss": 0.449,
+      "step": 1387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6194483612157232e-05,
+      "loss": 0.5519,
+      "step": 1388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6188812416165708e-05,
+      "loss": 0.5024,
+      "step": 1389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183137992211303e-05,
+      "loss": 0.4458,
+      "step": 1390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177460343253684e-05,
+      "loss": 0.4732,
+      "step": 1391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171779472254206e-05,
+      "loss": 0.4657,
+      "step": 1392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166095382175896e-05,
+      "loss": 0.4549,
+      "step": 1393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616040807598346e-05,
+      "loss": 0.4631,
+      "step": 1394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154717556643293e-05,
+      "loss": 0.5303,
+      "step": 1395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614902382712345e-05,
+      "loss": 0.4642,
+      "step": 1396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6143326890393675e-05,
+      "loss": 0.5017,
+      "step": 1397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6137626749425377e-05,
+      "loss": 0.5155,
+      "step": 1398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6131923407191635e-05,
+      "loss": 0.464,
+      "step": 1399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61262168666672e-05,
+      "loss": 0.4877,
+      "step": 1400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6120507130828493e-05,
+      "loss": 0.4924,
+      "step": 1401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114794202653602e-05,
+      "loss": 0.4337,
+      "step": 1402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109078085122272e-05,
+      "loss": 0.4611,
+      "step": 1403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610335878121592e-05,
+      "loss": 0.4286,
+      "step": 1404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097636293917626e-05,
+      "loss": 0.4135,
+      "step": 1405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6091910626212124e-05,
+      "loss": 0.4696,
+      "step": 1406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608618178108581e-05,
+      "loss": 0.5234,
+      "step": 1407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080449761526737e-05,
+      "loss": 0.4457,
+      "step": 1408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074714570524614e-05,
+      "loss": 0.5293,
+      "step": 1409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60689762110708e-05,
+      "loss": 0.4664,
+      "step": 1410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606323468615832e-05,
+      "loss": 0.5434,
+      "step": 1411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605748999878183e-05,
+      "loss": 0.4682,
+      "step": 1412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051742151937655e-05,
+      "loss": 0.4412,
+      "step": 1413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 0.4614,
+      "step": 1414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6040236991839733e-05,
+      "loss": 0.4316,
+      "step": 1415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6034479684586854e-05,
+      "loss": 0.5781,
+      "step": 1416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602871922986801e-05,
+      "loss": 0.4537,
+      "step": 1417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6022955630687752e-05,
+      "loss": 0.4176,
+      "step": 1418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017188890052247e-05,
+      "loss": 0.5089,
+      "step": 1419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6011419010969316e-05,
+      "loss": 0.4609,
+      "step": 1420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6005645996448422e-05,
+      "loss": 0.5167,
+      "step": 1421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599986984950065e-05,
+      "loss": 0.4458,
+      "step": 1422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5994090573138724e-05,
+      "loss": 0.4287,
+      "step": 1423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988308170377004e-05,
+      "loss": 0.4325,
+      "step": 1424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982522644231476e-05,
+      "loss": 0.4782,
+      "step": 1425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976733997719755e-05,
+      "loss": 0.4901,
+      "step": 1426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970942233861085e-05,
+      "loss": 0.5194,
+      "step": 1427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965147355676344e-05,
+      "loss": 0.4299,
+      "step": 1428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.595934936618802e-05,
+      "loss": 0.4744,
+      "step": 1429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5953548268420228e-05,
+      "loss": 0.4095,
+      "step": 1430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594774406539871e-05,
+      "loss": 0.436,
+      "step": 1431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941936760150825e-05,
+      "loss": 0.4517,
+      "step": 1432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593612635570555e-05,
+      "loss": 0.4473,
+      "step": 1433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593031285509347e-05,
+      "loss": 0.4548,
+      "step": 1434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59244962613468e-05,
+      "loss": 0.4544,
+      "step": 1435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591867657749936e-05,
+      "loss": 0.5642,
+      "step": 1436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912853806586585e-05,
+      "loss": 0.4781,
+      "step": 1437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590702795164551e-05,
+      "loss": 0.5116,
+      "step": 1438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901199015714796e-05,
+      "loss": 0.4935,
+      "step": 1439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5895367001834694e-05,
+      "loss": 0.5304,
+      "step": 1440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5889531913047072e-05,
+      "loss": 0.5054,
+      "step": 1441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5883693752395392e-05,
+      "loss": 0.4574,
+      "step": 1442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.5378,
+      "step": 1443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5872008227681757e-05,
+      "loss": 0.4769,
+      "step": 1444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.586616086971474e-05,
+      "loss": 0.4458,
+      "step": 1445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.586031045207354e-05,
+      "loss": 0.5277,
+      "step": 1446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585445697780963e-05,
+      "loss": 0.4624,
+      "step": 1447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584860044997606e-05,
+      "loss": 0.5109,
+      "step": 1448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5842740871627483e-05,
+      "loss": 0.4868,
+      "step": 1449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583687824582014e-05,
+      "loss": 0.4642,
+      "step": 1450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583101257561186e-05,
+      "loss": 0.4368,
+      "step": 1451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5825143864062063e-05,
+      "loss": 0.4784,
+      "step": 1452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581927211423175e-05,
+      "loss": 0.422,
+      "step": 1453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5813397329183514e-05,
+      "loss": 0.4995,
+      "step": 1454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5807519511981526e-05,
+      "loss": 0.4245,
+      "step": 1455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580163866569154e-05,
+      "loss": 0.4997,
+      "step": 1456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579575479338089e-05,
+      "loss": 0.4779,
+      "step": 1457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578986789811849e-05,
+      "loss": 0.461,
+      "step": 1458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5783977982974824e-05,
+      "loss": 0.4207,
+      "step": 1459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5778085051021957e-05,
+      "loss": 0.4902,
+      "step": 1460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5772189105333532e-05,
+      "loss": 0.4353,
+      "step": 1461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576629014898475e-05,
+      "loss": 0.4256,
+      "step": 1462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57603881850524e-05,
+      "loss": 0.4926,
+      "step": 1463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754483216614822e-05,
+      "loss": 0.4696,
+      "step": 1464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5748575246751935e-05,
+      "loss": 0.4718,
+      "step": 1465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574266427854522e-05,
+      "loss": 0.4914,
+      "step": 1466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573675031507772e-05,
+      "loss": 0.4025,
+      "step": 1467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573675031507772e-05,
+      "loss": 0.4814,
+      "step": 1468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573083335943405e-05,
+      "loss": 0.4741,
+      "step": 1469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5724913414700363e-05,
+      "loss": 0.4266,
+      "step": 1470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57189904839644e-05,
+      "loss": 0.5278,
+      "step": 1471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713064570315437e-05,
+      "loss": 0.4291,
+      "step": 1472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 0.396,
+      "step": 1473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5701203806643434e-05,
+      "loss": 0.4579,
+      "step": 1474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5695268962806736e-05,
+      "loss": 0.4997,
+      "step": 1475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5689331148429717e-05,
+      "loss": 0.4946,
+      "step": 1476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5683390366609428e-05,
+      "loss": 0.4748,
+      "step": 1477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677446620444464e-05,
+      "loss": 0.4736,
+      "step": 1478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671499913034962e-05,
+      "loss": 0.4084,
+      "step": 1479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566555024748261e-05,
+      "loss": 0.514,
+      "step": 1480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565959762689064e-05,
+      "loss": 0.4054,
+      "step": 1481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5653642054363818e-05,
+      "loss": 0.4305,
+      "step": 1482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647683533008455e-05,
+      "loss": 0.529,
+      "step": 1483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641722065932397e-05,
+      "loss": 0.5183,
+      "step": 1484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635757656245026e-05,
+      "loss": 0.5047,
+      "step": 1485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562979030705727e-05,
+      "loss": 0.464,
+      "step": 1486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562382002148157e-05,
+      "loss": 0.506,
+      "step": 1487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561784680263192e-05,
+      "loss": 0.4954,
+      "step": 1488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5611870653623826e-05,
+      "loss": 0.517,
+      "step": 1489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.560589157757433e-05,
+      "loss": 0.4627,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5599909577602003e-05,
+      "loss": 0.4918,
+      "step": 1491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5593924656826937e-05,
+      "loss": 0.4612,
+      "step": 1492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5587936818370754e-05,
+      "loss": 0.4149,
+      "step": 1493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5581946065356584e-05,
+      "loss": 0.4343,
+      "step": 1494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575952400909093e-05,
+      "loss": 0.4949,
+      "step": 1495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569955828154455e-05,
+      "loss": 0.4868,
+      "step": 1496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563956350220366e-05,
+      "loss": 0.5065,
+      "step": 1497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557953970236038e-05,
+      "loss": 0.4868,
+      "step": 1498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5551948691332192e-05,
+      "loss": 0.4342,
+      "step": 1499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5545940516641057e-05,
+      "loss": 0.4413,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5539929449296392e-05,
+      "loss": 0.3997,
+      "step": 1501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553391549243344e-05,
+      "loss": 0.4393,
+      "step": 1502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5527898649188974e-05,
+      "loss": 0.4289,
+      "step": 1503
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5521878922701244e-05,
+      "loss": 0.3845,
+      "step": 1504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5515856316110044e-05,
+      "loss": 0.4563,
+      "step": 1505
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5509830832556626e-05,
+      "loss": 0.4772,
+      "step": 1506
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503802475183773e-05,
+      "loss": 0.4196,
+      "step": 1507
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5497771247135757e-05,
+      "loss": 0.4221,
+      "step": 1508
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5491737151558352e-05,
+      "loss": 0.4357,
+      "step": 1509
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485700191598815e-05,
+      "loss": 0.4436,
+      "step": 1510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479660370405913e-05,
+      "loss": 0.4959,
+      "step": 1511
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5473617691129893e-05,
+      "loss": 0.5402,
+      "step": 1512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5467572156922504e-05,
+      "loss": 0.4534,
+      "step": 1513
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546152377093697e-05,
+      "loss": 0.4187,
+      "step": 1514
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455472536328016e-05,
+      "loss": 0.4216,
+      "step": 1515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5449418456251845e-05,
+      "loss": 0.4487,
+      "step": 1516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5443361533866146e-05,
+      "loss": 0.4607,
+      "step": 1517
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437301772330088e-05,
+      "loss": 0.4182,
+      "step": 1518
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.543123917480433e-05,
+      "loss": 0.4526,
+      "step": 1519
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5425173744450998e-05,
+      "loss": 0.452,
+      "step": 1520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5419105484433704e-05,
+      "loss": 0.423,
+      "step": 1521
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541303439791753e-05,
+      "loss": 0.4602,
+      "step": 1522
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540696048806904e-05,
+      "loss": 0.4719,
+      "step": 1523
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400883758056263e-05,
+      "loss": 0.4097,
+      "step": 1524
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5394804211048702e-05,
+      "loss": 0.4574,
+      "step": 1525
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538872185021733e-05,
+      "loss": 0.4495,
+      "step": 1526
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5382636678734584e-05,
+      "loss": 0.486,
+      "step": 1527
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5376548699774378e-05,
+      "loss": 0.4307,
+      "step": 1528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5370457916512072e-05,
+      "loss": 0.3872,
+      "step": 1529
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364364332124505e-05,
+      "loss": 0.4525,
+      "step": 1530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.3946,
+      "step": 1531
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352168772688216e-05,
+      "loss": 0.4393,
+      "step": 1532
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534606680400046e-05,
+      "loss": 0.4174,
+      "step": 1533
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5339962046909364e-05,
+      "loss": 0.3898,
+      "step": 1534
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.533385450459905e-05,
+      "loss": 0.4768,
+      "step": 1535
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5327744180255095e-05,
+      "loss": 0.3893,
+      "step": 1536
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5321631077064522e-05,
+      "loss": 0.4394,
+      "step": 1537
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315515198215807e-05,
+      "loss": 0.431,
+      "step": 1538
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530939654689887e-05,
+      "loss": 0.4795,
+      "step": 1539
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530327512630508e-05,
+      "loss": 0.4175,
+      "step": 1540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529715093962725e-05,
+      "loss": 0.4282,
+      "step": 1541
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291023990059633e-05,
+      "loss": 0.4379,
+      "step": 1542
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528489428079793e-05,
+      "loss": 0.433,
+      "step": 1543
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5278761815039276e-05,
+      "loss": 0.4259,
+      "step": 1544
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527262659598224e-05,
+      "loss": 0.4603,
+      "step": 1545
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5266488626826835e-05,
+      "loss": 0.4517,
+      "step": 1546
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5260347910774503e-05,
+      "loss": 0.4841,
+      "step": 1547
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5254204451028125e-05,
+      "loss": 0.4473,
+      "step": 1548
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5248058250792008e-05,
+      "loss": 0.5031,
+      "step": 1549
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5241909313271883e-05,
+      "loss": 0.3701,
+      "step": 1550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5235757641674922e-05,
+      "loss": 0.4607,
+      "step": 1551
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5229603239209713e-05,
+      "loss": 0.3636,
+      "step": 1552
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223446109086273e-05,
+      "loss": 0.3934,
+      "step": 1553
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5217286254516036e-05,
+      "loss": 0.3977,
+      "step": 1554
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211123678711865e-05,
+      "loss": 0.4212,
+      "step": 1555
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5204958384888035e-05,
+      "loss": 0.4604,
+      "step": 1556
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198790376260247e-05,
+      "loss": 0.3993,
+      "step": 1557
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192619656045606e-05,
+      "loss": 0.4048,
+      "step": 1558
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186446227462646e-05,
+      "loss": 0.473,
+      "step": 1559
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180270093731305e-05,
+      "loss": 0.4321,
+      "step": 1560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5174091258072926e-05,
+      "loss": 0.4053,
+      "step": 1561
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5167909723710275e-05,
+      "loss": 0.4626,
+      "step": 1562
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5161725493867514e-05,
+      "loss": 0.3532,
+      "step": 1563
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.515553857177022e-05,
+      "loss": 0.3604,
+      "step": 1564
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5149348960645367e-05,
+      "loss": 0.4407,
+      "step": 1565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5143156663721335e-05,
+      "loss": 0.4293,
+      "step": 1566
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136961684227905e-05,
+      "loss": 0.4215,
+      "step": 1567
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513076402539626e-05,
+      "loss": 0.4799,
+      "step": 1568
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124563690458966e-05,
+      "loss": 0.4245,
+      "step": 1569
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511836068265001e-05,
+      "loss": 0.3966,
+      "step": 1570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112155005204749e-05,
+      "loss": 0.3942,
+      "step": 1571
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105946661359945e-05,
+      "loss": 0.3678,
+      "step": 1572
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509973565435375e-05,
+      "loss": 0.4214,
+      "step": 1573
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093521987425698e-05,
+      "loss": 0.4659,
+      "step": 1574
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5087305663816723e-05,
+      "loss": 0.3993,
+      "step": 1575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508108668676913e-05,
+      "loss": 0.4878,
+      "step": 1576
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074865059526622e-05,
+      "loss": 0.4778,
+      "step": 1577
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5068640785334272e-05,
+      "loss": 0.5352,
+      "step": 1578
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.506241386743854e-05,
+      "loss": 0.4995,
+      "step": 1579
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5056184309087267e-05,
+      "loss": 0.4374,
+      "step": 1580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5049952113529666e-05,
+      "loss": 0.4191,
+      "step": 1581
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043717284016328e-05,
+      "loss": 0.47,
+      "step": 1582
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037479823799217e-05,
+      "loss": 0.4508,
+      "step": 1583
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031239736131667e-05,
+      "loss": 0.4491,
+      "step": 1584
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024997024268388e-05,
+      "loss": 0.3561,
+      "step": 1585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5018751691465458e-05,
+      "loss": 0.4043,
+      "step": 1586
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501250374098032e-05,
+      "loss": 0.4308,
+      "step": 1587
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006253176071775e-05,
+      "loss": 0.4661,
+      "step": 1588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4501,
+      "step": 1589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993744216026531e-05,
+      "loss": 0.4031,
+      "step": 1590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987485827414257e-05,
+      "loss": 0.4225,
+      "step": 1591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4981224837427433e-05,
+      "loss": 0.4039,
+      "step": 1592
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4974961249331666e-05,
+      "loss": 0.4549,
+      "step": 1593
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4968695066393923e-05,
+      "loss": 0.4576,
+      "step": 1594
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4962426291882524e-05,
+      "loss": 0.4299,
+      "step": 1595
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4956154929067128e-05,
+      "loss": 0.4503,
+      "step": 1596
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4949880981218768e-05,
+      "loss": 0.4535,
+      "step": 1597
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4943604451609805e-05,
+      "loss": 0.4235,
+      "step": 1598
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937325343513953e-05,
+      "loss": 0.4311,
+      "step": 1599
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931043660206274e-05,
+      "loss": 0.4447,
+      "step": 1600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492475940496317e-05,
+      "loss": 0.4352,
+      "step": 1601
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918472581062382e-05,
+      "loss": 0.4161,
+      "step": 1602
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912183191782995e-05,
+      "loss": 0.3784,
+      "step": 1603
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905891240405434e-05,
+      "loss": 0.506,
+      "step": 1604
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899596730211454e-05,
+      "loss": 0.3927,
+      "step": 1605
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893299664484151e-05,
+      "loss": 0.4003,
+      "step": 1606
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4887000046507945e-05,
+      "loss": 0.3848,
+      "step": 1607
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4880697879568597e-05,
+      "loss": 0.4759,
+      "step": 1608
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4874393166953194e-05,
+      "loss": 0.4752,
+      "step": 1609
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4868085911950147e-05,
+      "loss": 0.4133,
+      "step": 1610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4861776117849198e-05,
+      "loss": 0.4569,
+      "step": 1611
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855463787941414e-05,
+      "loss": 0.4894,
+      "step": 1612
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849148925519174e-05,
+      "loss": 0.4724,
+      "step": 1613
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4842831533876196e-05,
+      "loss": 0.426,
+      "step": 1614
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4836511616307501e-05,
+      "loss": 0.4089,
+      "step": 1615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830189176109438e-05,
+      "loss": 0.4959,
+      "step": 1616
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4823864216579664e-05,
+      "loss": 0.3918,
+      "step": 1617
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4817536741017153e-05,
+      "loss": 0.4639,
+      "step": 1618
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811206752722196e-05,
+      "loss": 0.4436,
+      "step": 1619
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4804874254996389e-05,
+      "loss": 0.4228,
+      "step": 1620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4798539251142634e-05,
+      "loss": 0.4664,
+      "step": 1621
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792201744465148e-05,
+      "loss": 0.457,
+      "step": 1622
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4785861738269451e-05,
+      "loss": 0.4409,
+      "step": 1623
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4779519235862365e-05,
+      "loss": 0.4013,
+      "step": 1624
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4773174240552016e-05,
+      "loss": 0.3952,
+      "step": 1625
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4766826755647821e-05,
+      "loss": 0.4405,
+      "step": 1626
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760476784460514e-05,
+      "loss": 0.4482,
+      "step": 1627
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754124330302106e-05,
+      "loss": 0.4929,
+      "step": 1628
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747769396485917e-05,
+      "loss": 0.4573,
+      "step": 1629
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741411986326558e-05,
+      "loss": 0.4517,
+      "step": 1630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735052103139927e-05,
+      "loss": 0.4695,
+      "step": 1631
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728689750243206e-05,
+      "loss": 0.3729,
+      "step": 1632
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722324930954885e-05,
+      "loss": 0.4114,
+      "step": 1633
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715957648594724e-05,
+      "loss": 0.3834,
+      "step": 1634
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470958790648377e-05,
+      "loss": 0.3854,
+      "step": 1635
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470321570794436e-05,
+      "loss": 0.4534,
+      "step": 1636
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4696841056300102e-05,
+      "loss": 0.4643,
+      "step": 1637
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4690463954875893e-05,
+      "loss": 0.4133,
+      "step": 1638
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 0.4634,
+      "step": 1639
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4677702415993578e-05,
+      "loss": 0.4546,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.467131798519164e-05,
+      "loss": 0.4603,
+      "step": 1641
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664931117922081e-05,
+      "loss": 0.4635,
+      "step": 1642
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658541817516166e-05,
+      "loss": 0.4274,
+      "step": 1643
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652150087306429e-05,
+      "loss": 0.3757,
+      "step": 1644
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645755930626671e-05,
+      "loss": 0.4297,
+      "step": 1645
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639359350811957e-05,
+      "loss": 0.4573,
+      "step": 1646
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 0.4066,
+      "step": 1647
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626558935124248e-05,
+      "loss": 0.4129,
+      "step": 1648
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4620155105927699e-05,
+      "loss": 0.3978,
+      "step": 1649
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4613748866949082e-05,
+      "loss": 0.4637,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607340221529762e-05,
+      "loss": 0.4747,
+      "step": 1651
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4600929173012368e-05,
+      "loss": 0.4794,
+      "step": 1652
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4594515724740775e-05,
+      "loss": 0.405,
+      "step": 1653
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.458809988006011e-05,
+      "loss": 0.4778,
+      "step": 1654
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4581681642316752e-05,
+      "loss": 0.5195,
+      "step": 1655
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.457526101485833e-05,
+      "loss": 0.4365,
+      "step": 1656
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568838001033711e-05,
+      "loss": 0.4673,
+      "step": 1657
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562412604193017e-05,
+      "loss": 0.5072,
+      "step": 1658
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4555984827687613e-05,
+      "loss": 0.447,
+      "step": 1659
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4549554674870092e-05,
+      "loss": 0.4054,
+      "step": 1660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4543122149094304e-05,
+      "loss": 0.394,
+      "step": 1661
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536687253715318e-05,
+      "loss": 0.4526,
+      "step": 1662
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453024999208946e-05,
+      "loss": 0.3596,
+      "step": 1663
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4523810367574271e-05,
+      "loss": 0.4645,
+      "step": 1664
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4517368383528534e-05,
+      "loss": 0.4875,
+      "step": 1665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510924043312265e-05,
+      "loss": 0.4594,
+      "step": 1666
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4504477350286701e-05,
+      "loss": 0.4151,
+      "step": 1667
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4498028307814311e-05,
+      "loss": 0.4629,
+      "step": 1668
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4491576919258794e-05,
+      "loss": 0.4742,
+      "step": 1669
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4485123187985057e-05,
+      "loss": 0.3885,
+      "step": 1670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4478667117359251e-05,
+      "loss": 0.3956,
+      "step": 1671
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472208710748726e-05,
+      "loss": 0.4736,
+      "step": 1672
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465747971522067e-05,
+      "loss": 0.4591,
+      "step": 1673
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445928490304906e-05,
+      "loss": 0.3995,
+      "step": 1674
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4452819508700724e-05,
+      "loss": 0.4229,
+      "step": 1675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4446351791849276e-05,
+      "loss": 0.4178,
+      "step": 1676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439881755868153e-05,
+      "loss": 0.4123,
+      "step": 1677
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4433409404131995e-05,
+      "loss": 0.4081,
+      "step": 1678
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426934740016655e-05,
+      "loss": 0.4276,
+      "step": 1679
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420457766899195e-05,
+      "loss": 0.4307,
+      "step": 1680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413978488157867e-05,
+      "loss": 0.3696,
+      "step": 1681
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407496907172145e-05,
+      "loss": 0.3478,
+      "step": 1682
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.440101302732269e-05,
+      "loss": 0.4027,
+      "step": 1683
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4394526851991363e-05,
+      "loss": 0.5213,
+      "step": 1684
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4388038384561234e-05,
+      "loss": 0.4326,
+      "step": 1685
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4381547628416552e-05,
+      "loss": 0.4055,
+      "step": 1686
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375054586942771e-05,
+      "loss": 0.5106,
+      "step": 1687
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4368559263526539e-05,
+      "loss": 0.4006,
+      "step": 1688
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362061661555675e-05,
+      "loss": 0.371,
+      "step": 1689
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355561784419215e-05,
+      "loss": 0.447,
+      "step": 1690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4349059635507359e-05,
+      "loss": 0.467,
+      "step": 1691
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4342555218211502e-05,
+      "loss": 0.4382,
+      "step": 1692
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336048535924223e-05,
+      "loss": 0.457,
+      "step": 1693
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329539592039275e-05,
+      "loss": 0.3613,
+      "step": 1694
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323028389951597e-05,
+      "loss": 0.4526,
+      "step": 1695
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316514933057304e-05,
+      "loss": 0.4247,
+      "step": 1696
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309999224753688e-05,
+      "loss": 0.5071,
+      "step": 1697
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4303481268439213e-05,
+      "loss": 0.3276,
+      "step": 1698
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4296961067513519e-05,
+      "loss": 0.3899,
+      "step": 1699
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.429043862537741e-05,
+      "loss": 0.4142,
+      "step": 1700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4283913945432871e-05,
+      "loss": 0.4012,
+      "step": 1701
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4277387031083039e-05,
+      "loss": 0.4454,
+      "step": 1702
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427085788573223e-05,
+      "loss": 0.4155,
+      "step": 1703
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264326512785914e-05,
+      "loss": 0.4112,
+      "step": 1704
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 0.4301,
+      "step": 1705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251257097734467e-05,
+      "loss": 0.4263,
+      "step": 1706
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424471906244609e-05,
+      "loss": 0.428,
+      "step": 1707
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42381788131957e-05,
+      "loss": 0.3901,
+      "step": 1708
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423163635339457e-05,
+      "loss": 0.432,
+      "step": 1709
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422509168645511e-05,
+      "loss": 0.416,
+      "step": 1710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4218544815790897e-05,
+      "loss": 0.3958,
+      "step": 1711
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211995744816647e-05,
+      "loss": 0.4593,
+      "step": 1712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4205444476948225e-05,
+      "loss": 0.4727,
+      "step": 1713
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4198891015602648e-05,
+      "loss": 0.3553,
+      "step": 1714
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.419233536419807e-05,
+      "loss": 0.4364,
+      "step": 1715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185777526153786e-05,
+      "loss": 0.5392,
+      "step": 1716
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417921750489024e-05,
+      "loss": 0.4851,
+      "step": 1717
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172655303829017e-05,
+      "loss": 0.5169,
+      "step": 1718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166090926392815e-05,
+      "loss": 0.4327,
+      "step": 1719
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41595243760055e-05,
+      "loss": 0.4163,
+      "step": 1720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152955656092046e-05,
+      "loss": 0.3984,
+      "step": 1721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146384770078574e-05,
+      "loss": 0.4379,
+      "step": 1722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139811721392325e-05,
+      "loss": 0.3986,
+      "step": 1723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4133236513461673e-05,
+      "loss": 0.3968,
+      "step": 1724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126659149716115e-05,
+      "loss": 0.4597,
+      "step": 1725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120079633586284e-05,
+      "loss": 0.3921,
+      "step": 1726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411349796850391e-05,
+      "loss": 0.4099,
+      "step": 1727
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4106914157901878e-05,
+      "loss": 0.4528,
+      "step": 1728
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4100328205214161e-05,
+      "loss": 0.3454,
+      "step": 1729
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.409374011387587e-05,
+      "loss": 0.4401,
+      "step": 1730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408714988732322e-05,
+      "loss": 0.4357,
+      "step": 1731
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080557528993544e-05,
+      "loss": 0.4328,
+      "step": 1732
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4073963042325284e-05,
+      "loss": 0.3736,
+      "step": 1733
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.4838,
+      "step": 1734
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4060767697732356e-05,
+      "loss": 0.3673,
+      "step": 1735
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4054166846690116e-05,
+      "loss": 0.3808,
+      "step": 1736
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047563881074157e-05,
+      "loss": 0.4226,
+      "step": 1737
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040958804328453e-05,
+      "loss": 0.4385,
+      "step": 1738
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4034351619898088e-05,
+      "loss": 0.3942,
+      "step": 1739
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027742331229233e-05,
+      "loss": 0.3983,
+      "step": 1740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4021130941769165e-05,
+      "loss": 0.3829,
+      "step": 1741
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014517454966257e-05,
+      "loss": 0.4376,
+      "step": 1742
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.400790187426997e-05,
+      "loss": 0.4994,
+      "step": 1743
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4001284203130868e-05,
+      "loss": 0.3322,
+      "step": 1744
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.399466444500059e-05,
+      "loss": 0.4369,
+      "step": 1745
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3988042603331875e-05,
+      "loss": 0.4017,
+      "step": 1746
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981418681578546e-05,
+      "loss": 0.4568,
+      "step": 1747
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974792683195513e-05,
+      "loss": 0.3966,
+      "step": 1748
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968164611638759e-05,
+      "loss": 0.3671,
+      "step": 1749
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3961534470365361e-05,
+      "loss": 0.3987,
+      "step": 1750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395490226283347e-05,
+      "loss": 0.4733,
+      "step": 1751
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948267992502314e-05,
+      "loss": 0.526,
+      "step": 1752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941631662832201e-05,
+      "loss": 0.4121,
+      "step": 1753
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934993277284503e-05,
+      "loss": 0.3893,
+      "step": 1754
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928352839321679e-05,
+      "loss": 0.3942,
+      "step": 1755
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921710352407248e-05,
+      "loss": 0.4457,
+      "step": 1756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3915065820005795e-05,
+      "loss": 0.4068,
+      "step": 1757
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.390841924558299e-05,
+      "loss": 0.3927,
+      "step": 1758
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3901770632605546e-05,
+      "loss": 0.4439,
+      "step": 1759
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3895119984541252e-05,
+      "loss": 0.4258,
+      "step": 1760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3888467304858952e-05,
+      "loss": 0.472,
+      "step": 1761
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3881812597028557e-05,
+      "loss": 0.3786,
+      "step": 1762
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 0.435,
+      "step": 1763
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3868497110808394e-05,
+      "loss": 0.4521,
+      "step": 1764
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3861836339363718e-05,
+      "loss": 0.3806,
+      "step": 1765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855173553661136e-05,
+      "loss": 0.4415,
+      "step": 1766
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848508757175825e-05,
+      "loss": 0.4572,
+      "step": 1767
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841841953384008e-05,
+      "loss": 0.4033,
+      "step": 1768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3835173145762961e-05,
+      "loss": 0.3619,
+      "step": 1769
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3828502337791005e-05,
+      "loss": 0.4458,
+      "step": 1770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821829532947497e-05,
+      "loss": 0.4351,
+      "step": 1771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815154734712848e-05,
+      "loss": 0.3742,
+      "step": 1772
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3808477946568494e-05,
+      "loss": 0.3802,
+      "step": 1773
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.380179917199692e-05,
+      "loss": 0.4052,
+      "step": 1774
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3795118414481648e-05,
+      "loss": 0.4154,
+      "step": 1775
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378843567750722e-05,
+      "loss": 0.4152,
+      "step": 1776
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3781750964559228e-05,
+      "loss": 0.3413,
+      "step": 1777
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3775064279124286e-05,
+      "loss": 0.4112,
+      "step": 1778
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768375624690033e-05,
+      "loss": 0.4978,
+      "step": 1779
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3761685004745144e-05,
+      "loss": 0.4337,
+      "step": 1780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754992422779317e-05,
+      "loss": 0.4531,
+      "step": 1781
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748297882283262e-05,
+      "loss": 0.3967,
+      "step": 1782
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741601386748728e-05,
+      "loss": 0.3827,
+      "step": 1783
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3734902939668473e-05,
+      "loss": 0.4698,
+      "step": 1784
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728202544536276e-05,
+      "loss": 0.3859,
+      "step": 1785
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372150020484693e-05,
+      "loss": 0.5314,
+      "step": 1786
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714795924096243e-05,
+      "loss": 0.4711,
+      "step": 1787
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3708089705781038e-05,
+      "loss": 0.3962,
+      "step": 1788
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3701381553399147e-05,
+      "loss": 0.5083,
+      "step": 1789
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3694671470449406e-05,
+      "loss": 0.4104,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3687959460431668e-05,
+      "loss": 0.4256,
+      "step": 1791
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3681245526846782e-05,
+      "loss": 0.4745,
+      "step": 1792
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674529673196606e-05,
+      "loss": 0.5476,
+      "step": 1793
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667811902983995e-05,
+      "loss": 0.463,
+      "step": 1794
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661092219712803e-05,
+      "loss": 0.4485,
+      "step": 1795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654370626887894e-05,
+      "loss": 0.3973,
+      "step": 1796
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647647128015111e-05,
+      "loss": 0.4212,
+      "step": 1797
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640921726601302e-05,
+      "loss": 0.4033,
+      "step": 1798
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3634194426154301e-05,
+      "loss": 0.4265,
+      "step": 1799
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3627465230182942e-05,
+      "loss": 0.4341,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620734142197031e-05,
+      "loss": 0.404,
+      "step": 1801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3614001165707383e-05,
+      "loss": 0.425,
+      "step": 1802
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3607266304225779e-05,
+      "loss": 0.4079,
+      "step": 1803
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.360052956126499e-05,
+      "loss": 0.4549,
+      "step": 1804
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.359379094033877e-05,
+      "loss": 0.4416,
+      "step": 1805
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587050444961853e-05,
+      "loss": 0.4292,
+      "step": 1806
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580308078649948e-05,
+      "loss": 0.456,
+      "step": 1807
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573563844919741e-05,
+      "loss": 0.3833,
+      "step": 1808
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3566817747288887e-05,
+      "loss": 0.4155,
+      "step": 1809
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560069789276026e-05,
+      "loss": 0.4667,
+      "step": 1810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553319974400756e-05,
+      "loss": 0.4929,
+      "step": 1811
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3546568306183644e-05,
+      "loss": 0.3362,
+      "step": 1812
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539814788146235e-05,
+      "loss": 0.4545,
+      "step": 1813
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3533059423811026e-05,
+      "loss": 0.4681,
+      "step": 1814
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352630221670148e-05,
+      "loss": 0.4095,
+      "step": 1815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3519543170342033e-05,
+      "loss": 0.44,
+      "step": 1816
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512782288258061e-05,
+      "loss": 0.4002,
+      "step": 1817
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3506019573975913e-05,
+      "loss": 0.4242,
+      "step": 1818
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3499255031022887e-05,
+      "loss": 0.4282,
+      "step": 1819
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.349248866292723e-05,
+      "loss": 0.4009,
+      "step": 1820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 0.4121,
+      "step": 1821
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478950465425808e-05,
+      "loss": 0.3613,
+      "step": 1822
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3472178643081303e-05,
+      "loss": 0.3548,
+      "step": 1823
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3465405009716676e-05,
+      "loss": 0.4573,
+      "step": 1824
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3458629568864931e-05,
+      "loss": 0.4729,
+      "step": 1825
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3451852324060002e-05,
+      "loss": 0.4378,
+      "step": 1826
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3445073278836767e-05,
+      "loss": 0.507,
+      "step": 1827
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343829243673104e-05,
+      "loss": 0.4322,
+      "step": 1828
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431509801279578e-05,
+      "loss": 0.4238,
+      "step": 1829
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3424725376020069e-05,
+      "loss": 0.4886,
+      "step": 1830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417939164491136e-05,
+      "loss": 0.457,
+      "step": 1831
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411151170232332e-05,
+      "loss": 0.4568,
+      "step": 1832
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3404361396784148e-05,
+      "loss": 0.4082,
+      "step": 1833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3397569847687985e-05,
+      "loss": 0.3811,
+      "step": 1834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3390776526486193e-05,
+      "loss": 0.5085,
+      "step": 1835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3383981436722025e-05,
+      "loss": 0.4289,
+      "step": 1836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377184581939672e-05,
+      "loss": 0.3784,
+      "step": 1837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3370385965684242e-05,
+      "loss": 0.3635,
+      "step": 1838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3363585591501751e-05,
+      "loss": 0.4237,
+      "step": 1839
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356783462939151e-05,
+      "loss": 0.4177,
+      "step": 1840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349979583544291e-05,
+      "loss": 0.3837,
+      "step": 1841
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349979583544291e-05,
+      "loss": 0.3984,
+      "step": 1842
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334317395686594e-05,
+      "loss": 0.4475,
+      "step": 1843
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336366586453783e-05,
+      "loss": 0.467,
+      "step": 1844
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329557475858413e-05,
+      "loss": 0.4044,
+      "step": 1845
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332274662863132e-05,
+      "loss": 0.3901,
+      "step": 1846
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315934048324919e-05,
+      "loss": 0.4112,
+      "step": 1847
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3309119738492503e-05,
+      "loss": 0.4158,
+      "step": 1848
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3302303702688299e-05,
+      "loss": 0.4408,
+      "step": 1849
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3295485944467405e-05,
+      "loss": 0.393,
+      "step": 1850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3288666467385834e-05,
+      "loss": 0.4312,
+      "step": 1851
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281845275000492e-05,
+      "loss": 0.4043,
+      "step": 1852
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275022370869176e-05,
+      "loss": 0.4225,
+      "step": 1853
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326819775855058e-05,
+      "loss": 0.4179,
+      "step": 1854
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261371441604292e-05,
+      "loss": 0.3778,
+      "step": 1855
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254543423590773e-05,
+      "loss": 0.4154,
+      "step": 1856
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247713708071394e-05,
+      "loss": 0.3667,
+      "step": 1857
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240882298608402e-05,
+      "loss": 0.3798,
+      "step": 1858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3234049198764914e-05,
+      "loss": 0.4773,
+      "step": 1859
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3227214412104952e-05,
+      "loss": 0.3944,
+      "step": 1860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3220377942193399e-05,
+      "loss": 0.4374,
+      "step": 1861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213539792596027e-05,
+      "loss": 0.3683,
+      "step": 1862
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3206699966879477e-05,
+      "loss": 0.3114,
+      "step": 1863
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.319985846861127e-05,
+      "loss": 0.3971,
+      "step": 1864
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31930153013598e-05,
+      "loss": 0.4219,
+      "step": 1865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3186170468694327e-05,
+      "loss": 0.3899,
+      "step": 1866
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179323974184975e-05,
+      "loss": 0.4389,
+      "step": 1867
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3172475821402748e-05,
+      "loss": 0.4999,
+      "step": 1868
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165626013919503e-05,
+      "loss": 0.4074,
+      "step": 1869
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158774555307972e-05,
+      "loss": 0.3841,
+      "step": 1870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3151921449141732e-05,
+      "loss": 0.3865,
+      "step": 1871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145066698995233e-05,
+      "loss": 0.4814,
+      "step": 1872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313821030844378e-05,
+      "loss": 0.3886,
+      "step": 1873
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313135228106353e-05,
+      "loss": 0.3827,
+      "step": 1874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312449262043149e-05,
+      "loss": 0.4703,
+      "step": 1875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311763133012553e-05,
+      "loss": 0.3859,
+      "step": 1876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3110768413724367e-05,
+      "loss": 0.3415,
+      "step": 1877
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3103903874807554e-05,
+      "loss": 0.4435,
+      "step": 1878
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3097037716955507e-05,
+      "loss": 0.3921,
+      "step": 1879
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.4229,
+      "step": 1880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3083300558771558e-05,
+      "loss": 0.3407,
+      "step": 1881
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076429565604685e-05,
+      "loss": 0.4592,
+      "step": 1882
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069556967832638e-05,
+      "loss": 0.3926,
+      "step": 1883
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062682769040026e-05,
+      "loss": 0.3678,
+      "step": 1884
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055806972812294e-05,
+      "loss": 0.3742,
+      "step": 1885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048929582735724e-05,
+      "loss": 0.426,
+      "step": 1886
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304205060239743e-05,
+      "loss": 0.4328,
+      "step": 1887
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035170035385347e-05,
+      "loss": 0.522,
+      "step": 1888
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028287885288248e-05,
+      "loss": 0.4037,
+      "step": 1889
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021404155695728e-05,
+      "loss": 0.4744,
+      "step": 1890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30145188501982e-05,
+      "loss": 0.4467,
+      "step": 1891
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3007631972386912e-05,
+      "loss": 0.3866,
+      "step": 1892
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3000743525853917e-05,
+      "loss": 0.476,
+      "step": 1893
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.299385351419209e-05,
+      "loss": 0.3677,
+      "step": 1894
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2986961940995138e-05,
+      "loss": 0.3898,
+      "step": 1895
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.298006880985756e-05,
+      "loss": 0.3769,
+      "step": 1896
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297317412437468e-05,
+      "loss": 0.4685,
+      "step": 1897
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966277888142633e-05,
+      "loss": 0.3924,
+      "step": 1898
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959380104758357e-05,
+      "loss": 0.4211,
+      "step": 1899
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29524807778196e-05,
+      "loss": 0.4555,
+      "step": 1900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294557991092492e-05,
+      "loss": 0.3297,
+      "step": 1901
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293867750767367e-05,
+      "loss": 0.4481,
+      "step": 1902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2931773571666007e-05,
+      "loss": 0.4066,
+      "step": 1903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292486810650289e-05,
+      "loss": 0.3659,
+      "step": 1904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917961115786072e-05,
+      "loss": 0.4108,
+      "step": 1905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2911052603118109e-05,
+      "loss": 0.4869,
+      "step": 1906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2904142572102338e-05,
+      "loss": 0.4195,
+      "step": 1907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2897231026342898e-05,
+      "loss": 0.3931,
+      "step": 1908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2890317969444716e-05,
+      "loss": 0.346,
+      "step": 1909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.288340340501351e-05,
+      "loss": 0.4557,
+      "step": 1910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2876487336655775e-05,
+      "loss": 0.3475,
+      "step": 1911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28695697679788e-05,
+      "loss": 0.4055,
+      "step": 1912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286265070259065e-05,
+      "loss": 0.4235,
+      "step": 1913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2855730144100174e-05,
+      "loss": 0.3983,
+      "step": 1914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848808096117003e-05,
+      "loss": 0.4302,
+      "step": 1915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841884562251532e-05,
+      "loss": 0.4314,
+      "step": 1916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2834959546114953e-05,
+      "loss": 0.4419,
+      "step": 1917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282803305131921e-05,
+      "loss": 0.4334,
+      "step": 1918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282110508147703e-05,
+      "loss": 0.3935,
+      "step": 1919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28141756402019e-05,
+      "loss": 0.4159,
+      "step": 1920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807244731108091e-05,
+      "loss": 0.298,
+      "step": 1921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2800312357810616e-05,
+      "loss": 0.3879,
+      "step": 1922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2793378523925275e-05,
+      "loss": 0.447,
+      "step": 1923
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2786443233068613e-05,
+      "loss": 0.422,
+      "step": 1924
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2779506488857945e-05,
+      "loss": 0.4472,
+      "step": 1925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2772568294911339e-05,
+      "loss": 0.412,
+      "step": 1926
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276562865484762e-05,
+      "loss": 0.4695,
+      "step": 1927
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758687572286367e-05,
+      "loss": 0.4283,
+      "step": 1928
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751745050847921e-05,
+      "loss": 0.4381,
+      "step": 1929
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274480109415335e-05,
+      "loss": 0.404,
+      "step": 1930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737855705824495e-05,
+      "loss": 0.4677,
+      "step": 1931
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2730908889483934e-05,
+      "loss": 0.3754,
+      "step": 1932
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2723960648754985e-05,
+      "loss": 0.4023,
+      "step": 1933
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717010987261716e-05,
+      "loss": 0.4038,
+      "step": 1934
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2710059908628936e-05,
+      "loss": 0.381,
+      "step": 1935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703107416482188e-05,
+      "loss": 0.376,
+      "step": 1936
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2696153514447756e-05,
+      "loss": 0.4644,
+      "step": 1937
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 0.4153,
+      "step": 1938
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2682241495224647e-05,
+      "loss": 0.4533,
+      "step": 1939
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 0.4541,
+      "step": 1940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2668323879984552e-05,
+      "loss": 0.39,
+      "step": 1941
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661362982931625e-05,
+      "loss": 0.3936,
+      "step": 1942
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654400697764087e-05,
+      "loss": 0.4102,
+      "step": 1943
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647437028113332e-05,
+      "loss": 0.4437,
+      "step": 1944
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2640471977611473e-05,
+      "loss": 0.4703,
+      "step": 1945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2633505549891345e-05,
+      "loss": 0.453,
+      "step": 1946
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2626537748586495e-05,
+      "loss": 0.4191,
+      "step": 1947
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26195685773312e-05,
+      "loss": 0.4205,
+      "step": 1948
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2612598039760433e-05,
+      "loss": 0.3893,
+      "step": 1949
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605626139509899e-05,
+      "loss": 0.4246,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2598652880215995e-05,
+      "loss": 0.4324,
+      "step": 1951
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2591678265515838e-05,
+      "loss": 0.3945,
+      "step": 1952
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2584702299047256e-05,
+      "loss": 0.4077,
+      "step": 1953
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.257772498444877e-05,
+      "loss": 0.3827,
+      "step": 1954
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2570746325359608e-05,
+      "loss": 0.405,
+      "step": 1955
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2563766325419708e-05,
+      "loss": 0.3746,
+      "step": 1956
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556784988269699e-05,
+      "loss": 0.4077,
+      "step": 1957
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549802317550907e-05,
+      "loss": 0.4581,
+      "step": 1958
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254281831690536e-05,
+      "loss": 0.3851,
+      "step": 1959
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2535832989975766e-05,
+      "loss": 0.4126,
+      "step": 1960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528846340405547e-05,
+      "loss": 0.3712,
+      "step": 1961
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2521858371838792e-05,
+      "loss": 0.3886,
+      "step": 1962
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514869087920292e-05,
+      "loss": 0.397,
+      "step": 1963
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250787849229552e-05,
+      "loss": 0.4772,
+      "step": 1964
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250088658861063e-05,
+      "loss": 0.3855,
+      "step": 1965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493893380512467e-05,
+      "loss": 0.3978,
+      "step": 1966
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2486898871648552e-05,
+      "loss": 0.4003,
+      "step": 1967
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2479903065667072e-05,
+      "loss": 0.3827,
+      "step": 1968
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2472905966216914e-05,
+      "loss": 0.464,
+      "step": 1969
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2465907576947623e-05,
+      "loss": 0.3228,
+      "step": 1970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2458907901509418e-05,
+      "loss": 0.4034,
+      "step": 1971
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451906943553195e-05,
+      "loss": 0.457,
+      "step": 1972
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444904706730521e-05,
+      "loss": 0.3512,
+      "step": 1973
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2437901194693616e-05,
+      "loss": 0.3807,
+      "step": 1974
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2430896411095381e-05,
+      "loss": 0.4627,
+      "step": 1975
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423890359589368e-05,
+      "loss": 0.4254,
+      "step": 1976
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2416883043829801e-05,
+      "loss": 0.4267,
+      "step": 1977
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409874467471556e-05,
+      "loss": 0.4726,
+      "step": 1978
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402864634170167e-05,
+      "loss": 0.412,
+      "step": 1979
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395853547581829e-05,
+      "loss": 0.3543,
+      "step": 1980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238884121136338e-05,
+      "loss": 0.4526,
+      "step": 1981
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2381827629172325e-05,
+      "loss": 0.4548,
+      "step": 1982
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2374812804666805e-05,
+      "loss": 0.3905,
+      "step": 1983
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2367796741505618e-05,
+      "loss": 0.3935,
+      "step": 1984
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2360779443348201e-05,
+      "loss": 0.3559,
+      "step": 1985
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2353760913854644e-05,
+      "loss": 0.4256,
+      "step": 1986
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346741156685666e-05,
+      "loss": 0.4127,
+      "step": 1987
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2339720175502643e-05,
+      "loss": 0.4373,
+      "step": 1988
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332697973967575e-05,
+      "loss": 0.3948,
+      "step": 1989
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325674555743106e-05,
+      "loss": 0.3973,
+      "step": 1990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2318649924492514e-05,
+      "loss": 0.4126,
+      "step": 1991
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311624083879706e-05,
+      "loss": 0.3838,
+      "step": 1992
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304597037569224e-05,
+      "loss": 0.455,
+      "step": 1993
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229756878922624e-05,
+      "loss": 0.4148,
+      "step": 1994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2290539342516541e-05,
+      "loss": 0.3628,
+      "step": 1995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 0.4289,
+      "step": 1996
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2276476868663334e-05,
+      "loss": 0.4649,
+      "step": 1997
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2269443848854528e-05,
+      "loss": 0.3549,
+      "step": 1998
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2262409645348432e-05,
+      "loss": 0.4164,
+      "step": 1999
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2255374261813944e-05,
+      "loss": 0.4016,
+      "step": 2000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2248337701920583e-05,
+      "loss": 0.4569,
+      "step": 2001
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224129996933848e-05,
+      "loss": 0.3383,
+      "step": 2002
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234261067738377e-05,
+      "loss": 0.3241,
+      "step": 2003
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227221000791631e-05,
+      "loss": 0.3424,
+      "step": 2004
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22201797721702e-05,
+      "loss": 0.4041,
+      "step": 2005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2213137385546648e-05,
+      "loss": 0.3556,
+      "step": 2006
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206093844594149e-05,
+      "loss": 0.3647,
+      "step": 2007
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199049152986473e-05,
+      "loss": 0.3972,
+      "step": 2008
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219200331439799e-05,
+      "loss": 0.4828,
+      "step": 2009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2184956332503679e-05,
+      "loss": 0.3964,
+      "step": 2010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177908210979101e-05,
+      "loss": 0.4361,
+      "step": 2011
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2170858953500417e-05,
+      "loss": 0.3711,
+      "step": 2012
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2163808563744388e-05,
+      "loss": 0.4104,
+      "step": 2013
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.215675704538835e-05,
+      "loss": 0.4145,
+      "step": 2014
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2149704402110243e-05,
+      "loss": 0.4766,
+      "step": 2015
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2142650637588582e-05,
+      "loss": 0.3906,
+      "step": 2016
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135595755502475e-05,
+      "loss": 0.3762,
+      "step": 2017
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212853975953161e-05,
+      "loss": 0.4275,
+      "step": 2018
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121482653356253e-05,
+      "loss": 0.4754,
+      "step": 2019
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114424440657255e-05,
+      "loss": 0.3653,
+      "step": 2020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210736512511604e-05,
+      "loss": 0.4114,
+      "step": 2021
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100304710414606e-05,
+      "loss": 0.3747,
+      "step": 2022
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209324320023553e-05,
+      "loss": 0.3975,
+      "step": 2023
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086180598261956e-05,
+      "loss": 0.416,
+      "step": 2024
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.3945,
+      "step": 2025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072052133666733e-05,
+      "loss": 0.3938,
+      "step": 2026
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2064986278414214e-05,
+      "loss": 0.4122,
+      "step": 2027
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2057919346105454e-05,
+      "loss": 0.3975,
+      "step": 2028
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2050851340426422e-05,
+      "loss": 0.436,
+      "step": 2029
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.204378226506365e-05,
+      "loss": 0.3779,
+      "step": 2030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2036712123704231e-05,
+      "loss": 0.3729,
+      "step": 2031
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2029640920035814e-05,
+      "loss": 0.4433,
+      "step": 2032
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2022568657746594e-05,
+      "loss": 0.4057,
+      "step": 2033
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2015495340525329e-05,
+      "loss": 0.4579,
+      "step": 2034
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200842097206132e-05,
+      "loss": 0.4092,
+      "step": 2035
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001345556044417e-05,
+      "loss": 0.4436,
+      "step": 2036
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994269096165021e-05,
+      "loss": 0.4205,
+      "step": 2037
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987191596114073e-05,
+      "loss": 0.4089,
+      "step": 2038
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980113059583058e-05,
+      "loss": 0.4111,
+      "step": 2039
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973033490264e-05,
+      "loss": 0.415,
+      "step": 2040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1965952891849468e-05,
+      "loss": 0.4371,
+      "step": 2041
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1958871268032555e-05,
+      "loss": 0.3576,
+      "step": 2042
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1951788622506905e-05,
+      "loss": 0.3802,
+      "step": 2043
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.194470495896668e-05,
+      "loss": 0.3566,
+      "step": 2044
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1937620281106585e-05,
+      "loss": 0.4433,
+      "step": 2045
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1930534592621845e-05,
+      "loss": 0.4293,
+      "step": 2046
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923447897208216e-05,
+      "loss": 0.3723,
+      "step": 2047
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191636019856198e-05,
+      "loss": 0.4262,
+      "step": 2048
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909271500379947e-05,
+      "loss": 0.3686,
+      "step": 2049
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1902181806359427e-05,
+      "loss": 0.4598,
+      "step": 2050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895091120198283e-05,
+      "loss": 0.3441,
+      "step": 2051
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1887999445594866e-05,
+      "loss": 0.3806,
+      "step": 2052
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880906786248057e-05,
+      "loss": 0.4296,
+      "step": 2053
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.4113,
+      "step": 2054
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1866718528122343e-05,
+      "loss": 0.3805,
+      "step": 2055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1859622936743756e-05,
+      "loss": 0.4329,
+      "step": 2056
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1852526375422404e-05,
+      "loss": 0.4067,
+      "step": 2057
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1845428847859718e-05,
+      "loss": 0.4054,
+      "step": 2058
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1838330357757627e-05,
+      "loss": 0.3762,
+      "step": 2059
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1831230908818563e-05,
+      "loss": 0.398,
+      "step": 2060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1824130504745459e-05,
+      "loss": 0.3611,
+      "step": 2061
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817029149241752e-05,
+      "loss": 0.3989,
+      "step": 2062
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180992684601136e-05,
+      "loss": 0.3412,
+      "step": 2063
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180282359875871e-05,
+      "loss": 0.4052,
+      "step": 2064
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795719411188717e-05,
+      "loss": 0.3221,
+      "step": 2065
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1788614287006785e-05,
+      "loss": 0.4288,
+      "step": 2066
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1781508229918804e-05,
+      "loss": 0.4052,
+      "step": 2067
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177440124363116e-05,
+      "loss": 0.3701,
+      "step": 2068
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1767293331850706e-05,
+      "loss": 0.3697,
+      "step": 2069
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760184498284801e-05,
+      "loss": 0.4405,
+      "step": 2070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753074746641269e-05,
+      "loss": 0.388,
+      "step": 2071
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1745964080628411e-05,
+      "loss": 0.4363,
+      "step": 2072
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1738852503955018e-05,
+      "loss": 0.381,
+      "step": 2073
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1731740020330347e-05,
+      "loss": 0.3619,
+      "step": 2074
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1724626633464127e-05,
+      "loss": 0.4122,
+      "step": 2075
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1717512347066562e-05,
+      "loss": 0.3164,
+      "step": 2076
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710397164848325e-05,
+      "loss": 0.364,
+      "step": 2077
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703281090520553e-05,
+      "loss": 0.3696,
+      "step": 2078
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1696164127794854e-05,
+      "loss": 0.3966,
+      "step": 2079
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168904628038329e-05,
+      "loss": 0.3841,
+      "step": 2080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681927551998398e-05,
+      "loss": 0.3419,
+      "step": 2081
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674807946353159e-05,
+      "loss": 0.3788,
+      "step": 2082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1667687467161025e-05,
+      "loss": 0.3539,
+      "step": 2083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660566118135894e-05,
+      "loss": 0.3984,
+      "step": 2084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1653443902992125e-05,
+      "loss": 0.3607,
+      "step": 2085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1646320825444524e-05,
+      "loss": 0.3629,
+      "step": 2086
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1639196889208348e-05,
+      "loss": 0.3609,
+      "step": 2087
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1632072097999301e-05,
+      "loss": 0.3558,
+      "step": 2088
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1624946455533538e-05,
+      "loss": 0.3259,
+      "step": 2089
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.161781996552765e-05,
+      "loss": 0.4532,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1610692631698675e-05,
+      "loss": 0.4202,
+      "step": 2091
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.160356445776409e-05,
+      "loss": 0.4044,
+      "step": 2092
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1596435447441809e-05,
+      "loss": 0.3611,
+      "step": 2093
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158930560445019e-05,
+      "loss": 0.4249,
+      "step": 2094
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582174932508012e-05,
+      "loss": 0.3676,
+      "step": 2095
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1575043435334494e-05,
+      "loss": 0.4266,
+      "step": 2096
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567911116649288e-05,
+      "loss": 0.3578,
+      "step": 2097
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560777980172473e-05,
+      "loss": 0.3576,
+      "step": 2098
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553644029624543e-05,
+      "loss": 0.4128,
+      "step": 2099
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546509268726437e-05,
+      "loss": 0.4247,
+      "step": 2100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539373701199501e-05,
+      "loss": 0.382,
+      "step": 2101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1532237330765508e-05,
+      "loss": 0.3854,
+      "step": 2102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1525100161146647e-05,
+      "loss": 0.4448,
+      "step": 2103
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1517962196065527e-05,
+      "loss": 0.4214,
+      "step": 2104
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1510823439245169e-05,
+      "loss": 0.4508,
+      "step": 2105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503683894409011e-05,
+      "loss": 0.3266,
+      "step": 2106
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496543565280892e-05,
+      "loss": 0.4154,
+      "step": 2107
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1489402455585078e-05,
+      "loss": 0.3397,
+      "step": 2108
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1482260569046221e-05,
+      "loss": 0.4374,
+      "step": 2109
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1475117909389396e-05,
+      "loss": 0.4135,
+      "step": 2110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467974480340072e-05,
+      "loss": 0.4419,
+      "step": 2111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 0.3809,
+      "step": 2112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1453685328967807e-05,
+      "loss": 0.3609,
+      "step": 2113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446539614097814e-05,
+      "loss": 0.3273,
+      "step": 2114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1439393144741192e-05,
+      "loss": 0.2859,
+      "step": 2115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432245924625407e-05,
+      "loss": 0.39,
+      "step": 2116
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1425097957478309e-05,
+      "loss": 0.352,
+      "step": 2117
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.141794924702813e-05,
+      "loss": 0.3971,
+      "step": 2118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1410799797003507e-05,
+      "loss": 0.392,
+      "step": 2119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1403649611133444e-05,
+      "loss": 0.3765,
+      "step": 2120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396498693147342e-05,
+      "loss": 0.3962,
+      "step": 2121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389347046774977e-05,
+      "loss": 0.4088,
+      "step": 2122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382194675746505e-05,
+      "loss": 0.4958,
+      "step": 2123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1375041583792467e-05,
+      "loss": 0.3832,
+      "step": 2124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367887774643775e-05,
+      "loss": 0.3036,
+      "step": 2125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136073325203171e-05,
+      "loss": 0.4268,
+      "step": 2126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1353578019687938e-05,
+      "loss": 0.396,
+      "step": 2127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134642208134448e-05,
+      "loss": 0.4339,
+      "step": 2128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339265440733736e-05,
+      "loss": 0.3711,
+      "step": 2129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332108101588468e-05,
+      "loss": 0.4902,
+      "step": 2130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1324950067641807e-05,
+      "loss": 0.4933,
+      "step": 2131
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1317791342627239e-05,
+      "loss": 0.3811,
+      "step": 2132
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1310631930278615e-05,
+      "loss": 0.3912,
+      "step": 2133
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1303471834330142e-05,
+      "loss": 0.3569,
+      "step": 2134
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.129631105851639e-05,
+      "loss": 0.3862,
+      "step": 2135
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1289149606572273e-05,
+      "loss": 0.3923,
+      "step": 2136
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281987482233068e-05,
+      "loss": 0.3633,
+      "step": 2137
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12748246892344e-05,
+      "loss": 0.4442,
+      "step": 2138
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1267661231312235e-05,
+      "loss": 0.4125,
+      "step": 2139
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260497112202895e-05,
+      "loss": 0.3562,
+      "step": 2140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1253332335643043e-05,
+      "loss": 0.3595,
+      "step": 2141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1246166905369686e-05,
+      "loss": 0.3961,
+      "step": 2142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1239000825120171e-05,
+      "loss": 0.399,
+      "step": 2143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123183409863219e-05,
+      "loss": 0.3648,
+      "step": 2144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224666729643755e-05,
+      "loss": 0.3963,
+      "step": 2145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217498721893238e-05,
+      "loss": 0.4078,
+      "step": 2146
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.121033007911932e-05,
+      "loss": 0.4134,
+      "step": 2147
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.120316080506103e-05,
+      "loss": 0.4065,
+      "step": 2148
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1195990903457719e-05,
+      "loss": 0.4109,
+      "step": 2149
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1188820378049066e-05,
+      "loss": 0.4212,
+      "step": 2150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1181649232575077e-05,
+      "loss": 0.409,
+      "step": 2151
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1174477470776082e-05,
+      "loss": 0.3954,
+      "step": 2152
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167305096392725e-05,
+      "loss": 0.3854,
+      "step": 2153
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1160132113165985e-05,
+      "loss": 0.4472,
+      "step": 2154
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115295852483714e-05,
+      "loss": 0.4242,
+      "step": 2155
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1145784335147793e-05,
+      "loss": 0.3298,
+      "step": 2156
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138609547839872e-05,
+      "loss": 0.3411,
+      "step": 2157
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113143416665559e-05,
+      "loss": 0.4131,
+      "step": 2158
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1124258195337495e-05,
+      "loss": 0.4091,
+      "step": 2159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117081637628426e-05,
+      "loss": 0.4627,
+      "step": 2160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110990449727154e-05,
+      "loss": 0.4222,
+      "step": 2161
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1102726778010288e-05,
+      "loss": 0.404,
+      "step": 2162
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1095548483588432e-05,
+      "loss": 0.4803,
+      "step": 2163
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1088369617750021e-05,
+      "loss": 0.4272,
+      "step": 2164
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1081190184239418e-05,
+      "loss": 0.4051,
+      "step": 2165
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1074010186801273e-05,
+      "loss": 0.4229,
+      "step": 2166
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1066829629180528e-05,
+      "loss": 0.3974,
+      "step": 2167
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1059648515122426e-05,
+      "loss": 0.4219,
+      "step": 2168
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1052466848372487e-05,
+      "loss": 0.463,
+      "step": 2169
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.3919,
+      "step": 2170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038101871780671e-05,
+      "loss": 0.4342,
+      "step": 2171
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030918569431277e-05,
+      "loss": 0.3803,
+      "step": 2172
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023734729375028e-05,
+      "loss": 0.3935,
+      "step": 2173
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016550355358872e-05,
+      "loss": 0.4238,
+      "step": 2174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009365451130033e-05,
+      "loss": 0.3654,
+      "step": 2175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002180020436025e-05,
+      "loss": 0.4206,
+      "step": 2176
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0994994067024619e-05,
+      "loss": 0.4073,
+      "step": 2177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0987807594643867e-05,
+      "loss": 0.4598,
+      "step": 2178
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0980620607042097e-05,
+      "loss": 0.3399,
+      "step": 2179
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0973433107967901e-05,
+      "loss": 0.3281,
+      "step": 2180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0966245101170128e-05,
+      "loss": 0.3649,
+      "step": 2181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0959056590397912e-05,
+      "loss": 0.3396,
+      "step": 2182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095186757940063e-05,
+      "loss": 0.3758,
+      "step": 2183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944678071927936e-05,
+      "loss": 0.3857,
+      "step": 2184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937488071729734e-05,
+      "loss": 0.4382,
+      "step": 2185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930297582556182e-05,
+      "loss": 0.3983,
+      "step": 2186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0923106608157703e-05,
+      "loss": 0.4257,
+      "step": 2187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0915915152284966e-05,
+      "loss": 0.3811,
+      "step": 2188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090872321868889e-05,
+      "loss": 0.4021,
+      "step": 2189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0901530811120655e-05,
+      "loss": 0.337,
+      "step": 2190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0894337933331669e-05,
+      "loss": 0.4303,
+      "step": 2191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0887144589073601e-05,
+      "loss": 0.3511,
+      "step": 2192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0879950782098359e-05,
+      "loss": 0.413,
+      "step": 2193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0872756516158087e-05,
+      "loss": 0.4553,
+      "step": 2194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0865561795005176e-05,
+      "loss": 0.3585,
+      "step": 2195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0858366622392252e-05,
+      "loss": 0.4478,
+      "step": 2196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851171002072169e-05,
+      "loss": 0.3953,
+      "step": 2197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.084397493779803e-05,
+      "loss": 0.4415,
+      "step": 2198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083677843332316e-05,
+      "loss": 0.3455,
+      "step": 2199
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829581492401103e-05,
+      "loss": 0.3366,
+      "step": 2200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822384118785654e-05,
+      "loss": 0.3702,
+      "step": 2201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815186316230814e-05,
+      "loss": 0.4574,
+      "step": 2202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807988088490822e-05,
+      "loss": 0.4005,
+      "step": 2203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0800789439320128e-05,
+      "loss": 0.3421,
+      "step": 2204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0793590372473403e-05,
+      "loss": 0.3558,
+      "step": 2205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786390891705544e-05,
+      "loss": 0.3463,
+      "step": 2206
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0779191000771661e-05,
+      "loss": 0.4389,
+      "step": 2207
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0771990703427067e-05,
+      "loss": 0.3633,
+      "step": 2208
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0764790003427302e-05,
+      "loss": 0.4076,
+      "step": 2209
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0757588904528106e-05,
+      "loss": 0.4095,
+      "step": 2210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0750387410485431e-05,
+      "loss": 0.3719,
+      "step": 2211
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0743185525055437e-05,
+      "loss": 0.3329,
+      "step": 2212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735983251994486e-05,
+      "loss": 0.3326,
+      "step": 2213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0728780595059138e-05,
+      "loss": 0.3969,
+      "step": 2214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0721577558006164e-05,
+      "loss": 0.4042,
+      "step": 2215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0714374144592516e-05,
+      "loss": 0.4093,
+      "step": 2216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0707170358575365e-05,
+      "loss": 0.3961,
+      "step": 2217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699966203712057e-05,
+      "loss": 0.3367,
+      "step": 2218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692761683760135e-05,
+      "loss": 0.3907,
+      "step": 2219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685556802477347e-05,
+      "loss": 0.3431,
+      "step": 2220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678351563621603e-05,
+      "loss": 0.3897,
+      "step": 2221
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.067114597095102e-05,
+      "loss": 0.452,
+      "step": 2222
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0663940028223896e-05,
+      "loss": 0.4185,
+      "step": 2223
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0656733739198705e-05,
+      "loss": 0.3894,
+      "step": 2224
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0649527107634107e-05,
+      "loss": 0.4587,
+      "step": 2225
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0642320137288944e-05,
+      "loss": 0.3745,
+      "step": 2226
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063511283192222e-05,
+      "loss": 0.407,
+      "step": 2227
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 0.4194,
+      "step": 2228
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0620697231161044e-05,
+      "loss": 0.4328,
+      "step": 2229
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061348894328548e-05,
+      "loss": 0.3404,
+      "step": 2230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0606280335426142e-05,
+      "loss": 0.355,
+      "step": 2231
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0599071411342903e-05,
+      "loss": 0.3676,
+      "step": 2232
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591862174795793e-05,
+      "loss": 0.4485,
+      "step": 2233
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584652629545011e-05,
+      "loss": 0.4068,
+      "step": 2234
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057744277935091e-05,
+      "loss": 0.4158,
+      "step": 2235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0570232627974008e-05,
+      "loss": 0.39,
+      "step": 2236
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0563022179174982e-05,
+      "loss": 0.3544,
+      "step": 2237
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0555811436714651e-05,
+      "loss": 0.3533,
+      "step": 2238
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0548600404354008e-05,
+      "loss": 0.3985,
+      "step": 2239
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 0.4006,
+      "step": 2240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0534177484976443e-05,
+      "loss": 0.3427,
+      "step": 2241
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526965605482236e-05,
+      "loss": 0.3261,
+      "step": 2242
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0519753451133128e-05,
+      "loss": 0.454,
+      "step": 2243
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0512541025690838e-05,
+      "loss": 0.326,
+      "step": 2244
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050532833291723e-05,
+      "loss": 0.3571,
+      "step": 2245
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498115376574294e-05,
+      "loss": 0.414,
+      "step": 2246
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490902160424175e-05,
+      "loss": 0.3689,
+      "step": 2247
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0483688688229137e-05,
+      "loss": 0.4254,
+      "step": 2248
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476474963751591e-05,
+      "loss": 0.3414,
+      "step": 2249
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046926099075407e-05,
+      "loss": 0.3838,
+      "step": 2250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0462046772999242e-05,
+      "loss": 0.4446,
+      "step": 2251
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0454832314249904e-05,
+      "loss": 0.3747,
+      "step": 2252
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.044761761826897e-05,
+      "loss": 0.3306,
+      "step": 2253
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0440402688819486e-05,
+      "loss": 0.3924,
+      "step": 2254
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0433187529664623e-05,
+      "loss": 0.3419,
+      "step": 2255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042597214456766e-05,
+      "loss": 0.4154,
+      "step": 2256
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0418756537291996e-05,
+      "loss": 0.3602,
+      "step": 2257
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041154071160116e-05,
+      "loss": 0.3885,
+      "step": 2258
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0404324671258781e-05,
+      "loss": 0.4195,
+      "step": 2259
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397108420028598e-05,
+      "loss": 0.4204,
+      "step": 2260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0389891961674475e-05,
+      "loss": 0.4249,
+      "step": 2261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382675299960367e-05,
+      "loss": 0.3861,
+      "step": 2262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375458438650348e-05,
+      "loss": 0.4074,
+      "step": 2263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036824138150859e-05,
+      "loss": 0.3403,
+      "step": 2264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0361024132299364e-05,
+      "loss": 0.4185,
+      "step": 2265
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0353806694787053e-05,
+      "loss": 0.4271,
+      "step": 2266
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0346589072736124e-05,
+      "loss": 0.4211,
+      "step": 2267
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0339371269911147e-05,
+      "loss": 0.4109,
+      "step": 2268
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.033215329007679e-05,
+      "loss": 0.3678,
+      "step": 2269
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0324935136997807e-05,
+      "loss": 0.4307,
+      "step": 2270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317716814439043e-05,
+      "loss": 0.3808,
+      "step": 2271
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310498326165439e-05,
+      "loss": 0.3591,
+      "step": 2272
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0303279675942008e-05,
+      "loss": 0.3635,
+      "step": 2273
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296060867533865e-05,
+      "loss": 0.3938,
+      "step": 2274
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0288841904706193e-05,
+      "loss": 0.4361,
+      "step": 2275
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0281622791224257e-05,
+      "loss": 0.413,
+      "step": 2276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0274403530853418e-05,
+      "loss": 0.3681,
+      "step": 2277
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0267184127359088e-05,
+      "loss": 0.389,
+      "step": 2278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259964584506769e-05,
+      "loss": 0.4412,
+      "step": 2279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252744906062038e-05,
+      "loss": 0.3657,
+      "step": 2280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.024552509579053e-05,
+      "loss": 0.3619,
+      "step": 2281
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.023830515745796e-05,
+      "loss": 0.3427,
+      "step": 2282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0231085094830107e-05,
+      "loss": 0.4342,
+      "step": 2283
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0223864911672807e-05,
+      "loss": 0.3675,
+      "step": 2284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0216644611751975e-05,
+      "loss": 0.4424,
+      "step": 2285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 0.3033,
+      "step": 2286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202203676683622e-05,
+      "loss": 0.3464,
+      "step": 2287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0194983049068212e-05,
+      "loss": 0.3572,
+      "step": 2288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187762319753476e-05,
+      "loss": 0.4336,
+      "step": 2289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0180541492505605e-05,
+      "loss": 0.3729,
+      "step": 2290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173320571090844e-05,
+      "loss": 0.3419,
+      "step": 2291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166099559275475e-05,
+      "loss": 0.3677,
+      "step": 2292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158878460825847e-05,
+      "loss": 0.3961,
+      "step": 2293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0151657279508335e-05,
+      "loss": 0.3504,
+      "step": 2294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014443601908937e-05,
+      "loss": 0.398,
+      "step": 2295
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0137214683335414e-05,
+      "loss": 0.3261,
+      "step": 2296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.012999327601298e-05,
+      "loss": 0.4237,
+      "step": 2297
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0122771800888606e-05,
+      "loss": 0.4208,
+      "step": 2298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0115550261728874e-05,
+      "loss": 0.4013,
+      "step": 2299
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0108328662300399e-05,
+      "loss": 0.4574,
+      "step": 2300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101107006369823e-05,
+      "loss": 0.3773,
+      "step": 2301
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093885297703825e-05,
+      "loss": 0.3734,
+      "step": 2302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0086663540069095e-05,
+      "loss": 0.395,
+      "step": 2303
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079441737232369e-05,
+      "loss": 0.3896,
+      "step": 2304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0072219892960395e-05,
+      "loss": 0.409,
+      "step": 2305
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064998011019945e-05,
+      "loss": 0.4083,
+      "step": 2306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0057776095177809e-05,
+      "loss": 0.4098,
+      "step": 2307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050554149200797e-05,
+      "loss": 0.4468,
+      "step": 2308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043332176855736e-05,
+      "loss": 0.3588,
+      "step": 2309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036110181909466e-05,
+      "loss": 0.4088,
+      "step": 2310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.002888816812883e-05,
+      "loss": 0.3274,
+      "step": 2311
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0021666139280697e-05,
+      "loss": 0.3439,
+      "step": 2312
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.001444409913193e-05,
+      "loss": 0.3658,
+      "step": 2313
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0007222051449406e-05,
+      "loss": 0.4425,
+      "step": 2314
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.3623,
+      "step": 2315
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992777948550599e-06,
+      "loss": 0.388,
+      "step": 2316
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985555900868071e-06,
+      "loss": 0.408,
+      "step": 2317
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.978333860719307e-06,
+      "loss": 0.3703,
+      "step": 2318
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971111831871173e-06,
+      "loss": 0.4306,
+      "step": 2319
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.963889818090539e-06,
+      "loss": 0.372,
+      "step": 2320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956667823144266e-06,
+      "loss": 0.3191,
+      "step": 2321
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949445850799206e-06,
+      "loss": 0.3544,
+      "step": 2322
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.942223904822194e-06,
+      "loss": 0.4114,
+      "step": 2323
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93500198898006e-06,
+      "loss": 0.4055,
+      "step": 2324
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.927780107039605e-06,
+      "loss": 0.3302,
+      "step": 2325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.920558262767633e-06,
+      "loss": 0.3272,
+      "step": 2326
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.913336459930909e-06,
+      "loss": 0.3946,
+      "step": 2327
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.90611470229618e-06,
+      "loss": 0.3966,
+      "step": 2328
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.898892993630175e-06,
+      "loss": 0.3876,
+      "step": 2329
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891671337699603e-06,
+      "loss": 0.428,
+      "step": 2330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884449738271127e-06,
+      "loss": 0.4388,
+      "step": 2331
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877228199111399e-06,
+      "loss": 0.3498,
+      "step": 2332
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870006723987024e-06,
+      "loss": 0.3631,
+      "step": 2333
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862785316664587e-06,
+      "loss": 0.4035,
+      "step": 2334
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855563980910634e-06,
+      "loss": 0.4556,
+      "step": 2335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848342720491666e-06,
+      "loss": 0.3634,
+      "step": 2336
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841121539174154e-06,
+      "loss": 0.3729,
+      "step": 2337
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833900440724526e-06,
+      "loss": 0.4398,
+      "step": 2338
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826679428909161e-06,
+      "loss": 0.3129,
+      "step": 2339
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819458507494395e-06,
+      "loss": 0.3767,
+      "step": 2340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.812237680246527e-06,
+      "loss": 0.4198,
+      "step": 2341
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.805016950931791e-06,
+      "loss": 0.3529,
+      "step": 2342
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.797796323316381e-06,
+      "loss": 0.3901,
+      "step": 2343
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 0.4451,
+      "step": 2344
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.783355388248026e-06,
+      "loss": 0.3472,
+      "step": 2345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776135088327195e-06,
+      "loss": 0.4096,
+      "step": 2346
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.768914905169898e-06,
+      "loss": 0.3457,
+      "step": 2347
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.761694842542042e-06,
+      "loss": 0.3248,
+      "step": 2348
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.754474904209474e-06,
+      "loss": 0.3742,
+      "step": 2349
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747255093937966e-06,
+      "loss": 0.3889,
+      "step": 2350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.740035415493234e-06,
+      "loss": 0.4075,
+      "step": 2351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.732815872640915e-06,
+      "loss": 0.4631,
+      "step": 2352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.725596469146585e-06,
+      "loss": 0.3908,
+      "step": 2353
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.718377208775744e-06,
+      "loss": 0.3238,
+      "step": 2354
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.711158095293812e-06,
+      "loss": 0.3462,
+      "step": 2355
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.711158095293812e-06,
+      "loss": 0.3652,
+      "step": 2356
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.703939132466138e-06,
+      "loss": 0.4049,
+      "step": 2357
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.696720324057995e-06,
+      "loss": 0.4544,
+      "step": 2358
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.689501673834566e-06,
+      "loss": 0.4463,
+      "step": 2359
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.682283185560957e-06,
+      "loss": 0.3041,
+      "step": 2360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.675064863002196e-06,
+      "loss": 0.3783,
+      "step": 2361
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667846709923212e-06,
+      "loss": 0.3426,
+      "step": 2362
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660628730088858e-06,
+      "loss": 0.3001,
+      "step": 2363
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65341092726388e-06,
+      "loss": 0.4994,
+      "step": 2364
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.646193305212949e-06,
+      "loss": 0.3958,
+      "step": 2365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.638975867700638e-06,
+      "loss": 0.4156,
+      "step": 2366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631758618491415e-06,
+      "loss": 0.3853,
+      "step": 2367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624541561349654e-06,
+      "loss": 0.4446,
+      "step": 2368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617324700039635e-06,
+      "loss": 0.4325,
+      "step": 2369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.610108038325529e-06,
+      "loss": 0.3976,
+      "step": 2370
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.602891579971405e-06,
+      "loss": 0.4292,
+      "step": 2371
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.595675328741222e-06,
+      "loss": 0.359,
+      "step": 2372
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.588459288398841e-06,
+      "loss": 0.4121,
+      "step": 2373
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.581243462708007e-06,
+      "loss": 0.3825,
+      "step": 2374
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.574027855432347e-06,
+      "loss": 0.3516,
+      "step": 2375
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566812470335379e-06,
+      "loss": 0.3831,
+      "step": 2376
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.559597311180515e-06,
+      "loss": 0.4206,
+      "step": 2377
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.552382381731033e-06,
+      "loss": 0.3685,
+      "step": 2378
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545167685750098e-06,
+      "loss": 0.3847,
+      "step": 2379
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.537953227000761e-06,
+      "loss": 0.3815,
+      "step": 2380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530739009245934e-06,
+      "loss": 0.4209,
+      "step": 2381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523525036248415e-06,
+      "loss": 0.4227,
+      "step": 2382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.516311311770867e-06,
+      "loss": 0.3303,
+      "step": 2383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.509097839575829e-06,
+      "loss": 0.3298,
+      "step": 2384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50188462342571e-06,
+      "loss": 0.4182,
+      "step": 2385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.494671667082776e-06,
+      "loss": 0.3816,
+      "step": 2386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.487458974309163e-06,
+      "loss": 0.3883,
+      "step": 2387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.480246548866876e-06,
+      "loss": 0.4355,
+      "step": 2388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.473034394517769e-06,
+      "loss": 0.3543,
+      "step": 2389
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.465822515023562e-06,
+      "loss": 0.3985,
+      "step": 2390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 0.3906,
+      "step": 2391
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.451399595645994e-06,
+      "loss": 0.3618,
+      "step": 2392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444188563285352e-06,
+      "loss": 0.4427,
+      "step": 2393
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436977820825023e-06,
+      "loss": 0.3653,
+      "step": 2394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.429767372025992e-06,
+      "loss": 0.4503,
+      "step": 2395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.422557220649094e-06,
+      "loss": 0.3924,
+      "step": 2396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415347370454992e-06,
+      "loss": 0.3574,
+      "step": 2397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.408137825204207e-06,
+      "loss": 0.3846,
+      "step": 2398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.4009285886571e-06,
+      "loss": 0.3863,
+      "step": 2399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.393719664573861e-06,
+      "loss": 0.3856,
+      "step": 2400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.386511056714527e-06,
+      "loss": 0.3809,
+      "step": 2401
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.379302768838959e-06,
+      "loss": 0.3719,
+      "step": 2402
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.3588,
+      "step": 2403
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.364887168077781e-06,
+      "loss": 0.3698,
+      "step": 2404
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.35767986271106e-06,
+      "loss": 0.3614,
+      "step": 2405
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.350472892365891e-06,
+      "loss": 0.3787,
+      "step": 2406
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343266260801297e-06,
+      "loss": 0.4144,
+      "step": 2407
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.336059971776108e-06,
+      "loss": 0.3892,
+      "step": 2408
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.328854029048985e-06,
+      "loss": 0.4785,
+      "step": 2409
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321648436378402e-06,
+      "loss": 0.3824,
+      "step": 2410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314443197522658e-06,
+      "loss": 0.3514,
+      "step": 2411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307238316239868e-06,
+      "loss": 0.4231,
+      "step": 2412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300033796287948e-06,
+      "loss": 0.3835,
+      "step": 2413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.292829641424637e-06,
+      "loss": 0.3891,
+      "step": 2414
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285625855407485e-06,
+      "loss": 0.3884,
+      "step": 2415
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.278422441993841e-06,
+      "loss": 0.3948,
+      "step": 2416
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.271219404940862e-06,
+      "loss": 0.3836,
+      "step": 2417
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.264016748005517e-06,
+      "loss": 0.3957,
+      "step": 2418
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.256814474944566e-06,
+      "loss": 0.3992,
+      "step": 2419
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.249612589514574e-06,
+      "loss": 0.348,
+      "step": 2420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.242411095471897e-06,
+      "loss": 0.4359,
+      "step": 2421
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235209996572701e-06,
+      "loss": 0.4318,
+      "step": 2422
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228009296572937e-06,
+      "loss": 0.4052,
+      "step": 2423
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220808999228344e-06,
+      "loss": 0.4222,
+      "step": 2424
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213609108294456e-06,
+      "loss": 0.4219,
+      "step": 2425
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206409627526599e-06,
+      "loss": 0.3885,
+      "step": 2426
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199210560679877e-06,
+      "loss": 0.3588,
+      "step": 2427
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192011911509183e-06,
+      "loss": 0.3649,
+      "step": 2428
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184813683769188e-06,
+      "loss": 0.4065,
+      "step": 2429
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17761588121435e-06,
+      "loss": 0.4647,
+      "step": 2430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.170418507598902e-06,
+      "loss": 0.4266,
+      "step": 2431
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.163221566676847e-06,
+      "loss": 0.41,
+      "step": 2432
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.156025062201971e-06,
+      "loss": 0.3771,
+      "step": 2433
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.148828997927833e-06,
+      "loss": 0.345,
+      "step": 2434
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.141633377607753e-06,
+      "loss": 0.3061,
+      "step": 2435
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.134438204994825e-06,
+      "loss": 0.3767,
+      "step": 2436
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.127243483841915e-06,
+      "loss": 0.3536,
+      "step": 2437
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120049217901645e-06,
+      "loss": 0.3815,
+      "step": 2438
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112855410926404e-06,
+      "loss": 0.4291,
+      "step": 2439
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105662066668333e-06,
+      "loss": 0.4189,
+      "step": 2440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098469188879348e-06,
+      "loss": 0.3468,
+      "step": 2441
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091276781311112e-06,
+      "loss": 0.3797,
+      "step": 2442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.084084847715037e-06,
+      "loss": 0.3492,
+      "step": 2443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.076893391842299e-06,
+      "loss": 0.3858,
+      "step": 2444
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069702417443821e-06,
+      "loss": 0.3775,
+      "step": 2445
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.06251192827027e-06,
+      "loss": 0.337,
+      "step": 2446
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.055321928072068e-06,
+      "loss": 0.3549,
+      "step": 2447
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.04813242059937e-06,
+      "loss": 0.4027,
+      "step": 2448
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.040943409602092e-06,
+      "loss": 0.3906,
+      "step": 2449
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.033754898829875e-06,
+      "loss": 0.3561,
+      "step": 2450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.026566892032104e-06,
+      "loss": 0.3738,
+      "step": 2451
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019379392957903e-06,
+      "loss": 0.3696,
+      "step": 2452
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.012192405356134e-06,
+      "loss": 0.3534,
+      "step": 2453
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.005005932975386e-06,
+      "loss": 0.3824,
+      "step": 2454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997819979563982e-06,
+      "loss": 0.4023,
+      "step": 2455
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990634548869968e-06,
+      "loss": 0.3414,
+      "step": 2456
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983449644641131e-06,
+      "loss": 0.3428,
+      "step": 2457
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976265270624975e-06,
+      "loss": 0.4051,
+      "step": 2458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969081430568724e-06,
+      "loss": 0.3921,
+      "step": 2459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96189812821933e-06,
+      "loss": 0.4307,
+      "step": 2460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.39,
+      "step": 2461
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.947533151627514e-06,
+      "loss": 0.428,
+      "step": 2462
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.940351484877576e-06,
+      "loss": 0.3783,
+      "step": 2463
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.933170370819473e-06,
+      "loss": 0.3937,
+      "step": 2464
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.92598981319873e-06,
+      "loss": 0.3523,
+      "step": 2465
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918809815760585e-06,
+      "loss": 0.351,
+      "step": 2466
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91163038224998e-06,
+      "loss": 0.3957,
+      "step": 2467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904451516411572e-06,
+      "loss": 0.3732,
+      "step": 2468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897273221989715e-06,
+      "loss": 0.3769,
+      "step": 2469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.890095502728463e-06,
+      "loss": 0.4012,
+      "step": 2470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882918362371573e-06,
+      "loss": 0.4295,
+      "step": 2471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875741804662507e-06,
+      "loss": 0.3496,
+      "step": 2472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868565833344411e-06,
+      "loss": 0.3731,
+      "step": 2473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861390452160133e-06,
+      "loss": 0.3836,
+      "step": 2474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.854215664852207e-06,
+      "loss": 0.4224,
+      "step": 2475
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.847041475162863e-06,
+      "loss": 0.3996,
+      "step": 2476
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.839867886834022e-06,
+      "loss": 0.435,
+      "step": 2477
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.832694903607275e-06,
+      "loss": 0.4607,
+      "step": 2478
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.825522529223921e-06,
+      "loss": 0.3433,
+      "step": 2479
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.818350767424925e-06,
+      "loss": 0.4649,
+      "step": 2480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811179621950937e-06,
+      "loss": 0.3757,
+      "step": 2481
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.804009096542281e-06,
+      "loss": 0.3687,
+      "step": 2482
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796839194938973e-06,
+      "loss": 0.4722,
+      "step": 2483
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789669920880683e-06,
+      "loss": 0.3913,
+      "step": 2484
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782501278106767e-06,
+      "loss": 0.3925,
+      "step": 2485
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775333270356245e-06,
+      "loss": 0.3944,
+      "step": 2486
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.768165901367811e-06,
+      "loss": 0.3254,
+      "step": 2487
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76099917487983e-06,
+      "loss": 0.3928,
+      "step": 2488
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.753833094630317e-06,
+      "loss": 0.3887,
+      "step": 2489
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.746667664356957e-06,
+      "loss": 0.3424,
+      "step": 2490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.739502887797108e-06,
+      "loss": 0.4199,
+      "step": 2491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.732338768687768e-06,
+      "loss": 0.4105,
+      "step": 2492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.725175310765606e-06,
+      "loss": 0.3416,
+      "step": 2493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.718012517766932e-06,
+      "loss": 0.4054,
+      "step": 2494
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.710850393427729e-06,
+      "loss": 0.3982,
+      "step": 2495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.703688941483616e-06,
+      "loss": 0.4203,
+      "step": 2496
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69652816566986e-06,
+      "loss": 0.3686,
+      "step": 2497
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.689368069721389e-06,
+      "loss": 0.356,
+      "step": 2498
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.682208657372767e-06,
+      "loss": 0.3593,
+      "step": 2499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.675049932358198e-06,
+      "loss": 0.4594,
+      "step": 2500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66789189841153e-06,
+      "loss": 0.4073,
+      "step": 2501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660734559266268e-06,
+      "loss": 0.3676,
+      "step": 2502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653577918655522e-06,
+      "loss": 0.3848,
+      "step": 2503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646421980312067e-06,
+      "loss": 0.4126,
+      "step": 2504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63926674796829e-06,
+      "loss": 0.3433,
+      "step": 2505
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.632112225356227e-06,
+      "loss": 0.3237,
+      "step": 2506
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.624958416207534e-06,
+      "loss": 0.4406,
+      "step": 2507
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.617805324253497e-06,
+      "loss": 0.3967,
+      "step": 2508
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.610652953225025e-06,
+      "loss": 0.3765,
+      "step": 2509
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.603501306852662e-06,
+      "loss": 0.4042,
+      "step": 2510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.596350388866557e-06,
+      "loss": 0.4125,
+      "step": 2511
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.589200202996497e-06,
+      "loss": 0.3964,
+      "step": 2512
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58205075297187e-06,
+      "loss": 0.3799,
+      "step": 2513
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574902042521693e-06,
+      "loss": 0.3445,
+      "step": 2514
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567754075374596e-06,
+      "loss": 0.3998,
+      "step": 2515
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.560606855258808e-06,
+      "loss": 0.4129,
+      "step": 2516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55346038590219e-06,
+      "loss": 0.3906,
+      "step": 2517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.546314671032194e-06,
+      "loss": 0.4074,
+      "step": 2518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 0.3299,
+      "step": 2519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53202551965993e-06,
+      "loss": 0.3164,
+      "step": 2520
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.524882090610607e-06,
+      "loss": 0.3746,
+      "step": 2521
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.51773943095378e-06,
+      "loss": 0.3074,
+      "step": 2522
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.510597544414927e-06,
+      "loss": 0.3661,
+      "step": 2523
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.503456434719108e-06,
+      "loss": 0.3195,
+      "step": 2524
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.496316105590992e-06,
+      "loss": 0.3449,
+      "step": 2525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.489176560754835e-06,
+      "loss": 0.4104,
+      "step": 2526
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482037803934477e-06,
+      "loss": 0.4277,
+      "step": 2527
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474899838853355e-06,
+      "loss": 0.3976,
+      "step": 2528
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467762669234496e-06,
+      "loss": 0.4141,
+      "step": 2529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.4606262988005e-06,
+      "loss": 0.3481,
+      "step": 2530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453490731273566e-06,
+      "loss": 0.3298,
+      "step": 2531
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446355970375457e-06,
+      "loss": 0.3795,
+      "step": 2532
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43922201982753e-06,
+      "loss": 0.3684,
+      "step": 2533
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432088883350715e-06,
+      "loss": 0.4322,
+      "step": 2534
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424956564665508e-06,
+      "loss": 0.3954,
+      "step": 2535
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.417825067491991e-06,
+      "loss": 0.3376,
+      "step": 2536
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.410694395549815e-06,
+      "loss": 0.3962,
+      "step": 2537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.403564552558193e-06,
+      "loss": 0.3686,
+      "step": 2538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.39643554223591e-06,
+      "loss": 0.4019,
+      "step": 2539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.389307368301329e-06,
+      "loss": 0.3586,
+      "step": 2540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 0.4015,
+      "step": 2541
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375053544466467e-06,
+      "loss": 0.4019,
+      "step": 2542
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3679279020007e-06,
+      "loss": 0.3839,
+      "step": 2543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.360803110791653e-06,
+      "loss": 0.3197,
+      "step": 2544
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35367917455548e-06,
+      "loss": 0.3862,
+      "step": 2545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346556097007878e-06,
+      "loss": 0.3753,
+      "step": 2546
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339433881864106e-06,
+      "loss": 0.3256,
+      "step": 2547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.332312532838978e-06,
+      "loss": 0.3737,
+      "step": 2548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325192053646843e-06,
+      "loss": 0.4155,
+      "step": 2549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318072448001609e-06,
+      "loss": 0.3605,
+      "step": 2550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.31095371961671e-06,
+      "loss": 0.3507,
+      "step": 2551
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.30383587220515e-06,
+      "loss": 0.276,
+      "step": 2552
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.29671890947945e-06,
+      "loss": 0.358,
+      "step": 2553
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.28960283515168e-06,
+      "loss": 0.4004,
+      "step": 2554
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.28248765293344e-06,
+      "loss": 0.3253,
+      "step": 2555
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.275373366535878e-06,
+      "loss": 0.3914,
+      "step": 2556
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.268259979669656e-06,
+      "loss": 0.3332,
+      "step": 2557
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.261147496044982e-06,
+      "loss": 0.3864,
+      "step": 2558
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25403591937159e-06,
+      "loss": 0.3458,
+      "step": 2559
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246925253358734e-06,
+      "loss": 0.4394,
+      "step": 2560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239815501715202e-06,
+      "loss": 0.3881,
+      "step": 2561
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.232706668149296e-06,
+      "loss": 0.3866,
+      "step": 2562
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225598756368844e-06,
+      "loss": 0.3379,
+      "step": 2563
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.218491770081198e-06,
+      "loss": 0.3946,
+      "step": 2564
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.211385712993219e-06,
+      "loss": 0.4632,
+      "step": 2565
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.204280588811283e-06,
+      "loss": 0.3749,
+      "step": 2566
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.197176401241292e-06,
+      "loss": 0.3662,
+      "step": 2567
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.190073153988643e-06,
+      "loss": 0.3819,
+      "step": 2568
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.182970850758255e-06,
+      "loss": 0.3821,
+      "step": 2569
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.175869495254541e-06,
+      "loss": 0.4403,
+      "step": 2570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.168769091181439e-06,
+      "loss": 0.4189,
+      "step": 2571
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161669642242378e-06,
+      "loss": 0.3763,
+      "step": 2572
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154571152140287e-06,
+      "loss": 0.4538,
+      "step": 2573
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147473624577598e-06,
+      "loss": 0.3896,
+      "step": 2574
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140377063256249e-06,
+      "loss": 0.3778,
+      "step": 2575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.133281471877659e-06,
+      "loss": 0.4266,
+      "step": 2576
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 0.2953,
+      "step": 2577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119093213751946e-06,
+      "loss": 0.3859,
+      "step": 2578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112000554405137e-06,
+      "loss": 0.3817,
+      "step": 2579
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.104908879801722e-06,
+      "loss": 0.401,
+      "step": 2580
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.097818193640571e-06,
+      "loss": 0.4251,
+      "step": 2581
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.090728499620058e-06,
+      "loss": 0.3834,
+      "step": 2582
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.083639801438021e-06,
+      "loss": 0.4061,
+      "step": 2583
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.076552102791789e-06,
+      "loss": 0.4347,
+      "step": 2584
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.069465407378155e-06,
+      "loss": 0.441,
+      "step": 2585
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.062379718893417e-06,
+      "loss": 0.357,
+      "step": 2586
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.055295041033324e-06,
+      "loss": 0.3546,
+      "step": 2587
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0482113774931e-06,
+      "loss": 0.3876,
+      "step": 2588
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041128731967445e-06,
+      "loss": 0.4055,
+      "step": 2589
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034047108150534e-06,
+      "loss": 0.3485,
+      "step": 2590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026966509736001e-06,
+      "loss": 0.359,
+      "step": 2591
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019886940416947e-06,
+      "loss": 0.3191,
+      "step": 2592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012808403885927e-06,
+      "loss": 0.3562,
+      "step": 2593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00573090383498e-06,
+      "loss": 0.3173,
+      "step": 2594
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.998654443955586e-06,
+      "loss": 0.4015,
+      "step": 2595
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.991579027938681e-06,
+      "loss": 0.296,
+      "step": 2596
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.984504659474673e-06,
+      "loss": 0.3486,
+      "step": 2597
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.977431342253409e-06,
+      "loss": 0.354,
+      "step": 2598
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.97035907996419e-06,
+      "loss": 0.3955,
+      "step": 2599
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.963287876295769e-06,
+      "loss": 0.3487,
+      "step": 2600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956217734936353e-06,
+      "loss": 0.4913,
+      "step": 2601
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.949148659573583e-06,
+      "loss": 0.3809,
+      "step": 2602
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942080653894551e-06,
+      "loss": 0.3593,
+      "step": 2603
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.935013721585787e-06,
+      "loss": 0.3527,
+      "step": 2604
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92794786633327e-06,
+      "loss": 0.408,
+      "step": 2605
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.3706,
+      "step": 2606
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91381940173805e-06,
+      "loss": 0.407,
+      "step": 2607
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.906756799764472e-06,
+      "loss": 0.2915,
+      "step": 2608
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899695289585397e-06,
+      "loss": 0.3913,
+      "step": 2609
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.892634874883964e-06,
+      "loss": 0.4912,
+      "step": 2610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.885575559342751e-06,
+      "loss": 0.3698,
+      "step": 2611
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.878517346643748e-06,
+      "loss": 0.408,
+      "step": 2612
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.871460240468395e-06,
+      "loss": 0.3467,
+      "step": 2613
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.86440424449753e-06,
+      "loss": 0.3728,
+      "step": 2614
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857349362411421e-06,
+      "loss": 0.3994,
+      "step": 2615
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85029559788976e-06,
+      "loss": 0.3979,
+      "step": 2616
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843242954611652e-06,
+      "loss": 0.3918,
+      "step": 2617
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.836191436255615e-06,
+      "loss": 0.4088,
+      "step": 2618
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.829141046499583e-06,
+      "loss": 0.3713,
+      "step": 2619
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822091789020902e-06,
+      "loss": 0.433,
+      "step": 2620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815043667496323e-06,
+      "loss": 0.2681,
+      "step": 2621
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807996685602013e-06,
+      "loss": 0.3938,
+      "step": 2622
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80095084701353e-06,
+      "loss": 0.4023,
+      "step": 2623
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.793906155405853e-06,
+      "loss": 0.4174,
+      "step": 2624
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.786862614453356e-06,
+      "loss": 0.3497,
+      "step": 2625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.779820227829804e-06,
+      "loss": 0.3457,
+      "step": 2626
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.772778999208369e-06,
+      "loss": 0.4266,
+      "step": 2627
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.765738932261624e-06,
+      "loss": 0.3689,
+      "step": 2628
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.758700030661523e-06,
+      "loss": 0.371,
+      "step": 2629
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.751662298079422e-06,
+      "loss": 0.4088,
+      "step": 2630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74462573818606e-06,
+      "loss": 0.3828,
+      "step": 2631
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.737590354651571e-06,
+      "loss": 0.3714,
+      "step": 2632
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730556151145477e-06,
+      "loss": 0.4408,
+      "step": 2633
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.723523131336673e-06,
+      "loss": 0.3786,
+      "step": 2634
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 0.3313,
+      "step": 2635
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709460657483462e-06,
+      "loss": 0.4063,
+      "step": 2636
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702431210773767e-06,
+      "loss": 0.3499,
+      "step": 2637
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695402962430778e-06,
+      "loss": 0.3759,
+      "step": 2638
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688375916120298e-06,
+      "loss": 0.3718,
+      "step": 2639
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.68135007550749e-06,
+      "loss": 0.3213,
+      "step": 2640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.674325444256899e-06,
+      "loss": 0.3153,
+      "step": 2641
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.667302026032428e-06,
+      "loss": 0.4405,
+      "step": 2642
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.660279824497359e-06,
+      "loss": 0.3751,
+      "step": 2643
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.653258843314337e-06,
+      "loss": 0.3469,
+      "step": 2644
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646239086145361e-06,
+      "loss": 0.4333,
+      "step": 2645
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6392205566518e-06,
+      "loss": 0.3396,
+      "step": 2646
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632203258494385e-06,
+      "loss": 0.3531,
+      "step": 2647
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625187195333197e-06,
+      "loss": 0.373,
+      "step": 2648
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61817237082768e-06,
+      "loss": 0.3942,
+      "step": 2649
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6111587886366216e-06,
+      "loss": 0.3734,
+      "step": 2650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.604146452418175e-06,
+      "loss": 0.3875,
+      "step": 2651
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597135365829837e-06,
+      "loss": 0.3641,
+      "step": 2652
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590125532528448e-06,
+      "loss": 0.4159,
+      "step": 2653
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5831169561702e-06,
+      "loss": 0.3348,
+      "step": 2654
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5761096404106335e-06,
+      "loss": 0.4549,
+      "step": 2655
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.569103588904623e-06,
+      "loss": 0.3426,
+      "step": 2656
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.562098805306385e-06,
+      "loss": 0.4023,
+      "step": 2657
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.555095293269483e-06,
+      "loss": 0.3481,
+      "step": 2658
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.548093056446806e-06,
+      "loss": 0.3909,
+      "step": 2659
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.541092098490587e-06,
+      "loss": 0.3758,
+      "step": 2660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534092423052382e-06,
+      "loss": 0.3514,
+      "step": 2661
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.527094033783089e-06,
+      "loss": 0.3818,
+      "step": 2662
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.520096934332931e-06,
+      "loss": 0.3908,
+      "step": 2663
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.513101128351454e-06,
+      "loss": 0.3704,
+      "step": 2664
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506106619487533e-06,
+      "loss": 0.3318,
+      "step": 2665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499113411389371e-06,
+      "loss": 0.3112,
+      "step": 2666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.492121507704484e-06,
+      "loss": 0.3727,
+      "step": 2667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4851309120797135e-06,
+      "loss": 0.348,
+      "step": 2668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.478141628161211e-06,
+      "loss": 0.3877,
+      "step": 2669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.471153659594456e-06,
+      "loss": 0.3346,
+      "step": 2670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.464167010024236e-06,
+      "loss": 0.3342,
+      "step": 2671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.457181683094645e-06,
+      "loss": 0.3224,
+      "step": 2672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.450197682449093e-06,
+      "loss": 0.3341,
+      "step": 2673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.443215011730303e-06,
+      "loss": 0.4555,
+      "step": 2674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.436233674580293e-06,
+      "loss": 0.4046,
+      "step": 2675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.429253674640391e-06,
+      "loss": 0.4434,
+      "step": 2676
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422275015551234e-06,
+      "loss": 0.2997,
+      "step": 2677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.415297700952748e-06,
+      "loss": 0.4127,
+      "step": 2678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.408321734484165e-06,
+      "loss": 0.3757,
+      "step": 2679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.401347119784008e-06,
+      "loss": 0.3606,
+      "step": 2680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394373860490105e-06,
+      "loss": 0.3627,
+      "step": 2681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.387401960239569e-06,
+      "loss": 0.4386,
+      "step": 2682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.380431422668803e-06,
+      "loss": 0.393,
+      "step": 2683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373462251413504e-06,
+      "loss": 0.4556,
+      "step": 2684
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.366494450108659e-06,
+      "loss": 0.3869,
+      "step": 2685
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.359528022388529e-06,
+      "loss": 0.4364,
+      "step": 2686
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.352562971886673e-06,
+      "loss": 0.4242,
+      "step": 2687
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.345599302235916e-06,
+      "loss": 0.3891,
+      "step": 2688
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.338637017068377e-06,
+      "loss": 0.384,
+      "step": 2689
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.331676120015449e-06,
+      "loss": 0.3999,
+      "step": 2690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.324716614707794e-06,
+      "loss": 0.3906,
+      "step": 2691
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.317758504775352e-06,
+      "loss": 0.4365,
+      "step": 2692
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 0.3483,
+      "step": 2693
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.303846485552249e-06,
+      "loss": 0.4153,
+      "step": 2694
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296892583517819e-06,
+      "loss": 0.3498,
+      "step": 2695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.289940091371068e-06,
+      "loss": 0.3271,
+      "step": 2696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2829890127382866e-06,
+      "loss": 0.3986,
+      "step": 2697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276039351245021e-06,
+      "loss": 0.417,
+      "step": 2698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26909111051607e-06,
+      "loss": 0.3789,
+      "step": 2699
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.262144294175506e-06,
+      "loss": 0.437,
+      "step": 2700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2551989058466545e-06,
+      "loss": 0.3978,
+      "step": 2701
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.248254949152084e-06,
+      "loss": 0.3368,
+      "step": 2702
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.241312427713632e-06,
+      "loss": 0.3476,
+      "step": 2703
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.234371345152382e-06,
+      "loss": 0.3892,
+      "step": 2704
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.227431705088665e-06,
+      "loss": 0.3725,
+      "step": 2705
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.22049351114206e-06,
+      "loss": 0.3617,
+      "step": 2706
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213556766931389e-06,
+      "loss": 0.3731,
+      "step": 2707
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.206621476074729e-06,
+      "loss": 0.3599,
+      "step": 2708
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199687642189388e-06,
+      "loss": 0.3739,
+      "step": 2709
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192755268891915e-06,
+      "loss": 0.3378,
+      "step": 2710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.185824359798099e-06,
+      "loss": 0.3946,
+      "step": 2711
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178894918522974e-06,
+      "loss": 0.3298,
+      "step": 2712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.171966948680792e-06,
+      "loss": 0.396,
+      "step": 2713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1650404538850504e-06,
+      "loss": 0.3191,
+      "step": 2714
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.158115437748467e-06,
+      "loss": 0.3813,
+      "step": 2715
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1511919038830016e-06,
+      "loss": 0.351,
+      "step": 2716
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.14426985589983e-06,
+      "loss": 0.3536,
+      "step": 2717
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.137349297409354e-06,
+      "loss": 0.3298,
+      "step": 2718
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.130430232021204e-06,
+      "loss": 0.3798,
+      "step": 2719
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1235126633442295e-06,
+      "loss": 0.421,
+      "step": 2720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1165965949864934e-06,
+      "loss": 0.2846,
+      "step": 2721
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.109682030555283e-06,
+      "loss": 0.4019,
+      "step": 2722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102768973657104e-06,
+      "loss": 0.4106,
+      "step": 2723
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.095857427897665e-06,
+      "loss": 0.3445,
+      "step": 2724
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.088947396881896e-06,
+      "loss": 0.3424,
+      "step": 2725
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0820388842139285e-06,
+      "loss": 0.3049,
+      "step": 2726
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0751318934971116e-06,
+      "loss": 0.3134,
+      "step": 2727
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.068226428333996e-06,
+      "loss": 0.3044,
+      "step": 2728
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.061322492326333e-06,
+      "loss": 0.3979,
+      "step": 2729
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.05442008907508e-06,
+      "loss": 0.363,
+      "step": 2730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0475192221804014e-06,
+      "loss": 0.3479,
+      "step": 2731
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.040619895241646e-06,
+      "loss": 0.3616,
+      "step": 2732
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.033722111857373e-06,
+      "loss": 0.3216,
+      "step": 2733
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.026825875625323e-06,
+      "loss": 0.3292,
+      "step": 2734
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.019931190142444e-06,
+      "loss": 0.4275,
+      "step": 2735
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.013038059004867e-06,
+      "loss": 0.3954,
+      "step": 2736
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.00614648580791e-06,
+      "loss": 0.3567,
+      "step": 2737
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9992564741460864e-06,
+      "loss": 0.4537,
+      "step": 2738
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992368027613092e-06,
+      "loss": 0.3531,
+      "step": 2739
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9854811498018e-06,
+      "loss": 0.4193,
+      "step": 2740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978595844304272e-06,
+      "loss": 0.4856,
+      "step": 2741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.971712114711753e-06,
+      "loss": 0.406,
+      "step": 2742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964829964614654e-06,
+      "loss": 0.3438,
+      "step": 2743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957949397602574e-06,
+      "loss": 0.4173,
+      "step": 2744
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.951070417264278e-06,
+      "loss": 0.4117,
+      "step": 2745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.944193027187708e-06,
+      "loss": 0.3862,
+      "step": 2746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.937317230959979e-06,
+      "loss": 0.3621,
+      "step": 2747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.930443032167365e-06,
+      "loss": 0.3246,
+      "step": 2748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.923570434395315e-06,
+      "loss": 0.3359,
+      "step": 2749
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916699441228446e-06,
+      "loss": 0.4152,
+      "step": 2750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.347,
+      "step": 2751
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902962283044498e-06,
+      "loss": 0.404,
+      "step": 2752
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.896096125192449e-06,
+      "loss": 0.4255,
+      "step": 2753
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.889231586275637e-06,
+      "loss": 0.3861,
+      "step": 2754
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882368669874473e-06,
+      "loss": 0.3463,
+      "step": 2755
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875507379568511e-06,
+      "loss": 0.3696,
+      "step": 2756
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868647718936474e-06,
+      "loss": 0.4536,
+      "step": 2757
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.861789691556225e-06,
+      "loss": 0.3324,
+      "step": 2758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8549333010047694e-06,
+      "loss": 0.3948,
+      "step": 2759
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.848078550858269e-06,
+      "loss": 0.293,
+      "step": 2760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.841225444692032e-06,
+      "loss": 0.321,
+      "step": 2761
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.834373986080498e-06,
+      "loss": 0.3643,
+      "step": 2762
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.827524178597257e-06,
+      "loss": 0.267,
+      "step": 2763
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.820676025815027e-06,
+      "loss": 0.4239,
+      "step": 2764
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.813829531305676e-06,
+      "loss": 0.3384,
+      "step": 2765
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806984698640202e-06,
+      "loss": 0.3743,
+      "step": 2766
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800141531388731e-06,
+      "loss": 0.3757,
+      "step": 2767
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793300033120524e-06,
+      "loss": 0.4155,
+      "step": 2768
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7864602074039775e-06,
+      "loss": 0.3567,
+      "step": 2769
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.779622057806604e-06,
+      "loss": 0.3503,
+      "step": 2770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772785587895055e-06,
+      "loss": 0.3327,
+      "step": 2771
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765950801235087e-06,
+      "loss": 0.3722,
+      "step": 2772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759117701391602e-06,
+      "loss": 0.3304,
+      "step": 2773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752286291928607e-06,
+      "loss": 0.3871,
+      "step": 2774
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.745456576409227e-06,
+      "loss": 0.3669,
+      "step": 2775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.738628558395712e-06,
+      "loss": 0.412,
+      "step": 2776
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7318022414494235e-06,
+      "loss": 0.3523,
+      "step": 2777
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.724977629130828e-06,
+      "loss": 0.3759,
+      "step": 2778
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.718154724999511e-06,
+      "loss": 0.3299,
+      "step": 2779
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.711333532614168e-06,
+      "loss": 0.3247,
+      "step": 2780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.704514055532597e-06,
+      "loss": 0.3834,
+      "step": 2781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697696297311706e-06,
+      "loss": 0.3472,
+      "step": 2782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690880261507496e-06,
+      "loss": 0.4262,
+      "step": 2783
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684065951675085e-06,
+      "loss": 0.4523,
+      "step": 2784
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677253371368681e-06,
+      "loss": 0.2744,
+      "step": 2785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670442524141592e-06,
+      "loss": 0.3108,
+      "step": 2786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663633413546215e-06,
+      "loss": 0.3702,
+      "step": 2787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.656826043134062e-06,
+      "loss": 0.4338,
+      "step": 2788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650020416455713e-06,
+      "loss": 0.3288,
+      "step": 2789
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.643216537060855e-06,
+      "loss": 0.3656,
+      "step": 2790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.636414408498249e-06,
+      "loss": 0.3811,
+      "step": 2791
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.629614034315761e-06,
+      "loss": 0.3717,
+      "step": 2792
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.62281541806033e-06,
+      "loss": 0.418,
+      "step": 2793
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.616018563277977e-06,
+      "loss": 0.3892,
+      "step": 2794
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6092234735138094e-06,
+      "loss": 0.3895,
+      "step": 2795
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602430152312017e-06,
+      "loss": 0.3346,
+      "step": 2796
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595638603215857e-06,
+      "loss": 0.3741,
+      "step": 2797
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588848829767666e-06,
+      "loss": 0.3106,
+      "step": 2798
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5820608355088664e-06,
+      "loss": 0.3732,
+      "step": 2799
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575274623979933e-06,
+      "loss": 0.3112,
+      "step": 2800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568490198720426e-06,
+      "loss": 0.3714,
+      "step": 2801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5617075632689616e-06,
+      "loss": 0.3862,
+      "step": 2802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.554926721163235e-06,
+      "loss": 0.4139,
+      "step": 2803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548147675940002e-06,
+      "loss": 0.3785,
+      "step": 2804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.541370431135073e-06,
+      "loss": 0.3507,
+      "step": 2805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.534594990283325e-06,
+      "loss": 0.3798,
+      "step": 2806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5278213569187025e-06,
+      "loss": 0.3993,
+      "step": 2807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.521049534574193e-06,
+      "loss": 0.3998,
+      "step": 2808
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 0.361,
+      "step": 2809
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507511337072772e-06,
+      "loss": 0.3851,
+      "step": 2810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500744968977116e-06,
+      "loss": 0.3255,
+      "step": 2811
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49398042602409e-06,
+      "loss": 0.3482,
+      "step": 2812
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4872177117419415e-06,
+      "loss": 0.3611,
+      "step": 2813
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4804568296579685e-06,
+      "loss": 0.325,
+      "step": 2814
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.47369778329852e-06,
+      "loss": 0.3692,
+      "step": 2815
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.466940576188978e-06,
+      "loss": 0.351,
+      "step": 2816
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460185211853767e-06,
+      "loss": 0.3916,
+      "step": 2817
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453431693816358e-06,
+      "loss": 0.3108,
+      "step": 2818
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446680025599248e-06,
+      "loss": 0.3902,
+      "step": 2819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.439930210723978e-06,
+      "loss": 0.3833,
+      "step": 2820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.433182252711113e-06,
+      "loss": 0.3718,
+      "step": 2821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.426436155080262e-06,
+      "loss": 0.3372,
+      "step": 2822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.419691921350055e-06,
+      "loss": 0.3883,
+      "step": 2823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.412949555038149e-06,
+      "loss": 0.3937,
+      "step": 2824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.40620905966123e-06,
+      "loss": 0.3759,
+      "step": 2825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399470438735014e-06,
+      "loss": 0.3421,
+      "step": 2826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.392733695774225e-06,
+      "loss": 0.3324,
+      "step": 2827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.385998834292621e-06,
+      "loss": 0.3914,
+      "step": 2828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37926585780297e-06,
+      "loss": 0.3993,
+      "step": 2829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372534769817061e-06,
+      "loss": 0.3374,
+      "step": 2830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365805573845701e-06,
+      "loss": 0.4592,
+      "step": 2831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359078273398702e-06,
+      "loss": 0.4086,
+      "step": 2832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35235287198489e-06,
+      "loss": 0.3557,
+      "step": 2833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34562937311211e-06,
+      "loss": 0.377,
+      "step": 2834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.338907780287198e-06,
+      "loss": 0.3529,
+      "step": 2835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.332188097016008e-06,
+      "loss": 0.3521,
+      "step": 2836
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.325470326803398e-06,
+      "loss": 0.4267,
+      "step": 2837
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.318754473153221e-06,
+      "loss": 0.3602,
+      "step": 2838
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.312040539568337e-06,
+      "loss": 0.3675,
+      "step": 2839
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.305328529550596e-06,
+      "loss": 0.3651,
+      "step": 2840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 0.3971,
+      "step": 2841
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291910294218966e-06,
+      "loss": 0.3467,
+      "step": 2842
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2852040759037605e-06,
+      "loss": 0.3586,
+      "step": 2843
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278499795153071e-06,
+      "loss": 0.3371,
+      "step": 2844
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271797455463726e-06,
+      "loss": 0.4007,
+      "step": 2845
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.265097060331529e-06,
+      "loss": 0.2913,
+      "step": 2846
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258398613251275e-06,
+      "loss": 0.3392,
+      "step": 2847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251702117716738e-06,
+      "loss": 0.4595,
+      "step": 2848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245007577220686e-06,
+      "loss": 0.3972,
+      "step": 2849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.238314995254857e-06,
+      "loss": 0.3779,
+      "step": 2850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.231624375309971e-06,
+      "loss": 0.4399,
+      "step": 2851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2249357208757155e-06,
+      "loss": 0.4142,
+      "step": 2852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.218249035440773e-06,
+      "loss": 0.4159,
+      "step": 2853
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.211564322492783e-06,
+      "loss": 0.3619,
+      "step": 2854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.204881585518354e-06,
+      "loss": 0.3968,
+      "step": 2855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19820082800308e-06,
+      "loss": 0.3608,
+      "step": 2856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.191522053431508e-06,
+      "loss": 0.3604,
+      "step": 2857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184845265287157e-06,
+      "loss": 0.3739,
+      "step": 2858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178170467052504e-06,
+      "loss": 0.3461,
+      "step": 2859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.171497662208998e-06,
+      "loss": 0.4052,
+      "step": 2860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.164826854237042e-06,
+      "loss": 0.385,
+      "step": 2861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.158158046615997e-06,
+      "loss": 0.439,
+      "step": 2862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151491242824177e-06,
+      "loss": 0.3784,
+      "step": 2863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.144826446338865e-06,
+      "loss": 0.4123,
+      "step": 2864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.138163660636285e-06,
+      "loss": 0.4115,
+      "step": 2865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.131502889191611e-06,
+      "loss": 0.4205,
+      "step": 2866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 0.3611,
+      "step": 2867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.118187402971445e-06,
+      "loss": 0.2671,
+      "step": 2868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.11153269514105e-06,
+      "loss": 0.4079,
+      "step": 2869
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104880015458754e-06,
+      "loss": 0.3537,
+      "step": 2870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.098229367394455e-06,
+      "loss": 0.4211,
+      "step": 2871
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091580754417013e-06,
+      "loss": 0.2867,
+      "step": 2872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084934179994206e-06,
+      "loss": 0.3324,
+      "step": 2873
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078289647592756e-06,
+      "loss": 0.3074,
+      "step": 2874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071647160678323e-06,
+      "loss": 0.3863,
+      "step": 2875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0650067227155006e-06,
+      "loss": 0.3667,
+      "step": 2876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058368337167805e-06,
+      "loss": 0.3506,
+      "step": 2877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051732007497687e-06,
+      "loss": 0.3645,
+      "step": 2878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045097737166533e-06,
+      "loss": 0.4147,
+      "step": 2879
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0384655296346425e-06,
+      "loss": 0.3863,
+      "step": 2880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0318353883612465e-06,
+      "loss": 0.3967,
+      "step": 2881
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.025207316804491e-06,
+      "loss": 0.3494,
+      "step": 2882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0185813184214544e-06,
+      "loss": 0.2784,
+      "step": 2883
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.011957396668129e-06,
+      "loss": 0.3617,
+      "step": 2884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.005335554999414e-06,
+      "loss": 0.3753,
+      "step": 2885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9987157968691345e-06,
+      "loss": 0.3416,
+      "step": 2886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.992098125730031e-06,
+      "loss": 0.3707,
+      "step": 2887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985482545033746e-06,
+      "loss": 0.3942,
+      "step": 2888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.978869058230841e-06,
+      "loss": 0.3897,
+      "step": 2889
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972257668770771e-06,
+      "loss": 0.3563,
+      "step": 2890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965648380101916e-06,
+      "loss": 0.3513,
+      "step": 2891
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959041195671551e-06,
+      "loss": 0.4081,
+      "step": 2892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9524361189258476e-06,
+      "loss": 0.3002,
+      "step": 2893
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9458331533098855e-06,
+      "loss": 0.3547,
+      "step": 2894
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.939232302267646e-06,
+      "loss": 0.3687,
+      "step": 2895
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.354,
+      "step": 2896
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.926036957674715e-06,
+      "loss": 0.3554,
+      "step": 2897
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.91944247100646e-06,
+      "loss": 0.3942,
+      "step": 2898
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.912850112676784e-06,
+      "loss": 0.3968,
+      "step": 2899
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906259886124135e-06,
+      "loss": 0.3285,
+      "step": 2900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8996717947858395e-06,
+      "loss": 0.4317,
+      "step": 2901
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893085842098125e-06,
+      "loss": 0.3527,
+      "step": 2902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886502031496091e-06,
+      "loss": 0.3009,
+      "step": 2903
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879920366413723e-06,
+      "loss": 0.3681,
+      "step": 2904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873340850283884e-06,
+      "loss": 0.3618,
+      "step": 2905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866763486538332e-06,
+      "loss": 0.3624,
+      "step": 2906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8601882786076795e-06,
+      "loss": 0.3994,
+      "step": 2907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.853615229921432e-06,
+      "loss": 0.3565,
+      "step": 2908
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.847044343907956e-06,
+      "loss": 0.3964,
+      "step": 2909
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.840475623994503e-06,
+      "loss": 0.3898,
+      "step": 2910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.833909073607187e-06,
+      "loss": 0.3673,
+      "step": 2911
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.827344696170991e-06,
+      "loss": 0.4343,
+      "step": 2912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.82078249510976e-06,
+      "loss": 0.4018,
+      "step": 2913
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.814222473846214e-06,
+      "loss": 0.3817,
+      "step": 2914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.807664635801935e-06,
+      "loss": 0.366,
+      "step": 2915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801108984397355e-06,
+      "loss": 0.4145,
+      "step": 2916
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794555523051777e-06,
+      "loss": 0.3981,
+      "step": 2917
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788004255183359e-06,
+      "loss": 0.3037,
+      "step": 2918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.781455184209106e-06,
+      "loss": 0.3142,
+      "step": 2919
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77490831354489e-06,
+      "loss": 0.306,
+      "step": 2920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.768363646605435e-06,
+      "loss": 0.3387,
+      "step": 2921
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7618211868043036e-06,
+      "loss": 0.4442,
+      "step": 2922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.755280937553917e-06,
+      "loss": 0.3261,
+      "step": 2923
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.748742902265535e-06,
+      "loss": 0.3664,
+      "step": 2924
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 0.3746,
+      "step": 2925
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7356734872140885e-06,
+      "loss": 0.4443,
+      "step": 2926
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.729142114267775e-06,
+      "loss": 0.3381,
+      "step": 2927
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.722612968916963e-06,
+      "loss": 0.4141,
+      "step": 2928
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.716086054567134e-06,
+      "loss": 0.4205,
+      "step": 2929
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70956137462259e-06,
+      "loss": 0.3298,
+      "step": 2930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7030389324864845e-06,
+      "loss": 0.3447,
+      "step": 2931
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696518731560787e-06,
+      "loss": 0.3594,
+      "step": 2932
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.690000775246315e-06,
+      "loss": 0.3607,
+      "step": 2933
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6834850669427e-06,
+      "loss": 0.3238,
+      "step": 2934
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.676971610048406e-06,
+      "loss": 0.3321,
+      "step": 2935
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670460407960726e-06,
+      "loss": 0.3257,
+      "step": 2936
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.663951464075781e-06,
+      "loss": 0.3429,
+      "step": 2937
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657444781788501e-06,
+      "loss": 0.3414,
+      "step": 2938
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.650940364492643e-06,
+      "loss": 0.3787,
+      "step": 2939
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.64443821558079e-06,
+      "loss": 0.3758,
+      "step": 2940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.637938338444325e-06,
+      "loss": 0.3966,
+      "step": 2941
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.631440736473468e-06,
+      "loss": 0.3806,
+      "step": 2942
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.624945413057229e-06,
+      "loss": 0.3964,
+      "step": 2943
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6184523715834505e-06,
+      "loss": 0.399,
+      "step": 2944
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6119616154387715e-06,
+      "loss": 0.367,
+      "step": 2945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6054731480086376e-06,
+      "loss": 0.3052,
+      "step": 2946
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598986972677312e-06,
+      "loss": 0.3763,
+      "step": 2947
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592503092827857e-06,
+      "loss": 0.3047,
+      "step": 2948
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586021511842136e-06,
+      "loss": 0.3871,
+      "step": 2949
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579542233100812e-06,
+      "loss": 0.2994,
+      "step": 2950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573065259983347e-06,
+      "loss": 0.3606,
+      "step": 2951
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566590595868006e-06,
+      "loss": 0.455,
+      "step": 2952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560118244131852e-06,
+      "loss": 0.3633,
+      "step": 2953
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.553648208150728e-06,
+      "loss": 0.3494,
+      "step": 2954
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.547180491299278e-06,
+      "loss": 0.403,
+      "step": 2955
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5407150969509434e-06,
+      "loss": 0.3508,
+      "step": 2956
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.534252028477938e-06,
+      "loss": 0.3142,
+      "step": 2957
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.527791289251274e-06,
+      "loss": 0.3724,
+      "step": 2958
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.521332882640753e-06,
+      "loss": 0.4414,
+      "step": 2959
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514876812014946e-06,
+      "loss": 0.3626,
+      "step": 2960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5084230807412135e-06,
+      "loss": 0.3634,
+      "step": 2961
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.50197169218569e-06,
+      "loss": 0.4517,
+      "step": 2962
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4955226497133e-06,
+      "loss": 0.4109,
+      "step": 2963
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4890759566877375e-06,
+      "loss": 0.3278,
+      "step": 2964
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48263161647147e-06,
+      "loss": 0.3152,
+      "step": 2965
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.476189632425732e-06,
+      "loss": 0.3157,
+      "step": 2966
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4697500079105445e-06,
+      "loss": 0.3645,
+      "step": 2967
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463312746284681e-06,
+      "loss": 0.4212,
+      "step": 2968
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.456877850905701e-06,
+      "loss": 0.3851,
+      "step": 2969
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.450445325129908e-06,
+      "loss": 0.3642,
+      "step": 2970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.444015172312391e-06,
+      "loss": 0.3415,
+      "step": 2971
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.437587395806987e-06,
+      "loss": 0.3913,
+      "step": 2972
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.431161998966292e-06,
+      "loss": 0.3571,
+      "step": 2973
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.424738985141673e-06,
+      "loss": 0.3153,
+      "step": 2974
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418318357683252e-06,
+      "loss": 0.3896,
+      "step": 2975
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.411900119939896e-06,
+      "loss": 0.303,
+      "step": 2976
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4054842752592285e-06,
+      "loss": 0.3731,
+      "step": 2977
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.399070826987636e-06,
+      "loss": 0.4104,
+      "step": 2978
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392659778470239e-06,
+      "loss": 0.3643,
+      "step": 2979
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386251133050924e-06,
+      "loss": 0.3926,
+      "step": 2980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.379844894072303e-06,
+      "loss": 0.3413,
+      "step": 2981
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.373441064875755e-06,
+      "loss": 0.3818,
+      "step": 2982
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 0.3344,
+      "step": 2983
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.360640649188045e-06,
+      "loss": 0.3715,
+      "step": 2984
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.35424406937333e-06,
+      "loss": 0.3274,
+      "step": 2985
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.347849912693573e-06,
+      "loss": 0.3466,
+      "step": 2986
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3414581824838385e-06,
+      "loss": 0.3564,
+      "step": 2987
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.335068882077925e-06,
+      "loss": 0.371,
+      "step": 2988
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.328682014808364e-06,
+      "loss": 0.3066,
+      "step": 2989
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.322297584006424e-06,
+      "loss": 0.4001,
+      "step": 2990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3159155930021e-06,
+      "loss": 0.307,
+      "step": 2991
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.309536045124111e-06,
+      "loss": 0.3277,
+      "step": 2992
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3031589436999e-06,
+      "loss": 0.358,
+      "step": 2993
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296784292055646e-06,
+      "loss": 0.3584,
+      "step": 2994
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290412093516232e-06,
+      "loss": 0.351,
+      "step": 2995
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284042351405276e-06,
+      "loss": 0.3421,
+      "step": 2996
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277675069045116e-06,
+      "loss": 0.3233,
+      "step": 2997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.271310249756796e-06,
+      "loss": 0.3229,
+      "step": 2998
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.264947896860081e-06,
+      "loss": 0.3538,
+      "step": 2999
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.258588013673445e-06,
+      "loss": 0.3398,
+      "step": 3000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.252230603514082e-06,
+      "loss": 0.3223,
+      "step": 3001
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.245875669697895e-06,
+      "loss": 0.3736,
+      "step": 3002
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.239523215539492e-06,
+      "loss": 0.3445,
+      "step": 3003
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.23317324435218e-06,
+      "loss": 0.3555,
+      "step": 3004
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22682575944799e-06,
+      "loss": 0.3099,
+      "step": 3005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220480764137636e-06,
+      "loss": 0.3787,
+      "step": 3006
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.21413826173055e-06,
+      "loss": 0.3672,
+      "step": 3007
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.207798255534852e-06,
+      "loss": 0.3327,
+      "step": 3008
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.201460748857369e-06,
+      "loss": 0.364,
+      "step": 3009
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.195125745003618e-06,
+      "loss": 0.3385,
+      "step": 3010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1887932472778056e-06,
+      "loss": 0.3545,
+      "step": 3011
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1824632589828465e-06,
+      "loss": 0.4028,
+      "step": 3012
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.176135783420339e-06,
+      "loss": 0.3673,
+      "step": 3013
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.169810823890566e-06,
+      "loss": 0.3583,
+      "step": 3014
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.163488383692499e-06,
+      "loss": 0.3716,
+      "step": 3015
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1571684661238075e-06,
+      "loss": 0.2917,
+      "step": 3016
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.150851074480826e-06,
+      "loss": 0.3375,
+      "step": 3017
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.144536212058591e-06,
+      "loss": 0.3598,
+      "step": 3018
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.138223882150803e-06,
+      "loss": 0.3926,
+      "step": 3019
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.131914088049855e-06,
+      "loss": 0.3406,
+      "step": 3020
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.12560683304681e-06,
+      "loss": 0.4148,
+      "step": 3021
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.119302120431404e-06,
+      "loss": 0.3796,
+      "step": 3022
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.112999953492055e-06,
+      "loss": 0.315,
+      "step": 3023
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1067003355158505e-06,
+      "loss": 0.359,
+      "step": 3024
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100403269788547e-06,
+      "loss": 0.4014,
+      "step": 3025
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094108759594569e-06,
+      "loss": 0.3381,
+      "step": 3026
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.087816808217005e-06,
+      "loss": 0.3399,
+      "step": 3027
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0815274189376186e-06,
+      "loss": 0.3439,
+      "step": 3028
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0752405950368325e-06,
+      "loss": 0.3731,
+      "step": 3029
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.068956339793729e-06,
+      "loss": 0.3702,
+      "step": 3030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.062674656486049e-06,
+      "loss": 0.3511,
+      "step": 3031
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.056395548390198e-06,
+      "loss": 0.3472,
+      "step": 3032
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.050119018781232e-06,
+      "loss": 0.3796,
+      "step": 3033
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.043845070932873e-06,
+      "loss": 0.3294,
+      "step": 3034
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03757370811748e-06,
+      "loss": 0.3622,
+      "step": 3035
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0313049336060805e-06,
+      "loss": 0.3758,
+      "step": 3036
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025038750668338e-06,
+      "loss": 0.2957,
+      "step": 3037
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.018775162572571e-06,
+      "loss": 0.2634,
+      "step": 3038
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012514172585744e-06,
+      "loss": 0.2996,
+      "step": 3039
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.006255783973471e-06,
+      "loss": 0.3313,
+      "step": 3040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.3357,
+      "step": 3041
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993746823928226e-06,
+      "loss": 0.3343,
+      "step": 3042
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987496259019684e-06,
+      "loss": 0.3375,
+      "step": 3043
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.981248308534541e-06,
+      "loss": 0.3452,
+      "step": 3044
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.975002975731613e-06,
+      "loss": 0.3532,
+      "step": 3045
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9687602638683326e-06,
+      "loss": 0.3777,
+      "step": 3046
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9625201762007865e-06,
+      "loss": 0.3116,
+      "step": 3047
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.956282715983677e-06,
+      "loss": 0.371,
+      "step": 3048
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.950047886470336e-06,
+      "loss": 0.393,
+      "step": 3049
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943815690912732e-06,
+      "loss": 0.3603,
+      "step": 3050
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.93758613256146e-06,
+      "loss": 0.2992,
+      "step": 3051
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931359214665731e-06,
+      "loss": 0.3529,
+      "step": 3052
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925134940473383e-06,
+      "loss": 0.3659,
+      "step": 3053
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.918913313230872e-06,
+      "loss": 0.3044,
+      "step": 3054
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.912694336183278e-06,
+      "loss": 0.3159,
+      "step": 3055
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906478012574304e-06,
+      "loss": 0.3247,
+      "step": 3056
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.900264345646253e-06,
+      "loss": 0.3378,
+      "step": 3057
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.894053338640058e-06,
+      "loss": 0.283,
+      "step": 3058
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.887844994795257e-06,
+      "loss": 0.3238,
+      "step": 3059
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.881639317349993e-06,
+      "loss": 0.3,
+      "step": 3060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.875436309541033e-06,
+      "loss": 0.2926,
+      "step": 3061
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.869235974603744e-06,
+      "loss": 0.3188,
+      "step": 3062
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8630383157720985e-06,
+      "loss": 0.3595,
+      "step": 3063
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.856843336278671e-06,
+      "loss": 0.3447,
+      "step": 3064
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8506510393546365e-06,
+      "loss": 0.3352,
+      "step": 3065
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844461428229782e-06,
+      "loss": 0.3575,
+      "step": 3066
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838274506132489e-06,
+      "loss": 0.3768,
+      "step": 3067
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.832090276289731e-06,
+      "loss": 0.3599,
+      "step": 3068
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.825908741927076e-06,
+      "loss": 0.3037,
+      "step": 3069
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8197299062687e-06,
+      "loss": 0.3658,
+      "step": 3070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813553772537354e-06,
+      "loss": 0.3088,
+      "step": 3071
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.807380343954394e-06,
+      "loss": 0.3897,
+      "step": 3072
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801209623739754e-06,
+      "loss": 0.3698,
+      "step": 3073
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.795041615111967e-06,
+      "loss": 0.3832,
+      "step": 3074
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.78887632128814e-06,
+      "loss": 0.3361,
+      "step": 3075
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.782713745483967e-06,
+      "loss": 0.3227,
+      "step": 3076
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.776553890913729e-06,
+      "loss": 0.4067,
+      "step": 3077
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.770396760790289e-06,
+      "loss": 0.3285,
+      "step": 3078
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.764242358325081e-06,
+      "loss": 0.3048,
+      "step": 3079
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758090686728118e-06,
+      "loss": 0.3376,
+      "step": 3080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751941749207996e-06,
+      "loss": 0.3212,
+      "step": 3081
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7457955489718745e-06,
+      "loss": 0.3224,
+      "step": 3082
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.739652089225497e-06,
+      "loss": 0.4286,
+      "step": 3083
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733511373173165e-06,
+      "loss": 0.2876,
+      "step": 3084
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.727373404017762e-06,
+      "loss": 0.3958,
+      "step": 3085
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7212381849607296e-06,
+      "loss": 0.387,
+      "step": 3086
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715105719202071e-06,
+      "loss": 0.3749,
+      "step": 3087
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708976009940365e-06,
+      "loss": 0.3607,
+      "step": 3088
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.702849060372752e-06,
+      "loss": 0.4202,
+      "step": 3089
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.696724873694923e-06,
+      "loss": 0.3171,
+      "step": 3090
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.690603453101134e-06,
+      "loss": 0.346,
+      "step": 3091
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.684484801784196e-06,
+      "loss": 0.385,
+      "step": 3092
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.678368922935478e-06,
+      "loss": 0.3492,
+      "step": 3093
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.672255819744907e-06,
+      "loss": 0.3889,
+      "step": 3094
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.66614549540095e-06,
+      "loss": 0.3631,
+      "step": 3095
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6600379530906395e-06,
+      "loss": 0.3414,
+      "step": 3096
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653933195999546e-06,
+      "loss": 0.3789,
+      "step": 3097
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6478312273117866e-06,
+      "loss": 0.3231,
+      "step": 3098
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 0.3022,
+      "step": 3099
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635635667875497e-06,
+      "loss": 0.3728,
+      "step": 3100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629542083487931e-06,
+      "loss": 0.332,
+      "step": 3101
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623451300225627e-06,
+      "loss": 0.3263,
+      "step": 3102
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.617363321265416e-06,
+      "loss": 0.3804,
+      "step": 3103
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.611278149782671e-06,
+      "loss": 0.2956,
+      "step": 3104
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6051957889513e-06,
+      "loss": 0.271,
+      "step": 3105
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.599116241943741e-06,
+      "loss": 0.3994,
+      "step": 3106
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.593039511930963e-06,
+      "loss": 0.2939,
+      "step": 3107
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.586965602082473e-06,
+      "loss": 0.3605,
+      "step": 3108
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5808945155662984e-06,
+      "loss": 0.3837,
+      "step": 3109
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574826255549005e-06,
+      "loss": 0.3801,
+      "step": 3110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.568760825195671e-06,
+      "loss": 0.3284,
+      "step": 3111
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562698227669914e-06,
+      "loss": 0.3318,
+      "step": 3112
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556638466133859e-06,
+      "loss": 0.3599,
+      "step": 3113
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.550581543748162e-06,
+      "loss": 0.3604,
+      "step": 3114
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.544527463671984e-06,
+      "loss": 0.3036,
+      "step": 3115
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53847622906303e-06,
+      "loss": 0.32,
+      "step": 3116
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5324278430775e-06,
+      "loss": 0.3068,
+      "step": 3117
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.526382308870106e-06,
+      "loss": 0.3452,
+      "step": 3118
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.52033962959409e-06,
+      "loss": 0.2846,
+      "step": 3119
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.514299808401185e-06,
+      "loss": 0.2854,
+      "step": 3120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.508262848441651e-06,
+      "loss": 0.3713,
+      "step": 3121
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.502228752864241e-06,
+      "loss": 0.3346,
+      "step": 3122
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.496197524816229e-06,
+      "loss": 0.3183,
+      "step": 3123
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.490169167443379e-06,
+      "loss": 0.3486,
+      "step": 3124
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.484143683889964e-06,
+      "loss": 0.3576,
+      "step": 3125
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4781210772987516e-06,
+      "loss": 0.3537,
+      "step": 3126
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.47210135081103e-06,
+      "loss": 0.2721,
+      "step": 3127
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46608450756656e-06,
+      "loss": 0.3039,
+      "step": 3128
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4600705507036125e-06,
+      "loss": 0.2601,
+      "step": 3129
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.454059483358944e-06,
+      "loss": 0.3334,
+      "step": 3130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44805130866781e-06,
+      "loss": 0.338,
+      "step": 3131
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.442046029763964e-06,
+      "loss": 0.3255,
+      "step": 3132
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436043649779637e-06,
+      "loss": 0.3218,
+      "step": 3133
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.430044171845546e-06,
+      "loss": 0.2969,
+      "step": 3134
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.42404759909091e-06,
+      "loss": 0.2976,
+      "step": 3135
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.418053934643421e-06,
+      "loss": 0.3339,
+      "step": 3136
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.412063181629247e-06,
+      "loss": 0.3178,
+      "step": 3137
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.406075343173063e-06,
+      "loss": 0.2851,
+      "step": 3138
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400090422397999e-06,
+      "loss": 0.3636,
+      "step": 3139
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394108422425675e-06,
+      "loss": 0.3443,
+      "step": 3140
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 0.3304,
+      "step": 3141
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382153197368084e-06,
+      "loss": 0.3554,
+      "step": 3142
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.376179978518429e-06,
+      "loss": 0.3196,
+      "step": 3143
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3702096929427325e-06,
+      "loss": 0.3349,
+      "step": 3144
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364242343754972e-06,
+      "loss": 0.3114,
+      "step": 3145
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358277934067606e-06,
+      "loss": 0.4029,
+      "step": 3146
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3523164669915495e-06,
+      "loss": 0.3452,
+      "step": 3147
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.346357945636183e-06,
+      "loss": 0.3921,
+      "step": 3148
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3404023731093604e-06,
+      "loss": 0.3264,
+      "step": 3149
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.33444975251739e-06,
+      "loss": 0.3793,
+      "step": 3150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.328500086965041e-06,
+      "loss": 0.2926,
+      "step": 3151
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.322553379555543e-06,
+      "loss": 0.3917,
+      "step": 3152
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3166096333905745e-06,
+      "loss": 0.3649,
+      "step": 3153
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310668851570283e-06,
+      "loss": 0.3077,
+      "step": 3154
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.304731037193267e-06,
+      "loss": 0.3372,
+      "step": 3155
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298796193356566e-06,
+      "loss": 0.3195,
+      "step": 3156
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.2843,
+      "step": 3157
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.286935429684567e-06,
+      "loss": 0.4817,
+      "step": 3158
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281009516035602e-06,
+      "loss": 0.3516,
+      "step": 3159
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.275086585299635e-06,
+      "loss": 0.3229,
+      "step": 3160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.269166640565953e-06,
+      "loss": 0.384,
+      "step": 3161
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263249684922281e-06,
+      "loss": 0.3303,
+      "step": 3162
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.257335721454785e-06,
+      "loss": 0.3504,
+      "step": 3163
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.251424753248069e-06,
+      "loss": 0.3574,
+      "step": 3164
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2455167833851804e-06,
+      "loss": 0.3421,
+      "step": 3165
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.239611814947605e-06,
+      "loss": 0.3258,
+      "step": 3166
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.233709851015254e-06,
+      "loss": 0.3727,
+      "step": 3167
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.227810894666472e-06,
+      "loss": 0.3157,
+      "step": 3168
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221914948978047e-06,
+      "loss": 0.3986,
+      "step": 3169
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.216022017025179e-06,
+      "loss": 0.4089,
+      "step": 3170
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210132101881516e-06,
+      "loss": 0.3104,
+      "step": 3171
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204245206619111e-06,
+      "loss": 0.3492,
+      "step": 3172
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.198361334308463e-06,
+      "loss": 0.3881,
+      "step": 3173
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.192480488018478e-06,
+      "loss": 0.3507,
+      "step": 3174
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186602670816488e-06,
+      "loss": 0.363,
+      "step": 3175
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1807278857682494e-06,
+      "loss": 0.3797,
+      "step": 3176
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17485613593794e-06,
+      "loss": 0.3225,
+      "step": 3177
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168987424388143e-06,
+      "loss": 0.361,
+      "step": 3178
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1631217541798614e-06,
+      "loss": 0.3347,
+      "step": 3179
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.157259128372521e-06,
+      "loss": 0.2776,
+      "step": 3180
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.151399550023941e-06,
+      "loss": 0.3349,
+      "step": 3181
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.145543022190374e-06,
+      "loss": 0.2986,
+      "step": 3182
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.139689547926461e-06,
+      "loss": 0.3327,
+      "step": 3183
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.133839130285265e-06,
+      "loss": 0.3489,
+      "step": 3184
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1279917723182476e-06,
+      "loss": 0.2881,
+      "step": 3185
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.264,
+      "step": 3186
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.116306247604606e-06,
+      "loss": 0.378,
+      "step": 3187
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11046808695293e-06,
+      "loss": 0.3537,
+      "step": 3188
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.104632998165309e-06,
+      "loss": 0.3767,
+      "step": 3189
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0988009842852085e-06,
+      "loss": 0.3712,
+      "step": 3190
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092972048354491e-06,
+      "loss": 0.2719,
+      "step": 3191
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.087146193413416e-06,
+      "loss": 0.3474,
+      "step": 3192
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08132342250064e-06,
+      "loss": 0.3274,
+      "step": 3193
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0755037386532e-06,
+      "loss": 0.387,
+      "step": 3194
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.069687144906532e-06,
+      "loss": 0.3078,
+      "step": 3195
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.063873644294457e-06,
+      "loss": 0.4277,
+      "step": 3196
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.058063239849178e-06,
+      "loss": 0.303,
+      "step": 3197
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.05225593460129e-06,
+      "loss": 0.3442,
+      "step": 3198
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.046451731579775e-06,
+      "loss": 0.3338,
+      "step": 3199
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.040650633811985e-06,
+      "loss": 0.3229,
+      "step": 3200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034852644323661e-06,
+      "loss": 0.3274,
+      "step": 3201
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0290577661389155e-06,
+      "loss": 0.4039,
+      "step": 3202
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023266002280246e-06,
+      "loss": 0.3878,
+      "step": 3203
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0174773557685284e-06,
+      "loss": 0.3278,
+      "step": 3204
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.011691829623001e-06,
+      "loss": 0.3437,
+      "step": 3205
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0059094268612785e-06,
+      "loss": 0.3815,
+      "step": 3206
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000130150499355e-06,
+      "loss": 0.3751,
+      "step": 3207
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.994354003551579e-06,
+      "loss": 0.3217,
+      "step": 3208
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.988580989030686e-06,
+      "loss": 0.2658,
+      "step": 3209
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.982811109947755e-06,
+      "loss": 0.3313,
+      "step": 3210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.977044369312253e-06,
+      "loss": 0.3221,
+      "step": 3211
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.971280770131992e-06,
+      "loss": 0.4058,
+      "step": 3212
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965520315413149e-06,
+      "loss": 0.3219,
+      "step": 3213
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959763008160269e-06,
+      "loss": 0.269,
+      "step": 3214
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 0.3427,
+      "step": 3215
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948257848062351e-06,
+      "loss": 0.3272,
+      "step": 3216
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.942510001218171e-06,
+      "loss": 0.2993,
+      "step": 3217
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.936765313841685e-06,
+      "loss": 0.2943,
+      "step": 3218
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9310237889292e-06,
+      "loss": 0.3488,
+      "step": 3219
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92528542947539e-06,
+      "loss": 0.3356,
+      "step": 3220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.919550238473265e-06,
+      "loss": 0.368,
+      "step": 3221
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913818218914194e-06,
+      "loss": 0.3014,
+      "step": 3222
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9080893737878815e-06,
+      "loss": 0.3359,
+      "step": 3223
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.902363706082377e-06,
+      "loss": 0.378,
+      "step": 3224
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.896641218784081e-06,
+      "loss": 0.3828,
+      "step": 3225
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.890921914877732e-06,
+      "loss": 0.2733,
+      "step": 3226
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8852057973464045e-06,
+      "loss": 0.346,
+      "step": 3227
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8794928691715125e-06,
+      "loss": 0.331,
+      "step": 3228
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873783133332803e-06,
+      "loss": 0.3828,
+      "step": 3229
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.868076592808366e-06,
+      "loss": 0.3037,
+      "step": 3230
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.862373250574626e-06,
+      "loss": 0.3166,
+      "step": 3231
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.856673109606328e-06,
+      "loss": 0.3069,
+      "step": 3232
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850976172876551e-06,
+      "loss": 0.3565,
+      "step": 3233
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845282443356711e-06,
+      "loss": 0.3341,
+      "step": 3234
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839591924016541e-06,
+      "loss": 0.3525,
+      "step": 3235
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833904617824105e-06,
+      "loss": 0.3324,
+      "step": 3236
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828220527745796e-06,
+      "loss": 0.2992,
+      "step": 3237
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822539656746318e-06,
+      "loss": 0.2779,
+      "step": 3238
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.816862007788702e-06,
+      "loss": 0.3706,
+      "step": 3239
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.811187583834296e-06,
+      "loss": 0.3311,
+      "step": 3240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8055163878427703e-06,
+      "loss": 0.3174,
+      "step": 3241
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.799848422772112e-06,
+      "loss": 0.2605,
+      "step": 3242
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.794183691578618e-06,
+      "loss": 0.2903,
+      "step": 3243
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7885221972168974e-06,
+      "loss": 0.3186,
+      "step": 3244
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7828639426398793e-06,
+      "loss": 0.3496,
+      "step": 3245
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7772089307987936e-06,
+      "loss": 0.3089,
+      "step": 3246
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771557164643189e-06,
+      "loss": 0.2777,
+      "step": 3247
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7659086471209092e-06,
+      "loss": 0.2603,
+      "step": 3248
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7602633811781165e-06,
+      "loss": 0.2894,
+      "step": 3249
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754621369759267e-06,
+      "loss": 0.3187,
+      "step": 3250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7489826158071196e-06,
+      "loss": 0.2949,
+      "step": 3251
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7433471222627414e-06,
+      "loss": 0.3132,
+      "step": 3252
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7377148920654973e-06,
+      "loss": 0.2611,
+      "step": 3253
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7320859281530453e-06,
+      "loss": 0.3283,
+      "step": 3254
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7264602334613385e-06,
+      "loss": 0.3116,
+      "step": 3255
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.720837810924637e-06,
+      "loss": 0.3688,
+      "step": 3256
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.715218663475476e-06,
+      "loss": 0.3401,
+      "step": 3257
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.709602794044702e-06,
+      "loss": 0.3519,
+      "step": 3258
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.703990205561434e-06,
+      "loss": 0.3791,
+      "step": 3259
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.698380900953095e-06,
+      "loss": 0.2853,
+      "step": 3260
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6927748831453835e-06,
+      "loss": 0.3458,
+      "step": 3261
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6871721550622865e-06,
+      "loss": 0.3655,
+      "step": 3262
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6815727196260796e-06,
+      "loss": 0.3647,
+      "step": 3263
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.675976579757321e-06,
+      "loss": 0.4113,
+      "step": 3264
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.670383738374844e-06,
+      "loss": 0.3581,
+      "step": 3265
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6647941983957647e-06,
+      "loss": 0.3308,
+      "step": 3266
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.659207962735475e-06,
+      "loss": 0.3735,
+      "step": 3267
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6536250343076475e-06,
+      "loss": 0.3161,
+      "step": 3268
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.648045416024232e-06,
+      "loss": 0.3033,
+      "step": 3269
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6424691107954446e-06,
+      "loss": 0.3255,
+      "step": 3270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6368961215297726e-06,
+      "loss": 0.2957,
+      "step": 3271
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.631326451133984e-06,
+      "loss": 0.3294,
+      "step": 3272
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.4195,
+      "step": 3273
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.620197078570432e-06,
+      "loss": 0.2926,
+      "step": 3274
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6146373822075366e-06,
+      "loss": 0.3416,
+      "step": 3275
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.609081016324243e-06,
+      "loss": 0.3895,
+      "step": 3276
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.603527983818641e-06,
+      "loss": 0.2818,
+      "step": 3277
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5979782875870817e-06,
+      "loss": 0.3747,
+      "step": 3278
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5924319305241797e-06,
+      "loss": 0.3399,
+      "step": 3279
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.586888915522809e-06,
+      "loss": 0.2831,
+      "step": 3280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.581349245474095e-06,
+      "loss": 0.3161,
+      "step": 3281
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5758129232674156e-06,
+      "loss": 0.2725,
+      "step": 3282
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.570279951790416e-06,
+      "loss": 0.366,
+      "step": 3283
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5647503339289778e-06,
+      "loss": 0.3184,
+      "step": 3284
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5592240725672476e-06,
+      "loss": 0.3662,
+      "step": 3285
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.553701170587609e-06,
+      "loss": 0.2892,
+      "step": 3286
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5481816308707052e-06,
+      "loss": 0.3449,
+      "step": 3287
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5426654562954155e-06,
+      "loss": 0.2852,
+      "step": 3288
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.537152649738864e-06,
+      "loss": 0.3368,
+      "step": 3289
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.531643214076429e-06,
+      "loss": 0.354,
+      "step": 3290
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5261371521817247e-06,
+      "loss": 0.3397,
+      "step": 3291
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5206344669266004e-06,
+      "loss": 0.3821,
+      "step": 3292
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.515135161181151e-06,
+      "loss": 0.3991,
+      "step": 3293
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.509639237813701e-06,
+      "loss": 0.3379,
+      "step": 3294
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5041466996908203e-06,
+      "loss": 0.3255,
+      "step": 3295
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4986575496773122e-06,
+      "loss": 0.2751,
+      "step": 3296
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.493171790636203e-06,
+      "loss": 0.2927,
+      "step": 3297
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.487689425428762e-06,
+      "loss": 0.3409,
+      "step": 3298
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.482210456914481e-06,
+      "loss": 0.2628,
+      "step": 3299
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.476734887951079e-06,
+      "loss": 0.35,
+      "step": 3300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4712627213945084e-06,
+      "loss": 0.3389,
+      "step": 3301
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4657939600989453e-06,
+      "loss": 0.3822,
+      "step": 3302
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4603286069167873e-06,
+      "loss": 0.3179,
+      "step": 3303
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.454866664698654e-06,
+      "loss": 0.3056,
+      "step": 3304
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.449408136293383e-06,
+      "loss": 0.3664,
+      "step": 3305
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4439530245480403e-06,
+      "loss": 0.3315,
+      "step": 3306
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4385013323079066e-06,
+      "loss": 0.362,
+      "step": 3307
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4330530624164747e-06,
+      "loss": 0.3362,
+      "step": 3308
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4276082177154536e-06,
+      "loss": 0.2646,
+      "step": 3309
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4221668010447705e-06,
+      "loss": 0.2445,
+      "step": 3310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.416728815242557e-06,
+      "loss": 0.3438,
+      "step": 3311
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.411294263145166e-06,
+      "loss": 0.3086,
+      "step": 3312
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.405863147587146e-06,
+      "loss": 0.3027,
+      "step": 3313
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4004354714012667e-06,
+      "loss": 0.3125,
+      "step": 3314
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.395011237418494e-06,
+      "loss": 0.3367,
+      "step": 3315
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3895904484679986e-06,
+      "loss": 0.3254,
+      "step": 3316
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.38417310737716e-06,
+      "loss": 0.3111,
+      "step": 3317
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3787592169715612e-06,
+      "loss": 0.3904,
+      "step": 3318
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3733487800749775e-06,
+      "loss": 0.3358,
+      "step": 3319
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.367941799509382e-06,
+      "loss": 0.3423,
+      "step": 3320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3625382780949576e-06,
+      "loss": 0.401,
+      "step": 3321
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.357138218650067e-06,
+      "loss": 0.324,
+      "step": 3322
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3517416239912826e-06,
+      "loss": 0.3624,
+      "step": 3323
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3463484969333547e-06,
+      "loss": 0.4257,
+      "step": 3324
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.340958840289239e-06,
+      "loss": 0.3103,
+      "step": 3325
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.335572656870072e-06,
+      "loss": 0.3433,
+      "step": 3326
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3301899494851764e-06,
+      "loss": 0.3213,
+      "step": 3327
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.324810720942071e-06,
+      "loss": 0.3523,
+      "step": 3328
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.319434974046458e-06,
+      "loss": 0.3311,
+      "step": 3329
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3140627116022194e-06,
+      "loss": 0.3331,
+      "step": 3330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.3048,
+      "step": 3331
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3033286512743066e-06,
+      "loss": 0.3864,
+      "step": 3332
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2979668589893064e-06,
+      "loss": 0.3854,
+      "step": 3333
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2926085623530314e-06,
+      "loss": 0.3635,
+      "step": 3334
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2872537641602542e-06,
+      "loss": 0.3625,
+      "step": 3335
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.28190246720394e-06,
+      "loss": 0.3708,
+      "step": 3336
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2765546742752173e-06,
+      "loss": 0.4195,
+      "step": 3337
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2712103881633873e-06,
+      "loss": 0.3543,
+      "step": 3338
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2658696116559253e-06,
+      "loss": 0.3627,
+      "step": 3339
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.260532347538481e-06,
+      "loss": 0.3291,
+      "step": 3340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.255198598594862e-06,
+      "loss": 0.4208,
+      "step": 3341
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.249868367607046e-06,
+      "loss": 0.3641,
+      "step": 3342
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.244541657355176e-06,
+      "loss": 0.2526,
+      "step": 3343
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.23921847061756e-06,
+      "loss": 0.3012,
+      "step": 3344
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2338988101706727e-06,
+      "loss": 0.3729,
+      "step": 3345
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2285826787891395e-06,
+      "loss": 0.4035,
+      "step": 3346
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2232700792457485e-06,
+      "loss": 0.3215,
+      "step": 3347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2179610143114517e-06,
+      "loss": 0.3278,
+      "step": 3348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2126554867553483e-06,
+      "loss": 0.333,
+      "step": 3349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.207353499344703e-06,
+      "loss": 0.3064,
+      "step": 3350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2020550548449213e-06,
+      "loss": 0.3256,
+      "step": 3351
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196760156019575e-06,
+      "loss": 0.3861,
+      "step": 3352
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1914688056303743e-06,
+      "loss": 0.317,
+      "step": 3353
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1861810064371825e-06,
+      "loss": 0.3593,
+      "step": 3354
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1808967611980133e-06,
+      "loss": 0.3477,
+      "step": 3355
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.17561607266903e-06,
+      "loss": 0.313,
+      "step": 3356
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1703389436045307e-06,
+      "loss": 0.3242,
+      "step": 3357
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1650653767569596e-06,
+      "loss": 0.3499,
+      "step": 3358
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1597953748769126e-06,
+      "loss": 0.3913,
+      "step": 3359
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1545289407131128e-06,
+      "loss": 0.3436,
+      "step": 3360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1492660770124337e-06,
+      "loss": 0.3638,
+      "step": 3361
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1440067865198776e-06,
+      "loss": 0.3176,
+      "step": 3362
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1387510719785907e-06,
+      "loss": 0.3983,
+      "step": 3363
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1334989361298486e-06,
+      "loss": 0.3389,
+      "step": 3364
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1282503817130583e-06,
+      "loss": 0.2781,
+      "step": 3365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.123005411465766e-06,
+      "loss": 0.3308,
+      "step": 3366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1177640281236486e-06,
+      "loss": 0.3253,
+      "step": 3367
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1125262344205044e-06,
+      "loss": 0.313,
+      "step": 3368
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.107292033088265e-06,
+      "loss": 0.3484,
+      "step": 3369
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.102061426856984e-06,
+      "loss": 0.3359,
+      "step": 3370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.096834418454844e-06,
+      "loss": 0.3655,
+      "step": 3371
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091611010608154e-06,
+      "loss": 0.3099,
+      "step": 3372
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.086391206041338e-06,
+      "loss": 0.339,
+      "step": 3373
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.081175007476941e-06,
+      "loss": 0.2935,
+      "step": 3374
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.075962417635634e-06,
+      "loss": 0.3373,
+      "step": 3375
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.070753439236196e-06,
+      "loss": 0.2741,
+      "step": 3376
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0655480749955303e-06,
+      "loss": 0.3419,
+      "step": 3377
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.060346327628657e-06,
+      "loss": 0.291,
+      "step": 3378
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.055148199848702e-06,
+      "loss": 0.3217,
+      "step": 3379
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.049953694366905e-06,
+      "loss": 0.324,
+      "step": 3380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0447628138926153e-06,
+      "loss": 0.3485,
+      "step": 3381
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0395755611332987e-06,
+      "loss": 0.241,
+      "step": 3382
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0343919387945265e-06,
+      "loss": 0.3041,
+      "step": 3383
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.02921194957997e-06,
+      "loss": 0.2959,
+      "step": 3384
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.024035596191408e-06,
+      "loss": 0.3776,
+      "step": 3385
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.01886288132873e-06,
+      "loss": 0.3323,
+      "step": 3386
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.013693807689917e-06,
+      "loss": 0.3265,
+      "step": 3387
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.008528377971062e-06,
+      "loss": 0.3576,
+      "step": 3388
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 0.326,
+      "step": 3389
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9982084610680574e-06,
+      "loss": 0.3085,
+      "step": 3390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9930539792665767e-06,
+      "loss": 0.319,
+      "step": 3391
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9879031521503808e-06,
+      "loss": 0.3322,
+      "step": 3392
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9827559824060314e-06,
+      "loss": 0.3157,
+      "step": 3393
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9776124727182043e-06,
+      "loss": 0.2714,
+      "step": 3394
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.972472625769648e-06,
+      "loss": 0.3563,
+      "step": 3395
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9673364442412e-06,
+      "loss": 0.3867,
+      "step": 3396
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9622039308118e-06,
+      "loss": 0.3111,
+      "step": 3397
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.95707508815846e-06,
+      "loss": 0.3067,
+      "step": 3398
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9519499189562884e-06,
+      "loss": 0.3083,
+      "step": 3399
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9468284258784663e-06,
+      "loss": 0.328,
+      "step": 3400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.941710611596271e-06,
+      "loss": 0.3098,
+      "step": 3401
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9365964787790515e-06,
+      "loss": 0.278,
+      "step": 3402
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9314860300942372e-06,
+      "loss": 0.3667,
+      "step": 3403
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9263792682073323e-06,
+      "loss": 0.3363,
+      "step": 3404
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9212761957819347e-06,
+      "loss": 0.3279,
+      "step": 3405
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9161768154797023e-06,
+      "loss": 0.3421,
+      "step": 3406
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.91108112996037e-06,
+      "loss": 0.345,
+      "step": 3407
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9059891418817453e-06,
+      "loss": 0.2944,
+      "step": 3408
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9009008538997142e-06,
+      "loss": 0.34,
+      "step": 3409
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.895816268668222e-06,
+      "loss": 0.3752,
+      "step": 3410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8907353888392955e-06,
+      "loss": 0.3441,
+      "step": 3411
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.885658217063013e-06,
+      "loss": 0.3541,
+      "step": 3412
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8805847559875365e-06,
+      "loss": 0.3572,
+      "step": 3413
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8755150082590786e-06,
+      "loss": 0.3672,
+      "step": 3414
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8704489765219157e-06,
+      "loss": 0.3284,
+      "step": 3415
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8653866634184e-06,
+      "loss": 0.3535,
+      "step": 3416
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8603280715889304e-06,
+      "loss": 0.3338,
+      "step": 3417
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.855273203671969e-06,
+      "loss": 0.3186,
+      "step": 3418
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.850222062304031e-06,
+      "loss": 0.3471,
+      "step": 3419
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8451746501196998e-06,
+      "loss": 0.389,
+      "step": 3420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8401309697515987e-06,
+      "loss": 0.2552,
+      "step": 3421
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.835091023830421e-06,
+      "loss": 0.2924,
+      "step": 3422
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.830054814984895e-06,
+      "loss": 0.3019,
+      "step": 3423
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8250223458418145e-06,
+      "loss": 0.3206,
+      "step": 3424
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819993619026015e-06,
+      "loss": 0.3396,
+      "step": 3425
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.814968637160377e-06,
+      "loss": 0.3121,
+      "step": 3426
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8099474028658356e-06,
+      "loss": 0.3494,
+      "step": 3427
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.804929918761371e-06,
+      "loss": 0.2149,
+      "step": 3428
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7999161874640026e-06,
+      "loss": 0.3445,
+      "step": 3429
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7949062115887916e-06,
+      "loss": 0.3236,
+      "step": 3430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7898999937488415e-06,
+      "loss": 0.3467,
+      "step": 3431
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.784897536555299e-06,
+      "loss": 0.3834,
+      "step": 3432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.779898842617351e-06,
+      "loss": 0.3652,
+      "step": 3433
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.774903914542213e-06,
+      "loss": 0.3297,
+      "step": 3434
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.769912754935146e-06,
+      "loss": 0.3537,
+      "step": 3435
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.764925366399438e-06,
+      "loss": 0.3302,
+      "step": 3436
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7599417515364103e-06,
+      "loss": 0.3512,
+      "step": 3437
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7549619129454207e-06,
+      "loss": 0.3624,
+      "step": 3438
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7499858532238587e-06,
+      "loss": 0.313,
+      "step": 3439
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7450135749671357e-06,
+      "loss": 0.3241,
+      "step": 3440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 0.3345,
+      "step": 3441
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7350803732200017e-06,
+      "loss": 0.3241,
+      "step": 3442
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7301194549105523e-06,
+      "loss": 0.3511,
+      "step": 3443
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7251623284278684e-06,
+      "loss": 0.2612,
+      "step": 3444
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7202089963574863e-06,
+      "loss": 0.3287,
+      "step": 3445
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715259461282964e-06,
+      "loss": 0.382,
+      "step": 3446
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.3797,
+      "step": 3447
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7053717924458478e-06,
+      "loss": 0.3659,
+      "step": 3448
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7004336638404683e-06,
+      "loss": 0.3015,
+      "step": 3449
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6954993425453726e-06,
+      "loss": 0.3082,
+      "step": 3450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6905688311342113e-06,
+      "loss": 0.3438,
+      "step": 3451
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.68564213217864e-06,
+      "loss": 0.3004,
+      "step": 3452
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6807192482483246e-06,
+      "loss": 0.267,
+      "step": 3453
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.675800181910948e-06,
+      "loss": 0.316,
+      "step": 3454
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.670884935732202e-06,
+      "loss": 0.3539,
+      "step": 3455
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.665973512275778e-06,
+      "loss": 0.3329,
+      "step": 3456
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6610659141033758e-06,
+      "loss": 0.3661,
+      "step": 3457
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6561621437747074e-06,
+      "loss": 0.4345,
+      "step": 3458
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6512622038474778e-06,
+      "loss": 0.366,
+      "step": 3459
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.646366096877405e-06,
+      "loss": 0.387,
+      "step": 3460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641473825418197e-06,
+      "loss": 0.3477,
+      "step": 3461
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.63658539202157e-06,
+      "loss": 0.2888,
+      "step": 3462
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6317007992372346e-06,
+      "loss": 0.2973,
+      "step": 3463
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6268200496128937e-06,
+      "loss": 0.2941,
+      "step": 3464
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6219431456942536e-06,
+      "loss": 0.3533,
+      "step": 3465
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6170700900250146e-06,
+      "loss": 0.2759,
+      "step": 3466
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6122008851468627e-06,
+      "loss": 0.3464,
+      "step": 3467
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.607335533599482e-06,
+      "loss": 0.3531,
+      "step": 3468
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.602474037920539e-06,
+      "loss": 0.3126,
+      "step": 3469
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5976164006456984e-06,
+      "loss": 0.3713,
+      "step": 3470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.59276262430861e-06,
+      "loss": 0.2966,
+      "step": 3471
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5879127114409064e-06,
+      "loss": 0.33,
+      "step": 3472
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5830666645722034e-06,
+      "loss": 0.2971,
+      "step": 3473
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.57822448623011e-06,
+      "loss": 0.2853,
+      "step": 3474
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5733861789402047e-06,
+      "loss": 0.3464,
+      "step": 3475
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.3473,
+      "step": 3476
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.563721187609214e-06,
+      "loss": 0.3623,
+      "step": 3477
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5588945086091985e-06,
+      "loss": 0.3534,
+      "step": 3478
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.554071710743511e-06,
+      "loss": 0.2945,
+      "step": 3479
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5492527965276246e-06,
+      "loss": 0.3463,
+      "step": 3480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.544437768474993e-06,
+      "loss": 0.3297,
+      "step": 3481
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.539626629097044e-06,
+      "loss": 0.3443,
+      "step": 3482
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5348193809031695e-06,
+      "loss": 0.2983,
+      "step": 3483
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5300160264007324e-06,
+      "loss": 0.3349,
+      "step": 3484
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5252165680950747e-06,
+      "loss": 0.3245,
+      "step": 3485
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5204210084894944e-06,
+      "loss": 0.3442,
+      "step": 3486
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5156293500852645e-06,
+      "loss": 0.3233,
+      "step": 3487
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.510841595381616e-06,
+      "loss": 0.2985,
+      "step": 3488
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.506057746875753e-06,
+      "loss": 0.2971,
+      "step": 3489
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.501277807062833e-06,
+      "loss": 0.3979,
+      "step": 3490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496501778435977e-06,
+      "loss": 0.2927,
+      "step": 3491
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4917296634862697e-06,
+      "loss": 0.2646,
+      "step": 3492
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.486961464702756e-06,
+      "loss": 0.3571,
+      "step": 3493
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4821971845724313e-06,
+      "loss": 0.3058,
+      "step": 3494
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4774368255802483e-06,
+      "loss": 0.314,
+      "step": 3495
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4726803902091213e-06,
+      "loss": 0.3255,
+      "step": 3496
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4679278809399076e-06,
+      "loss": 0.3451,
+      "step": 3497
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4631793002514293e-06,
+      "loss": 0.3617,
+      "step": 3498
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.458434650620447e-06,
+      "loss": 0.3097,
+      "step": 3499
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4536939345216814e-06,
+      "loss": 0.3553,
+      "step": 3500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4489571544277944e-06,
+      "loss": 0.3258,
+      "step": 3501
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.444224312809396e-06,
+      "loss": 0.3504,
+      "step": 3502
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4394954121350433e-06,
+      "loss": 0.3275,
+      "step": 3503
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.434770454871244e-06,
+      "loss": 0.3243,
+      "step": 3504
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 0.3041,
+      "step": 3505
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4253323804310115e-06,
+      "loss": 0.3695,
+      "step": 3506
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.42061926817729e-06,
+      "loss": 0.3266,
+      "step": 3507
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.415910109179543e-06,
+      "loss": 0.325,
+      "step": 3508
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.411204905893979e-06,
+      "loss": 0.3349,
+      "step": 3509
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4065036607747348e-06,
+      "loss": 0.3093,
+      "step": 3510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4018063762738863e-06,
+      "loss": 0.3123,
+      "step": 3511
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.397113054841449e-06,
+      "loss": 0.2789,
+      "step": 3512
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3924236989253623e-06,
+      "loss": 0.3106,
+      "step": 3513
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3877383109715047e-06,
+      "loss": 0.3315,
+      "step": 3514
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3830568934236854e-06,
+      "loss": 0.398,
+      "step": 3515
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3783794487236367e-06,
+      "loss": 0.3731,
+      "step": 3516
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3737059793110216e-06,
+      "loss": 0.3067,
+      "step": 3517
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3690364876234273e-06,
+      "loss": 0.3479,
+      "step": 3518
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.364370976096372e-06,
+      "loss": 0.3658,
+      "step": 3519
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3597094471632966e-06,
+      "loss": 0.3202,
+      "step": 3520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.355051903255562e-06,
+      "loss": 0.3729,
+      "step": 3521
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.350398346802446e-06,
+      "loss": 0.3728,
+      "step": 3522
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3457487802311597e-06,
+      "loss": 0.3328,
+      "step": 3523
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3411032059668203e-06,
+      "loss": 0.3494,
+      "step": 3524
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3364616264324725e-06,
+      "loss": 0.3639,
+      "step": 3525
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3318240440490667e-06,
+      "loss": 0.356,
+      "step": 3526
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3271904612354823e-06,
+      "loss": 0.3706,
+      "step": 3527
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.322560880408502e-06,
+      "loss": 0.3092,
+      "step": 3528
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.317935303982819e-06,
+      "loss": 0.2978,
+      "step": 3529
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3133137343710487e-06,
+      "loss": 0.3644,
+      "step": 3530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.308696173983711e-06,
+      "loss": 0.262,
+      "step": 3531
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3040826252292347e-06,
+      "loss": 0.3509,
+      "step": 3532
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.299473090513954e-06,
+      "loss": 0.2904,
+      "step": 3533
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2948675722421086e-06,
+      "loss": 0.3515,
+      "step": 3534
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.290266072815851e-06,
+      "loss": 0.3345,
+      "step": 3535
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2856685946352343e-06,
+      "loss": 0.2861,
+      "step": 3536
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281075140098208e-06,
+      "loss": 0.3527,
+      "step": 3537
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.276485711600631e-06,
+      "loss": 0.2944,
+      "step": 3538
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2719003115362593e-06,
+      "loss": 0.3154,
+      "step": 3539
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2673189422967425e-06,
+      "loss": 0.3452,
+      "step": 3540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2627416062716366e-06,
+      "loss": 0.3363,
+      "step": 3541
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2581683058483915e-06,
+      "loss": 0.3843,
+      "step": 3542
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25359904341235e-06,
+      "loss": 0.3267,
+      "step": 3543
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.249033821346748e-06,
+      "loss": 0.3132,
+      "step": 3544
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2444726420327136e-06,
+      "loss": 0.34,
+      "step": 3545
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2399155078492695e-06,
+      "loss": 0.3313,
+      "step": 3546
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2353624211733328e-06,
+      "loss": 0.3239,
+      "step": 3547
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2308133843796996e-06,
+      "loss": 0.3663,
+      "step": 3548
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226268399841055e-06,
+      "loss": 0.3739,
+      "step": 3549
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.221727469927979e-06,
+      "loss": 0.2889,
+      "step": 3550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.217190597008927e-06,
+      "loss": 0.3803,
+      "step": 3551
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.212657783450247e-06,
+      "loss": 0.2881,
+      "step": 3552
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2081290316161606e-06,
+      "loss": 0.2774,
+      "step": 3553
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2036043438687815e-06,
+      "loss": 0.34,
+      "step": 3554
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.199083722568095e-06,
+      "loss": 0.3768,
+      "step": 3555
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.194567170071965e-06,
+      "loss": 0.3891,
+      "step": 3556
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1900546887361398e-06,
+      "loss": 0.2801,
+      "step": 3557
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1855462809142423e-06,
+      "loss": 0.4434,
+      "step": 3558
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.181041948957768e-06,
+      "loss": 0.3422,
+      "step": 3559
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1765416952160857e-06,
+      "loss": 0.3566,
+      "step": 3560
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1720455220364443e-06,
+      "loss": 0.3538,
+      "step": 3561
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1675534317639525e-06,
+      "loss": 0.2944,
+      "step": 3562
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 0.3993,
+      "step": 3563
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.158581509310246e-06,
+      "loss": 0.3858,
+      "step": 3564
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154101681808608e-06,
+      "loss": 0.3813,
+      "step": 3565
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1496259465732783e-06,
+      "loss": 0.346,
+      "step": 3566
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1451543059387094e-06,
+      "loss": 0.2931,
+      "step": 3567
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1406867622372227e-06,
+      "loss": 0.3839,
+      "step": 3568
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1362233177990055e-06,
+      "loss": 0.3277,
+      "step": 3569
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.131763974952099e-06,
+      "loss": 0.3768,
+      "step": 3570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1273087360224086e-06,
+      "loss": 0.3129,
+      "step": 3571
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1228576033336968e-06,
+      "loss": 0.3301,
+      "step": 3572
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1184105792075894e-06,
+      "loss": 0.3146,
+      "step": 3573
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1139676659635698e-06,
+      "loss": 0.292,
+      "step": 3574
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1095288659189696e-06,
+      "loss": 0.3351,
+      "step": 3575
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1050941813889837e-06,
+      "loss": 0.3304,
+      "step": 3576
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100663614686653e-06,
+      "loss": 0.2472,
+      "step": 3577
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0962371681228732e-06,
+      "loss": 0.3485,
+      "step": 3578
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0918148440063948e-06,
+      "loss": 0.365,
+      "step": 3579
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0873966446438156e-06,
+      "loss": 0.3214,
+      "step": 3580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0829825723395803e-06,
+      "loss": 0.3031,
+      "step": 3581
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0785726293959807e-06,
+      "loss": 0.3626,
+      "step": 3582
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0741668181131557e-06,
+      "loss": 0.3491,
+      "step": 3583
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0697651407890928e-06,
+      "loss": 0.3275,
+      "step": 3584
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.065367599719621e-06,
+      "loss": 0.3639,
+      "step": 3585
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0609741971984087e-06,
+      "loss": 0.3716,
+      "step": 3586
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0565849355169675e-06,
+      "loss": 0.3349,
+      "step": 3587
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.052199816964653e-06,
+      "loss": 0.3339,
+      "step": 3588
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047818843828654e-06,
+      "loss": 0.2869,
+      "step": 3589
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043442018394003e-06,
+      "loss": 0.301,
+      "step": 3590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0390693429435626e-06,
+      "loss": 0.3416,
+      "step": 3591
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0347008197580376e-06,
+      "loss": 0.3359,
+      "step": 3592
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.030336451115964e-06,
+      "loss": 0.348,
+      "step": 3593
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0259762392937053e-06,
+      "loss": 0.2952,
+      "step": 3594
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.021620186565466e-06,
+      "loss": 0.3319,
+      "step": 3595
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0172682952032807e-06,
+      "loss": 0.3835,
+      "step": 3596
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.012920567477007e-06,
+      "loss": 0.2419,
+      "step": 3597
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0085770056543317e-06,
+      "loss": 0.3797,
+      "step": 3598
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0042376120007766e-06,
+      "loss": 0.3045,
+      "step": 3599
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9999023887796797e-06,
+      "loss": 0.3978,
+      "step": 3600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9955713382522113e-06,
+      "loss": 0.3513,
+      "step": 3601
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9912444626773598e-06,
+      "loss": 0.3997,
+      "step": 3602
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986921764311941e-06,
+      "loss": 0.3475,
+      "step": 3603
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9826032454105883e-06,
+      "loss": 0.3492,
+      "step": 3604
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.978288908225754e-06,
+      "loss": 0.2903,
+      "step": 3605
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9739787550077116e-06,
+      "loss": 0.3821,
+      "step": 3606
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9696727880045587e-06,
+      "loss": 0.3106,
+      "step": 3607
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.965371009462197e-06,
+      "loss": 0.3874,
+      "step": 3608
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.961073421624352e-06,
+      "loss": 0.2976,
+      "step": 3609
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9567800267325567e-06,
+      "loss": 0.2955,
+      "step": 3610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9524908270261655e-06,
+      "loss": 0.2742,
+      "step": 3611
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.948205824742342e-06,
+      "loss": 0.2855,
+      "step": 3612
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9439250221160576e-06,
+      "loss": 0.2963,
+      "step": 3613
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9396484213800914e-06,
+      "loss": 0.3488,
+      "step": 3614
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.93537602476504e-06,
+      "loss": 0.3488,
+      "step": 3615
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.931107834499296e-06,
+      "loss": 0.2904,
+      "step": 3616
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.926843852809067e-06,
+      "loss": 0.3354,
+      "step": 3617
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9225840819183626e-06,
+      "loss": 0.3045,
+      "step": 3618
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9183285240489945e-06,
+      "loss": 0.4146,
+      "step": 3619
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9140771814205783e-06,
+      "loss": 0.3727,
+      "step": 3620
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.3086,
+      "step": 3621
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9055871507540603e-06,
+      "loss": 0.3166,
+      "step": 3622
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9013484671441973e-06,
+      "loss": 0.2985,
+      "step": 3623
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8971140076317497e-06,
+      "loss": 0.2593,
+      "step": 3624
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8928837744253259e-06,
+      "loss": 0.3161,
+      "step": 3625
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8886577697313357e-06,
+      "loss": 0.3135,
+      "step": 3626
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8844359957539783e-06,
+      "loss": 0.4362,
+      "step": 3627
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8802184546952496e-06,
+      "loss": 0.3294,
+      "step": 3628
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8760051487549325e-06,
+      "loss": 0.325,
+      "step": 3629
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8717960801306101e-06,
+      "loss": 0.3264,
+      "step": 3630
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8675912510176476e-06,
+      "loss": 0.2918,
+      "step": 3631
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.863390663609197e-06,
+      "loss": 0.3653,
+      "step": 3632
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8591943200962059e-06,
+      "loss": 0.2961,
+      "step": 3633
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8550022226674059e-06,
+      "loss": 0.3147,
+      "step": 3634
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.850814373509312e-06,
+      "loss": 0.3035,
+      "step": 3635
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8466307748062206e-06,
+      "loss": 0.3737,
+      "step": 3636
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8424514287402184e-06,
+      "loss": 0.3908,
+      "step": 3637
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.838276337491166e-06,
+      "loss": 0.4086,
+      "step": 3638
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8341055032367149e-06,
+      "loss": 0.2418,
+      "step": 3639
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8299389281522838e-06,
+      "loss": 0.3538,
+      "step": 3640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8257766144110823e-06,
+      "loss": 0.2782,
+      "step": 3641
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8216185641840878e-06,
+      "loss": 0.3075,
+      "step": 3642
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8174647796400568e-06,
+      "loss": 0.3425,
+      "step": 3643
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8133152629455219e-06,
+      "loss": 0.3531,
+      "step": 3644
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.809170016264794e-06,
+      "loss": 0.298,
+      "step": 3645
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8050290417599491e-06,
+      "loss": 0.3381,
+      "step": 3646
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8008923415908375e-06,
+      "loss": 0.3443,
+      "step": 3647
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7967599179150786e-06,
+      "loss": 0.3636,
+      "step": 3648
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.792631772888065e-06,
+      "loss": 0.3582,
+      "step": 3649
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7885079086629598e-06,
+      "loss": 0.3174,
+      "step": 3650
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.784388327390687e-06,
+      "loss": 0.3549,
+      "step": 3651
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7802730312199357e-06,
+      "loss": 0.261,
+      "step": 3652
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7761620222971698e-06,
+      "loss": 0.3606,
+      "step": 3653
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7720553027666042e-06,
+      "loss": 0.3189,
+      "step": 3654
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7679528747702257e-06,
+      "loss": 0.3208,
+      "step": 3655
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7638547404477845e-06,
+      "loss": 0.3842,
+      "step": 3656
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7597609019367822e-06,
+      "loss": 0.3134,
+      "step": 3657
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7556713613724852e-06,
+      "loss": 0.3857,
+      "step": 3658
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7515861208879147e-06,
+      "loss": 0.3299,
+      "step": 3659
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7475051826138555e-06,
+      "loss": 0.2933,
+      "step": 3660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.743428548678847e-06,
+      "loss": 0.2679,
+      "step": 3661
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7393562212091775e-06,
+      "loss": 0.3235,
+      "step": 3662
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.735288202328891e-06,
+      "loss": 0.3159,
+      "step": 3663
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.731224494159791e-06,
+      "loss": 0.2883,
+      "step": 3664
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.727165098821425e-06,
+      "loss": 0.3961,
+      "step": 3665
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7231100184310955e-06,
+      "loss": 0.302,
+      "step": 3666
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.71905925510385e-06,
+      "loss": 0.3043,
+      "step": 3667
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.71501281095249e-06,
+      "loss": 0.2865,
+      "step": 3668
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.710970688087561e-06,
+      "loss": 0.28,
+      "step": 3669
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.706932888617353e-06,
+      "loss": 0.3249,
+      "step": 3670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7028994146478983e-06,
+      "loss": 0.276,
+      "step": 3671
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6988702682829894e-06,
+      "loss": 0.3229,
+      "step": 3672
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.694845451624142e-06,
+      "loss": 0.3239,
+      "step": 3673
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.69082496677062e-06,
+      "loss": 0.2987,
+      "step": 3674
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6868088158194351e-06,
+      "loss": 0.2273,
+      "step": 3675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6827970008653294e-06,
+      "loss": 0.3306,
+      "step": 3676
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6787895240007833e-06,
+      "loss": 0.3411,
+      "step": 3677
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.674786387316023e-06,
+      "loss": 0.3232,
+      "step": 3678
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 0.2664,
+      "step": 3679
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6667931428354222e-06,
+      "loss": 0.4062,
+      "step": 3680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6628030392087001e-06,
+      "loss": 0.3107,
+      "step": 3681
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6588172840999938e-06,
+      "loss": 0.3632,
+      "step": 3682
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6548358795882037e-06,
+      "loss": 0.387,
+      "step": 3683
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6508588277499481e-06,
+      "loss": 0.3639,
+      "step": 3684
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6468861306595786e-06,
+      "loss": 0.3813,
+      "step": 3685
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6429177903891734e-06,
+      "loss": 0.2873,
+      "step": 3686
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6389538090085467e-06,
+      "loss": 0.3485,
+      "step": 3687
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.634994188585226e-06,
+      "loss": 0.4085,
+      "step": 3688
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6310389311844787e-06,
+      "loss": 0.3341,
+      "step": 3689
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6270880388692833e-06,
+      "loss": 0.3676,
+      "step": 3690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6231415137003536e-06,
+      "loss": 0.3324,
+      "step": 3691
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6191993577361155e-06,
+      "loss": 0.3496,
+      "step": 3692
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6152615730327159e-06,
+      "loss": 0.3625,
+      "step": 3693
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6113281616440357e-06,
+      "loss": 0.3804,
+      "step": 3694
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6073991256216603e-06,
+      "loss": 0.3657,
+      "step": 3695
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6034744670148972e-06,
+      "loss": 0.338,
+      "step": 3696
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5995541878707677e-06,
+      "loss": 0.3749,
+      "step": 3697
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5956382902340195e-06,
+      "loss": 0.3617,
+      "step": 3698
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5917267761471022e-06,
+      "loss": 0.3125,
+      "step": 3699
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5878196476501883e-06,
+      "loss": 0.3612,
+      "step": 3700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5839169067811555e-06,
+      "loss": 0.3364,
+      "step": 3701
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.580018555575601e-06,
+      "loss": 0.3395,
+      "step": 3702
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.576124596066826e-06,
+      "loss": 0.3705,
+      "step": 3703
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5722350302858401e-06,
+      "loss": 0.4205,
+      "step": 3704
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5683498602613689e-06,
+      "loss": 0.4033,
+      "step": 3705
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.56446908801984e-06,
+      "loss": 0.3625,
+      "step": 3706
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5605927155853884e-06,
+      "loss": 0.3555,
+      "step": 3707
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5567207449798517e-06,
+      "loss": 0.2865,
+      "step": 3708
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5528531782227718e-06,
+      "loss": 0.3713,
+      "step": 3709
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5489900173313987e-06,
+      "loss": 0.3958,
+      "step": 3710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5451312643206828e-06,
+      "loss": 0.325,
+      "step": 3711
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5412769212032718e-06,
+      "loss": 0.2895,
+      "step": 3712
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5374269899895122e-06,
+      "loss": 0.2765,
+      "step": 3713
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5335814726874576e-06,
+      "loss": 0.3018,
+      "step": 3714
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5297403713028502e-06,
+      "loss": 0.3551,
+      "step": 3715
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5259036878391342e-06,
+      "loss": 0.2968,
+      "step": 3716
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.522071424297451e-06,
+      "loss": 0.3239,
+      "step": 3717
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5182435826766318e-06,
+      "loss": 0.311,
+      "step": 3718
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5144201649732037e-06,
+      "loss": 0.3392,
+      "step": 3719
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.510601173181383e-06,
+      "loss": 0.3224,
+      "step": 3720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.506786609293084e-06,
+      "loss": 0.3392,
+      "step": 3721
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5029764752979093e-06,
+      "loss": 0.3005,
+      "step": 3722
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4991707731831494e-06,
+      "loss": 0.2874,
+      "step": 3723
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.495369504933779e-06,
+      "loss": 0.3056,
+      "step": 3724
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.491572672532473e-06,
+      "loss": 0.2934,
+      "step": 3725
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4877802779595762e-06,
+      "loss": 0.3332,
+      "step": 3726
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4839923231931342e-06,
+      "loss": 0.3444,
+      "step": 3727
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4802088102088642e-06,
+      "loss": 0.3837,
+      "step": 3728
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4764297409801764e-06,
+      "loss": 0.3524,
+      "step": 3729
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4726551174781567e-06,
+      "loss": 0.3302,
+      "step": 3730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4688849416715734e-06,
+      "loss": 0.3346,
+      "step": 3731
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4651192155268768e-06,
+      "loss": 0.3499,
+      "step": 3732
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.461357941008199e-06,
+      "loss": 0.3713,
+      "step": 3733
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4576011200773443e-06,
+      "loss": 0.3195,
+      "step": 3734
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4538487546937951e-06,
+      "loss": 0.3981,
+      "step": 3735
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4501008468147148e-06,
+      "loss": 0.3233,
+      "step": 3736
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 0.3129,
+      "step": 3737
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4426184113869667e-06,
+      "loss": 0.3525,
+      "step": 3738
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4388838877409906e-06,
+      "loss": 0.3071,
+      "step": 3739
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.435153829404864e-06,
+      "loss": 0.3218,
+      "step": 3740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 0.2982,
+      "step": 3741
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.427707116441921e-06,
+      "loss": 0.2778,
+      "step": 3742
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4239904656991633e-06,
+      "loss": 0.3613,
+      "step": 3743
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4202782880343713e-06,
+      "loss": 0.3585,
+      "step": 3744
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.416570585383742e-06,
+      "loss": 0.3316,
+      "step": 3745
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4128673596811404e-06,
+      "loss": 0.2486,
+      "step": 3746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4091686128580927e-06,
+      "loss": 0.2902,
+      "step": 3747
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4054743468437971e-06,
+      "loss": 0.3206,
+      "step": 3748
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4017845635651096e-06,
+      "loss": 0.2809,
+      "step": 3749
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3980992649465475e-06,
+      "loss": 0.3621,
+      "step": 3750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3944184529102878e-06,
+      "loss": 0.3525,
+      "step": 3751
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3907421293761747e-06,
+      "loss": 0.3171,
+      "step": 3752
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.387070296261701e-06,
+      "loss": 0.3031,
+      "step": 3753
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3834029554820261e-06,
+      "loss": 0.3508,
+      "step": 3754
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.379740108949964e-06,
+      "loss": 0.3717,
+      "step": 3755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.376081758575981e-06,
+      "loss": 0.3185,
+      "step": 3756
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3724279062682022e-06,
+      "loss": 0.3406,
+      "step": 3757
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3687785539324005e-06,
+      "loss": 0.3493,
+      "step": 3758
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3651337034720113e-06,
+      "loss": 0.3756,
+      "step": 3759
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3614933567881172e-06,
+      "loss": 0.3529,
+      "step": 3760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3578575157794504e-06,
+      "loss": 0.3182,
+      "step": 3761
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3542261823423908e-06,
+      "loss": 0.3484,
+      "step": 3762
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3505993583709753e-06,
+      "loss": 0.3027,
+      "step": 3763
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.346977045756881e-06,
+      "loss": 0.3463,
+      "step": 3764
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3433592463894375e-06,
+      "loss": 0.3495,
+      "step": 3765
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.2171,
+      "step": 3766
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3361371949400325e-06,
+      "loss": 0.4176,
+      "step": 3767
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3325329466249536e-06,
+      "loss": 0.2562,
+      "step": 3768
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3289332190902793e-06,
+      "loss": 0.3542,
+      "step": 3769
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3253380142135608e-06,
+      "loss": 0.3392,
+      "step": 3770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.321747333869986e-06,
+      "loss": 0.3195,
+      "step": 3771
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3181611799323824e-06,
+      "loss": 0.3404,
+      "step": 3772
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3145795542712147e-06,
+      "loss": 0.3268,
+      "step": 3773
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3110024587545933e-06,
+      "loss": 0.3534,
+      "step": 3774
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3074298952482568e-06,
+      "loss": 0.3159,
+      "step": 3775
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3038618656155877e-06,
+      "loss": 0.3786,
+      "step": 3776
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3002983717175955e-06,
+      "loss": 0.3432,
+      "step": 3777
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2967394154129332e-06,
+      "loss": 0.3091,
+      "step": 3778
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2931849985578804e-06,
+      "loss": 0.3559,
+      "step": 3779
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2896351230063497e-06,
+      "loss": 0.3007,
+      "step": 3780
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2860897906098869e-06,
+      "loss": 0.3224,
+      "step": 3781
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2825490032176714e-06,
+      "loss": 0.2721,
+      "step": 3782
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2790127626765048e-06,
+      "loss": 0.3488,
+      "step": 3783
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2754810708308208e-06,
+      "loss": 0.3688,
+      "step": 3784
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2719539295226802e-06,
+      "loss": 0.277,
+      "step": 3785
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2684313405917702e-06,
+      "loss": 0.3549,
+      "step": 3786
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2649133058754071e-06,
+      "loss": 0.2941,
+      "step": 3787
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.261399827208527e-06,
+      "loss": 0.4215,
+      "step": 3788
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2578909064236887e-06,
+      "loss": 0.3759,
+      "step": 3789
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2543865453510795e-06,
+      "loss": 0.4101,
+      "step": 3790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2508867458185037e-06,
+      "loss": 0.2958,
+      "step": 3791
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2473915096513888e-06,
+      "loss": 0.2816,
+      "step": 3792
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2439008386727802e-06,
+      "loss": 0.3321,
+      "step": 3793
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2404147347033458e-06,
+      "loss": 0.3255,
+      "step": 3794
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 0.3276,
+      "step": 3795
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.233456235062741e-06,
+      "loss": 0.2827,
+      "step": 3796
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2299838430209876e-06,
+      "loss": 0.337,
+      "step": 3797
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2265160252472396e-06,
+      "loss": 0.3235,
+      "step": 3798
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2230527835502414e-06,
+      "loss": 0.3427,
+      "step": 3799
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2195941197363491e-06,
+      "loss": 0.3282,
+      "step": 3800
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2161400356095376e-06,
+      "loss": 0.3527,
+      "step": 3801
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2126905329713857e-06,
+      "loss": 0.3688,
+      "step": 3802
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2092456136210894e-06,
+      "loss": 0.2903,
+      "step": 3803
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2058052793554476e-06,
+      "loss": 0.3771,
+      "step": 3804
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2023695319688755e-06,
+      "loss": 0.3452,
+      "step": 3805
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1989383732533876e-06,
+      "loss": 0.3201,
+      "step": 3806
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1955118049986091e-06,
+      "loss": 0.2958,
+      "step": 3807
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1920898289917715e-06,
+      "loss": 0.3035,
+      "step": 3808
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1886724470177113e-06,
+      "loss": 0.3654,
+      "step": 3809
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1852596608588674e-06,
+      "loss": 0.2885,
+      "step": 3810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1818514722952812e-06,
+      "loss": 0.2959,
+      "step": 3811
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1784478831045941e-06,
+      "loss": 0.3298,
+      "step": 3812
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1750488950620543e-06,
+      "loss": 0.3021,
+      "step": 3813
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1716545099405096e-06,
+      "loss": 0.334,
+      "step": 3814
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1682647295104e-06,
+      "loss": 0.2466,
+      "step": 3815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1648795555397719e-06,
+      "loss": 0.2892,
+      "step": 3816
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.161498989794263e-06,
+      "loss": 0.3492,
+      "step": 3817
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1581230340371097e-06,
+      "loss": 0.3017,
+      "step": 3818
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1547516900291445e-06,
+      "loss": 0.3815,
+      "step": 3819
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1513849595287962e-06,
+      "loss": 0.4004,
+      "step": 3820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1480228442920848e-06,
+      "loss": 0.2723,
+      "step": 3821
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1446653460726232e-06,
+      "loss": 0.3172,
+      "step": 3822
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1413124666216124e-06,
+      "loss": 0.3404,
+      "step": 3823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1379642076878528e-06,
+      "loss": 0.2608,
+      "step": 3824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1346205710177304e-06,
+      "loss": 0.2821,
+      "step": 3825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1312815583552205e-06,
+      "loss": 0.3143,
+      "step": 3826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1279471714418833e-06,
+      "loss": 0.3167,
+      "step": 3827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.124617412016873e-06,
+      "loss": 0.4076,
+      "step": 3828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1212922818169224e-06,
+      "loss": 0.3384,
+      "step": 3829
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1179717825763602e-06,
+      "loss": 0.3418,
+      "step": 3830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1146559160270875e-06,
+      "loss": 0.3584,
+      "step": 3831
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1113446838985997e-06,
+      "loss": 0.3532,
+      "step": 3832
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1080380879179686e-06,
+      "loss": 0.3506,
+      "step": 3833
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.104736129809847e-06,
+      "loss": 0.3694,
+      "step": 3834
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1014388112964747e-06,
+      "loss": 0.3241,
+      "step": 3835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0981461340976684e-06,
+      "loss": 0.3189,
+      "step": 3836
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0948580999308234e-06,
+      "loss": 0.3201,
+      "step": 3837
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0915747105109097e-06,
+      "loss": 0.3443,
+      "step": 3838
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0882959675504845e-06,
+      "loss": 0.286,
+      "step": 3839
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0850218727596718e-06,
+      "loss": 0.3038,
+      "step": 3840
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0817524278461777e-06,
+      "loss": 0.3053,
+      "step": 3841
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0784876345152772e-06,
+      "loss": 0.3365,
+      "step": 3842
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.075227494469826e-06,
+      "loss": 0.3354,
+      "step": 3843
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0719720094102482e-06,
+      "loss": 0.3203,
+      "step": 3844
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0687211810345366e-06,
+      "loss": 0.3072,
+      "step": 3845
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0654750110382627e-06,
+      "loss": 0.3447,
+      "step": 3846
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0622335011145678e-06,
+      "loss": 0.2835,
+      "step": 3847
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0589966529541562e-06,
+      "loss": 0.3313,
+      "step": 3848
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.055764468245304e-06,
+      "loss": 0.3782,
+      "step": 3849
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0525369486738546e-06,
+      "loss": 0.3482,
+      "step": 3850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0493140959232206e-06,
+      "loss": 0.3591,
+      "step": 3851
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0460959116743807e-06,
+      "loss": 0.3161,
+      "step": 3852
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 0.2584,
+      "step": 3853
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.039673555393802e-06,
+      "loss": 0.3411,
+      "step": 3854
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0364693867118425e-06,
+      "loss": 0.3165,
+      "step": 3855
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0332698932312191e-06,
+      "loss": 0.3496,
+      "step": 3856
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.030075076620729e-06,
+      "loss": 0.3494,
+      "step": 3857
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0268849385467262e-06,
+      "loss": 0.3574,
+      "step": 3858
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0236994806731225e-06,
+      "loss": 0.2873,
+      "step": 3859
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0205187046613897e-06,
+      "loss": 0.3464,
+      "step": 3860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0173426121705577e-06,
+      "loss": 0.2931,
+      "step": 3861
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.014171204857214e-06,
+      "loss": 0.3358,
+      "step": 3862
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.011004484375504e-06,
+      "loss": 0.2782,
+      "step": 3863
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0078424523771268e-06,
+      "loss": 0.2911,
+      "step": 3864
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0046851105113331e-06,
+      "loss": 0.3551,
+      "step": 3865
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0015324604249343e-06,
+      "loss": 0.3564,
+      "step": 3866
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.983845037622876e-07,
+      "loss": 0.3716,
+      "step": 3867
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.95241242165308e-07,
+      "loss": 0.3256,
+      "step": 3868
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.921026772734543e-07,
+      "loss": 0.2986,
+      "step": 3869
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.889688107237462e-07,
+      "loss": 0.2699,
+      "step": 3870
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.858396441507445e-07,
+      "loss": 0.3244,
+      "step": 3871
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.827151791865576e-07,
+      "loss": 0.3075,
+      "step": 3872
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.795954174608491e-07,
+      "loss": 0.298,
+      "step": 3873
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.764803606008255e-07,
+      "loss": 0.3457,
+      "step": 3874
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.733700102312372e-07,
+      "loss": 0.3305,
+      "step": 3875
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.702643679743817e-07,
+      "loss": 0.3359,
+      "step": 3876
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.671634354501036e-07,
+      "loss": 0.2802,
+      "step": 3877
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.640672142757834e-07,
+      "loss": 0.3714,
+      "step": 3878
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.60975706066356e-07,
+      "loss": 0.3206,
+      "step": 3879
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.57888912434284e-07,
+      "loss": 0.3389,
+      "step": 3880
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.548068349895845e-07,
+      "loss": 0.3746,
+      "step": 3881
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.517294753398066e-07,
+      "loss": 0.2861,
+      "step": 3882
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.48656835090036e-07,
+      "loss": 0.3218,
+      "step": 3883
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.455889158429065e-07,
+      "loss": 0.3163,
+      "step": 3884
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.425257191985859e-07,
+      "loss": 0.2549,
+      "step": 3885
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.394672467547739e-07,
+      "loss": 0.3958,
+      "step": 3886
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.364135001067098e-07,
+      "loss": 0.3507,
+      "step": 3887
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.333644808471676e-07,
+      "loss": 0.3232,
+      "step": 3888
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.30320190566456e-07,
+      "loss": 0.338,
+      "step": 3889
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.272806308524196e-07,
+      "loss": 0.3248,
+      "step": 3890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.242458032904311e-07,
+      "loss": 0.3373,
+      "step": 3891
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.212157094633945e-07,
+      "loss": 0.3437,
+      "step": 3892
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.181903509517509e-07,
+      "loss": 0.3081,
+      "step": 3893
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.151697293334649e-07,
+      "loss": 0.2747,
+      "step": 3894
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.121538461840352e-07,
+      "loss": 0.2732,
+      "step": 3895
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.091427030764888e-07,
+      "loss": 0.3291,
+      "step": 3896
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.061363015813763e-07,
+      "loss": 0.2385,
+      "step": 3897
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.031346432667775e-07,
+      "loss": 0.3474,
+      "step": 3898
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.001377296982972e-07,
+      "loss": 0.2994,
+      "step": 3899
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.97145562439068e-07,
+      "loss": 0.4047,
+      "step": 3900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.941581430497481e-07,
+      "loss": 0.3022,
+      "step": 3901
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.911754730885147e-07,
+      "loss": 0.3294,
+      "step": 3902
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.881975541110666e-07,
+      "loss": 0.3616,
+      "step": 3903
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.852243876706334e-07,
+      "loss": 0.3147,
+      "step": 3904
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.82255975317956e-07,
+      "loss": 0.3271,
+      "step": 3905
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.792923186013025e-07,
+      "loss": 0.3662,
+      "step": 3906
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.763334190664552e-07,
+      "loss": 0.2877,
+      "step": 3907
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.733792782567219e-07,
+      "loss": 0.35,
+      "step": 3908
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.70429897712921e-07,
+      "loss": 0.3307,
+      "step": 3909
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.674852789733923e-07,
+      "loss": 0.3633,
+      "step": 3910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.2775,
+      "step": 3911
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.616103330480885e-07,
+      "loss": 0.3069,
+      "step": 3912
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.586800089265712e-07,
+      "loss": 0.3054,
+      "step": 3913
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.557544527378359e-07,
+      "loss": 0.3649,
+      "step": 3914
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.528336660077974e-07,
+      "loss": 0.3382,
+      "step": 3915
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.499176502598783e-07,
+      "loss": 0.3078,
+      "step": 3916
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.470064070150186e-07,
+      "loss": 0.324,
+      "step": 3917
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.440999377916614e-07,
+      "loss": 0.2791,
+      "step": 3918
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.411982441057687e-07,
+      "loss": 0.3767,
+      "step": 3919
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.38301327470804e-07,
+      "loss": 0.3258,
+      "step": 3920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.354091893977401e-07,
+      "loss": 0.2883,
+      "step": 3921
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.325218313950611e-07,
+      "loss": 0.4023,
+      "step": 3922
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.296392549687593e-07,
+      "loss": 0.2962,
+      "step": 3923
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.267614616223274e-07,
+      "loss": 0.2797,
+      "step": 3924
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.238884528567648e-07,
+      "loss": 0.354,
+      "step": 3925
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.210202301705761e-07,
+      "loss": 0.3998,
+      "step": 3926
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.181567950597713e-07,
+      "loss": 0.2989,
+      "step": 3927
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.152981490178625e-07,
+      "loss": 0.2848,
+      "step": 3928
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.124442935358623e-07,
+      "loss": 0.267,
+      "step": 3929
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.095952301022847e-07,
+      "loss": 0.3301,
+      "step": 3930
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.067509602031465e-07,
+      "loss": 0.3425,
+      "step": 3931
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.039114853219621e-07,
+      "loss": 0.2884,
+      "step": 3932
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.010768069397457e-07,
+      "loss": 0.3683,
+      "step": 3933
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.982469265350124e-07,
+      "loss": 0.3456,
+      "step": 3934
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.954218455837703e-07,
+      "loss": 0.3041,
+      "step": 3935
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.926015655595254e-07,
+      "loss": 0.3326,
+      "step": 3936
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.897860879332797e-07,
+      "loss": 0.3111,
+      "step": 3937
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.86975414173532e-07,
+      "loss": 0.3585,
+      "step": 3938
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.84169545746275e-07,
+      "loss": 0.3585,
+      "step": 3939
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.81368484114996e-07,
+      "loss": 0.3447,
+      "step": 3940
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.785722307406685e-07,
+      "loss": 0.391,
+      "step": 3941
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.757807870817691e-07,
+      "loss": 0.3622,
+      "step": 3942
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.729941545942577e-07,
+      "loss": 0.3024,
+      "step": 3943
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.702123347315837e-07,
+      "loss": 0.3406,
+      "step": 3944
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.674353289446946e-07,
+      "loss": 0.3577,
+      "step": 3945
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.646631386820202e-07,
+      "loss": 0.3425,
+      "step": 3946
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.618957653894809e-07,
+      "loss": 0.331,
+      "step": 3947
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.591332105104832e-07,
+      "loss": 0.3481,
+      "step": 3948
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.563754754859187e-07,
+      "loss": 0.3283,
+      "step": 3949
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.536225617541736e-07,
+      "loss": 0.307,
+      "step": 3950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.508744707511118e-07,
+      "loss": 0.4131,
+      "step": 3951
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.48131203910083e-07,
+      "loss": 0.2952,
+      "step": 3952
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.453927626619184e-07,
+      "loss": 0.3101,
+      "step": 3953
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.426591484349388e-07,
+      "loss": 0.3513,
+      "step": 3954
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.399303626549415e-07,
+      "loss": 0.3283,
+      "step": 3955
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.372064067452078e-07,
+      "loss": 0.3194,
+      "step": 3956
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.344872821265025e-07,
+      "loss": 0.3522,
+      "step": 3957
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.317729902170656e-07,
+      "loss": 0.271,
+      "step": 3958
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.290635324326168e-07,
+      "loss": 0.3543,
+      "step": 3959
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.263589101863544e-07,
+      "loss": 0.3485,
+      "step": 3960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.236591248889624e-07,
+      "loss": 0.3296,
+      "step": 3961
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.209641779485921e-07,
+      "loss": 0.3481,
+      "step": 3962
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.182740707708757e-07,
+      "loss": 0.3211,
+      "step": 3963
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.155888047589177e-07,
+      "loss": 0.359,
+      "step": 3964
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.129083813133031e-07,
+      "loss": 0.3016,
+      "step": 3965
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.102328018320859e-07,
+      "loss": 0.2886,
+      "step": 3966
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.075620677107975e-07,
+      "loss": 0.2822,
+      "step": 3967
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.048961803424393e-07,
+      "loss": 0.3268,
+      "step": 3968
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.3324,
+      "step": 3969
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.995789514238849e-07,
+      "loss": 0.3531,
+      "step": 3970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.969276126470493e-07,
+      "loss": 0.3586,
+      "step": 3971
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.942811261698657e-07,
+      "loss": 0.3341,
+      "step": 3972
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.916394933726923e-07,
+      "loss": 0.3396,
+      "step": 3973
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.890027156333512e-07,
+      "loss": 0.3923,
+      "step": 3974
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.863707943271325e-07,
+      "loss": 0.3095,
+      "step": 3975
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.83743730826798e-07,
+      "loss": 0.3209,
+      "step": 3976
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.811215265025673e-07,
+      "loss": 0.3382,
+      "step": 3977
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.785041827221361e-07,
+      "loss": 0.3525,
+      "step": 3978
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.758917008506549e-07,
+      "loss": 0.3952,
+      "step": 3979
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.732840822507469e-07,
+      "loss": 0.3082,
+      "step": 3980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.70681328282492e-07,
+      "loss": 0.3649,
+      "step": 3981
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.680834403034342e-07,
+      "loss": 0.328,
+      "step": 3982
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.654904196685841e-07,
+      "loss": 0.2733,
+      "step": 3983
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.629022677304098e-07,
+      "loss": 0.3313,
+      "step": 3984
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.603189858388393e-07,
+      "loss": 0.3301,
+      "step": 3985
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.577405753412625e-07,
+      "loss": 0.316,
+      "step": 3986
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.551670375825259e-07,
+      "loss": 0.3411,
+      "step": 3987
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.525983739049369e-07,
+      "loss": 0.3209,
+      "step": 3988
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.500345856482615e-07,
+      "loss": 0.3869,
+      "step": 3989
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.474756741497212e-07,
+      "loss": 0.3055,
+      "step": 3990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.449216407439906e-07,
+      "loss": 0.2998,
+      "step": 3991
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.423724867632075e-07,
+      "loss": 0.375,
+      "step": 3992
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.398282135369571e-07,
+      "loss": 0.3599,
+      "step": 3993
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.372888223922835e-07,
+      "loss": 0.3313,
+      "step": 3994
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.34754314653685e-07,
+      "loss": 0.2959,
+      "step": 3995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.322246916431108e-07,
+      "loss": 0.2897,
+      "step": 3996
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.296999546799599e-07,
+      "loss": 0.2845,
+      "step": 3997
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.271801050810856e-07,
+      "loss": 0.3286,
+      "step": 3998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.246651441607931e-07,
+      "loss": 0.3057,
+      "step": 3999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.221550732308379e-07,
+      "loss": 0.3684,
+      "step": 4000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.196498936004236e-07,
+      "loss": 0.3014,
+      "step": 4001
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.171496065761983e-07,
+      "loss": 0.3694,
+      "step": 4002
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.14654213462268e-07,
+      "loss": 0.3172,
+      "step": 4003
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.121637155601767e-07,
+      "loss": 0.3323,
+      "step": 4004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.096781141689223e-07,
+      "loss": 0.3338,
+      "step": 4005
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.071974105849432e-07,
+      "loss": 0.3089,
+      "step": 4006
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.047216061021266e-07,
+      "loss": 0.3621,
+      "step": 4007
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.02250702011804e-07,
+      "loss": 0.2894,
+      "step": 4008
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.99784699602749e-07,
+      "loss": 0.3633,
+      "step": 4009
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.973236001611805e-07,
+      "loss": 0.3443,
+      "step": 4010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.948674049707604e-07,
+      "loss": 0.279,
+      "step": 4011
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.924161153125918e-07,
+      "loss": 0.3245,
+      "step": 4012
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.899697324652165e-07,
+      "loss": 0.3109,
+      "step": 4013
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.875282577046249e-07,
+      "loss": 0.3017,
+      "step": 4014
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.85091692304236e-07,
+      "loss": 0.3837,
+      "step": 4015
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.826600375349201e-07,
+      "loss": 0.3249,
+      "step": 4016
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.802332946649757e-07,
+      "loss": 0.3024,
+      "step": 4017
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.778114649601485e-07,
+      "loss": 0.3594,
+      "step": 4018
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.753945496836144e-07,
+      "loss": 0.3236,
+      "step": 4019
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.729825500959885e-07,
+      "loss": 0.3501,
+      "step": 4020
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.705754674553243e-07,
+      "loss": 0.3554,
+      "step": 4021
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.681733030171089e-07,
+      "loss": 0.3159,
+      "step": 4022
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.65776058034263e-07,
+      "loss": 0.3526,
+      "step": 4023
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.633837337571435e-07,
+      "loss": 0.2885,
+      "step": 4024
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.609963314335365e-07,
+      "loss": 0.3799,
+      "step": 4025
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.586138523086671e-07,
+      "loss": 0.2785,
+      "step": 4026
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 0.3323,
+      "step": 4027
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.538636686231902e-07,
+      "loss": 0.3272,
+      "step": 4028
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.51495966540182e-07,
+      "loss": 0.3451,
+      "step": 4029
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.491331926111166e-07,
+      "loss": 0.3201,
+      "step": 4030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.467753480683658e-07,
+      "loss": 0.2877,
+      "step": 4031
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.444224341417392e-07,
+      "loss": 0.306,
+      "step": 4032
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.420744520584664e-07,
+      "loss": 0.3064,
+      "step": 4033
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.397314030432133e-07,
+      "loss": 0.2742,
+      "step": 4034
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.373932883180655e-07,
+      "loss": 0.3411,
+      "step": 4035
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.350601091025365e-07,
+      "loss": 0.305,
+      "step": 4036
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.327318666135672e-07,
+      "loss": 0.3116,
+      "step": 4037
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.304085620655263e-07,
+      "loss": 0.3806,
+      "step": 4038
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.280901966702012e-07,
+      "loss": 0.3338,
+      "step": 4039
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.257767716368045e-07,
+      "loss": 0.3913,
+      "step": 4040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 0.3071,
+      "step": 4041
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.211647474797732e-07,
+      "loss": 0.3315,
+      "step": 4042
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.188661507616799e-07,
+      "loss": 0.3716,
+      "step": 4043
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.165724992165955e-07,
+      "loss": 0.2761,
+      "step": 4044
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.142837940408474e-07,
+      "loss": 0.3258,
+      "step": 4045
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.120000364281763e-07,
+      "loss": 0.3388,
+      "step": 4046
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.097212275697461e-07,
+      "loss": 0.2973,
+      "step": 4047
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.074473686541381e-07,
+      "loss": 0.3221,
+      "step": 4048
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.051784608673549e-07,
+      "loss": 0.3346,
+      "step": 4049
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.029145053928131e-07,
+      "loss": 0.3958,
+      "step": 4050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.006555034113459e-07,
+      "loss": 0.3131,
+      "step": 4051
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.984014561012051e-07,
+      "loss": 0.3285,
+      "step": 4052
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.961523646380562e-07,
+      "loss": 0.4182,
+      "step": 4053
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.939082301949849e-07,
+      "loss": 0.3045,
+      "step": 4054
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916690539424828e-07,
+      "loss": 0.3166,
+      "step": 4055
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.3116,
+      "step": 4056
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.872055806782517e-07,
+      "loss": 0.2773,
+      "step": 4057
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.849812859945791e-07,
+      "loss": 0.2815,
+      "step": 4058
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.827619541575967e-07,
+      "loss": 0.3304,
+      "step": 4059
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.805475863248665e-07,
+      "loss": 0.3646,
+      "step": 4060
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.783381836513557e-07,
+      "loss": 0.3417,
+      "step": 4061
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7613374728944784e-07,
+      "loss": 0.3676,
+      "step": 4062
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7393427838893e-07,
+      "loss": 0.4519,
+      "step": 4063
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7173977809700455e-07,
+      "loss": 0.3346,
+      "step": 4064
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.695502475582814e-07,
+      "loss": 0.4199,
+      "step": 4065
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6736568791477367e-07,
+      "loss": 0.3787,
+      "step": 4066
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.651861003059055e-07,
+      "loss": 0.3017,
+      "step": 4067
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.630114858685075e-07,
+      "loss": 0.3275,
+      "step": 4068
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6084184573681465e-07,
+      "loss": 0.2565,
+      "step": 4069
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.586771810424706e-07,
+      "loss": 0.3588,
+      "step": 4070
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5651749291451884e-07,
+      "loss": 0.3631,
+      "step": 4071
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.543627824794139e-07,
+      "loss": 0.373,
+      "step": 4072
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.52213050861009e-07,
+      "loss": 0.3495,
+      "step": 4073
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.500682991805594e-07,
+      "loss": 0.2922,
+      "step": 4074
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.479285285567281e-07,
+      "loss": 0.3081,
+      "step": 4075
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.457937401055779e-07,
+      "loss": 0.3487,
+      "step": 4076
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4366393494057135e-07,
+      "loss": 0.3344,
+      "step": 4077
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.415391141725711e-07,
+      "loss": 0.3392,
+      "step": 4078
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.39419278909845e-07,
+      "loss": 0.3258,
+      "step": 4079
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3730443025805537e-07,
+      "loss": 0.3423,
+      "step": 4080
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.351945693202686e-07,
+      "loss": 0.3592,
+      "step": 4081
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.330896971969423e-07,
+      "loss": 0.3088,
+      "step": 4082
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.309898149859393e-07,
+      "loss": 0.2884,
+      "step": 4083
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2889492378251687e-07,
+      "loss": 0.3069,
+      "step": 4084
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 0.3344,
+      "step": 4085
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2472011876642184e-07,
+      "loss": 0.346,
+      "step": 4086
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2264020713124854e-07,
+      "loss": 0.313,
+      "step": 4087
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.205652908586455e-07,
+      "loss": 0.3483,
+      "step": 4088
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1849537103084924e-07,
+      "loss": 0.3193,
+      "step": 4089
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.164304487274884e-07,
+      "loss": 0.3478,
+      "step": 4090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1437052502558693e-07,
+      "loss": 0.3052,
+      "step": 4091
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1231560099956103e-07,
+      "loss": 0.3066,
+      "step": 4092
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.102656777212166e-07,
+      "loss": 0.3159,
+      "step": 4093
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0822075625975623e-07,
+      "loss": 0.3224,
+      "step": 4094
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0618083768176996e-07,
+      "loss": 0.3313,
+      "step": 4095
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0414592305123566e-07,
+      "loss": 0.3286,
+      "step": 4096
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0211601342952876e-07,
+      "loss": 0.33,
+      "step": 4097
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.00091109875409e-07,
+      "loss": 0.3239,
+      "step": 4098
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9807121344502706e-07,
+      "loss": 0.3908,
+      "step": 4099
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9605632519192027e-07,
+      "loss": 0.3503,
+      "step": 4100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.940464461670135e-07,
+      "loss": 0.3308,
+      "step": 4101
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.920415774186215e-07,
+      "loss": 0.2845,
+      "step": 4102
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9004171999244553e-07,
+      "loss": 0.3816,
+      "step": 4103
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8804687493157136e-07,
+      "loss": 0.3076,
+      "step": 4104
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8605704327646877e-07,
+      "loss": 0.3216,
+      "step": 4105
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.840722260649976e-07,
+      "loss": 0.3857,
+      "step": 4106
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.8209242433239755e-07,
+      "loss": 0.3369,
+      "step": 4107
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.8011763911129595e-07,
+      "loss": 0.285,
+      "step": 4108
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7814787143169885e-07,
+      "loss": 0.311,
+      "step": 4109
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7618312232100327e-07,
+      "loss": 0.276,
+      "step": 4110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.742233928039796e-07,
+      "loss": 0.2917,
+      "step": 4111
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.722686839027856e-07,
+      "loss": 0.3444,
+      "step": 4112
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7031899663695804e-07,
+      "loss": 0.3425,
+      "step": 4113
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.68374332023419e-07,
+      "loss": 0.3218,
+      "step": 4114
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.664346910764638e-07,
+      "loss": 0.2743,
+      "step": 4115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.645000748077709e-07,
+      "loss": 0.2839,
+      "step": 4116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6257048422639995e-07,
+      "loss": 0.3732,
+      "step": 4117
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6064592033878577e-07,
+      "loss": 0.3294,
+      "step": 4118
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.587263841487454e-07,
+      "loss": 0.3409,
+      "step": 4119
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.56811876657468e-07,
+      "loss": 0.3385,
+      "step": 4120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5490239886352584e-07,
+      "loss": 0.3789,
+      "step": 4121
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5299795176286453e-07,
+      "loss": 0.308,
+      "step": 4122
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.510985363488051e-07,
+      "loss": 0.3262,
+      "step": 4123
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4920415361204494e-07,
+      "loss": 0.3338,
+      "step": 4124
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4731480454065823e-07,
+      "loss": 0.2953,
+      "step": 4125
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4543049012009336e-07,
+      "loss": 0.3427,
+      "step": 4126
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4355121133316847e-07,
+      "loss": 0.3515,
+      "step": 4127
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4167696916008075e-07,
+      "loss": 0.3005,
+      "step": 4128
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3980776457839593e-07,
+      "loss": 0.3047,
+      "step": 4129
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3794359856305747e-07,
+      "loss": 0.3105,
+      "step": 4130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.360844720863765e-07,
+      "loss": 0.2676,
+      "step": 4131
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3423038611803406e-07,
+      "loss": 0.2954,
+      "step": 4132
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3238134162508984e-07,
+      "loss": 0.3343,
+      "step": 4133
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3053733957196464e-07,
+      "loss": 0.3455,
+      "step": 4134
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.286983809204558e-07,
+      "loss": 0.3811,
+      "step": 4135
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2686446662972824e-07,
+      "loss": 0.3024,
+      "step": 4136
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2503559765631577e-07,
+      "loss": 0.2961,
+      "step": 4137
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.232117749541208e-07,
+      "loss": 0.4101,
+      "step": 4138
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2139299947441026e-07,
+      "loss": 0.3242,
+      "step": 4139
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1957927216582307e-07,
+      "loss": 0.3053,
+      "step": 4140
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1777059397436693e-07,
+      "loss": 0.2743,
+      "step": 4141
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.159669658434106e-07,
+      "loss": 0.3911,
+      "step": 4142
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 0.3147,
+      "step": 4143
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.123748635233082e-07,
+      "loss": 0.4112,
+      "step": 4144
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1058639120773225e-07,
+      "loss": 0.3541,
+      "step": 4145
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0880297269979654e-07,
+      "loss": 0.2931,
+      "step": 4146
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0702460892969356e-07,
+      "loss": 0.2996,
+      "step": 4147
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0525130082498575e-07,
+      "loss": 0.3319,
+      "step": 4148
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0348304931059556e-07,
+      "loss": 0.3121,
+      "step": 4149
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0171985530880633e-07,
+      "loss": 0.3454,
+      "step": 4150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9996171973926567e-07,
+      "loss": 0.3697,
+      "step": 4151
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.982086435189857e-07,
+      "loss": 0.4037,
+      "step": 4152
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9646062756233273e-07,
+      "loss": 0.3243,
+      "step": 4153
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9471767278103855e-07,
+      "loss": 0.4074,
+      "step": 4154
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9297978008419604e-07,
+      "loss": 0.297,
+      "step": 4155
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.912469503782522e-07,
+      "loss": 0.3053,
+      "step": 4156
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8951918456702087e-07,
+      "loss": 0.2948,
+      "step": 4157
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8779648355166776e-07,
+      "loss": 0.3947,
+      "step": 4158
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8607884823072196e-07,
+      "loss": 0.294,
+      "step": 4159
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.843662795000668e-07,
+      "loss": 0.3124,
+      "step": 4160
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.826587782529444e-07,
+      "loss": 0.3534,
+      "step": 4161
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8095634537995465e-07,
+      "loss": 0.2595,
+      "step": 4162
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7925898176905275e-07,
+      "loss": 0.3367,
+      "step": 4163
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7756668830555164e-07,
+      "loss": 0.3472,
+      "step": 4164
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.758794658721164e-07,
+      "loss": 0.3874,
+      "step": 4165
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.741973153487687e-07,
+      "loss": 0.2704,
+      "step": 4166
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.725202376128855e-07,
+      "loss": 0.2777,
+      "step": 4167
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.708482335391993e-07,
+      "loss": 0.3249,
+      "step": 4168
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.691813039997937e-07,
+      "loss": 0.2865,
+      "step": 4169
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.675194498641054e-07,
+      "loss": 0.3619,
+      "step": 4170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.658626719989266e-07,
+      "loss": 0.3301,
+      "step": 4171
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6421097126839714e-07,
+      "loss": 0.3165,
+      "step": 4172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6256434853401455e-07,
+      "loss": 0.3632,
+      "step": 4173
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6092280465462416e-07,
+      "loss": 0.3449,
+      "step": 4174
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5928634048642323e-07,
+      "loss": 0.3234,
+      "step": 4175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5765495688295785e-07,
+      "loss": 0.3142,
+      "step": 4176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5602865469512626e-07,
+      "loss": 0.3416,
+      "step": 4177
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5440743477117556e-07,
+      "loss": 0.3508,
+      "step": 4178
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.527912979567038e-07,
+      "loss": 0.3487,
+      "step": 4179
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5118024509465434e-07,
+      "loss": 0.3208,
+      "step": 4180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4957427702532177e-07,
+      "loss": 0.3118,
+      "step": 4181
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4797339458634717e-07,
+      "loss": 0.3149,
+      "step": 4182
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4637759861271704e-07,
+      "loss": 0.3332,
+      "step": 4183
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.447868899367722e-07,
+      "loss": 0.3598,
+      "step": 4184
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4320126938819023e-07,
+      "loss": 0.3148,
+      "step": 4185
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4162073779400274e-07,
+      "loss": 0.368,
+      "step": 4186
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.400452959785815e-07,
+      "loss": 0.3443,
+      "step": 4187
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.384749447636481e-07,
+      "loss": 0.2827,
+      "step": 4188
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3690968496826505e-07,
+      "loss": 0.3065,
+      "step": 4189
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.353495174088427e-07,
+      "loss": 0.3286,
+      "step": 4190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3379444289913344e-07,
+      "loss": 0.3182,
+      "step": 4191
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.322444622502329e-07,
+      "loss": 0.25,
+      "step": 4192
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3069957627058104e-07,
+      "loss": 0.3923,
+      "step": 4193
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2915978576595888e-07,
+      "loss": 0.4447,
+      "step": 4194
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2762509153949285e-07,
+      "loss": 0.3257,
+      "step": 4195
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2609549439164713e-07,
+      "loss": 0.361,
+      "step": 4196
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2457099512023129e-07,
+      "loss": 0.3051,
+      "step": 4197
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.230515945203937e-07,
+      "loss": 0.3311,
+      "step": 4198
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2153729338462272e-07,
+      "loss": 0.3098,
+      "step": 4199
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2002809250274982e-07,
+      "loss": 0.318,
+      "step": 4200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.3056,
+      "step": 4201
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.170249946467129e-07,
+      "loss": 0.3491,
+      "step": 4202
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1553109923890703e-07,
+      "loss": 0.3286,
+      "step": 4203
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1404230721770891e-07,
+      "loss": 0.3316,
+      "step": 4204
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1255861935964628e-07,
+      "loss": 0.3972,
+      "step": 4205
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1108003643858123e-07,
+      "loss": 0.2529,
+      "step": 4206
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0960655922571238e-07,
+      "loss": 0.3215,
+      "step": 4207
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0813818848957724e-07,
+      "loss": 0.3036,
+      "step": 4208
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.066749249960498e-07,
+      "loss": 0.2565,
+      "step": 4209
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0521676950833736e-07,
+      "loss": 0.3292,
+      "step": 4210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.037637227869871e-07,
+      "loss": 0.2921,
+      "step": 4211
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0231578558987942e-07,
+      "loss": 0.2492,
+      "step": 4212
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0087295867222912e-07,
+      "loss": 0.3024,
+      "step": 4213
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9943524278658755e-07,
+      "loss": 0.2949,
+      "step": 4214
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9800263868283708e-07,
+      "loss": 0.2967,
+      "step": 4215
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.965751471081978e-07,
+      "loss": 0.3995,
+      "step": 4216
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.951527688072208e-07,
+      "loss": 0.2858,
+      "step": 4217
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9373550452178925e-07,
+      "loss": 0.3052,
+      "step": 4218
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9232335499112186e-07,
+      "loss": 0.2677,
+      "step": 4219
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9091632095176837e-07,
+      "loss": 0.3228,
+      "step": 4220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8951440313760838e-07,
+      "loss": 0.2892,
+      "step": 4221
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8811760227985588e-07,
+      "loss": 0.3362,
+      "step": 4222
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8672591910705363e-07,
+      "loss": 0.2917,
+      "step": 4223
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8533935434507876e-07,
+      "loss": 0.343,
+      "step": 4224
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8395790871713393e-07,
+      "loss": 0.338,
+      "step": 4225
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8258158294375385e-07,
+      "loss": 0.3888,
+      "step": 4226
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8121037774280314e-07,
+      "loss": 0.2593,
+      "step": 4227
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7984429382947643e-07,
+      "loss": 0.3226,
+      "step": 4228
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.784833319162971e-07,
+      "loss": 0.3465,
+      "step": 4229
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7712749271311392e-07,
+      "loss": 0.2553,
+      "step": 4230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7577677692710682e-07,
+      "loss": 0.3325,
+      "step": 4231
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7443118526278225e-07,
+      "loss": 0.338,
+      "step": 4232
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7309071842197544e-07,
+      "loss": 0.3295,
+      "step": 4233
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7175537710384604e-07,
+      "loss": 0.3344,
+      "step": 4234
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7042516200488357e-07,
+      "loss": 0.329,
+      "step": 4235
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.691000738189008e-07,
+      "loss": 0.3179,
+      "step": 4236
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6778011323703825e-07,
+      "loss": 0.4275,
+      "step": 4237
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6646528094776071e-07,
+      "loss": 0.3241,
+      "step": 4238
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6515557763685964e-07,
+      "loss": 0.3491,
+      "step": 4239
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.638510039874508e-07,
+      "loss": 0.3268,
+      "step": 4240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6255156067997325e-07,
+      "loss": 0.3094,
+      "step": 4241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6125724839219036e-07,
+      "loss": 0.3506,
+      "step": 4242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5996806779919215e-07,
+      "loss": 0.3117,
+      "step": 4243
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5868401957338853e-07,
+      "loss": 0.3179,
+      "step": 4244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5740510438451374e-07,
+      "loss": 0.3039,
+      "step": 4245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.561313228996253e-07,
+      "loss": 0.3218,
+      "step": 4246
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5486267578310177e-07,
+      "loss": 0.3059,
+      "step": 4247
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5359916369664607e-07,
+      "loss": 0.3574,
+      "step": 4248
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5234078729927992e-07,
+      "loss": 0.3676,
+      "step": 4249
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5108754724734832e-07,
+      "loss": 0.3834,
+      "step": 4250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4983944419451613e-07,
+      "loss": 0.2845,
+      "step": 4251
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.485964787917704e-07,
+      "loss": 0.2863,
+      "step": 4252
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.473586516874159e-07,
+      "loss": 0.2932,
+      "step": 4253
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4612596352708064e-07,
+      "loss": 0.3861,
+      "step": 4254
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4489841495370804e-07,
+      "loss": 0.3524,
+      "step": 4255
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4367600660756598e-07,
+      "loss": 0.3129,
+      "step": 4256
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4245873912623777e-07,
+      "loss": 0.3378,
+      "step": 4257
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4124661314462552e-07,
+      "loss": 0.3624,
+      "step": 4258
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 0.3379,
+      "step": 4259
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3883778820675266e-07,
+      "loss": 0.3034,
+      "step": 4260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.37641090506887e-07,
+      "loss": 0.3352,
+      "step": 4261
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3644953681952954e-07,
+      "loss": 0.3168,
+      "step": 4262
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.352631277661709e-07,
+      "loss": 0.3847,
+      "step": 4263
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3408186396561716e-07,
+      "loss": 0.2992,
+      "step": 4264
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.329057460339933e-07,
+      "loss": 0.3281,
+      "step": 4265
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.317347745847386e-07,
+      "loss": 0.3402,
+      "step": 4266
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3056895022861005e-07,
+      "loss": 0.3547,
+      "step": 4267
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2940827357367792e-07,
+      "loss": 0.3329,
+      "step": 4268
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2825274522532795e-07,
+      "loss": 0.285,
+      "step": 4269
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2710236578626246e-07,
+      "loss": 0.3239,
+      "step": 4270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2595713585649482e-07,
+      "loss": 0.3194,
+      "step": 4271
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2481705603335504e-07,
+      "loss": 0.2644,
+      "step": 4272
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2368212691148852e-07,
+      "loss": 0.3616,
+      "step": 4273
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2255234908284953e-07,
+      "loss": 0.3538,
+      "step": 4274
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.214277231367078e-07,
+      "loss": 0.3453,
+      "step": 4275
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.203082496596475e-07,
+      "loss": 0.3825,
+      "step": 4276
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1919392923556261e-07,
+      "loss": 0.346,
+      "step": 4277
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1808476244566269e-07,
+      "loss": 0.2866,
+      "step": 4278
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1698074986846608e-07,
+      "loss": 0.3038,
+      "step": 4279
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1588189207980327e-07,
+      "loss": 0.3915,
+      "step": 4280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1478818965281912e-07,
+      "loss": 0.3166,
+      "step": 4281
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.13699643157964e-07,
+      "loss": 0.2861,
+      "step": 4282
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1261625316300596e-07,
+      "loss": 0.299,
+      "step": 4283
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1153802023301741e-07,
+      "loss": 0.3576,
+      "step": 4284
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1046494493038517e-07,
+      "loss": 0.3667,
+      "step": 4285
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0939702781480266e-07,
+      "loss": 0.399,
+      "step": 4286
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.083342694432754e-07,
+      "loss": 0.34,
+      "step": 4287
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0727667037011668e-07,
+      "loss": 0.3235,
+      "step": 4288
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0622423114694968e-07,
+      "loss": 0.2471,
+      "step": 4289
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0517695232270752e-07,
+      "loss": 0.3283,
+      "step": 4290
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.041348344436277e-07,
+      "loss": 0.2842,
+      "step": 4291
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0309787805325877e-07,
+      "loss": 0.2655,
+      "step": 4292
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0206608369245807e-07,
+      "loss": 0.3982,
+      "step": 4293
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0103945189938847e-07,
+      "loss": 0.3451,
+      "step": 4294
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0001798320952161e-07,
+      "loss": 0.3396,
+      "step": 4295
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.900167815563467e-08,
+      "loss": 0.3732,
+      "step": 4296
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.79905372678136e-08,
+      "loss": 0.2568,
+      "step": 4297
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.698456107344767e-08,
+      "loss": 0.2779,
+      "step": 4298
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.598375009723603e-08,
+      "loss": 0.3068,
+      "step": 4299
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.498810486118115e-08,
+      "loss": 0.392,
+      "step": 4300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.399762588459205e-08,
+      "loss": 0.3009,
+      "step": 4301
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.301231368408326e-08,
+      "loss": 0.3234,
+      "step": 4302
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.20321687735748e-08,
+      "loss": 0.3187,
+      "step": 4303
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.105719166428883e-08,
+      "loss": 0.3439,
+      "step": 4304
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.008738286475748e-08,
+      "loss": 0.3706,
+      "step": 4305
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.912274288081168e-08,
+      "loss": 0.3403,
+      "step": 4306
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.816327221558896e-08,
+      "loss": 0.2932,
+      "step": 4307
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.720897136953121e-08,
+      "loss": 0.2664,
+      "step": 4308
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.625984084038253e-08,
+      "loss": 0.3076,
+      "step": 4309
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.531588112319022e-08,
+      "loss": 0.3213,
+      "step": 4310
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.437709271030603e-08,
+      "loss": 0.3076,
+      "step": 4311
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.34434760913827e-08,
+      "loss": 0.3476,
+      "step": 4312
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.251503175337739e-08,
+      "loss": 0.3392,
+      "step": 4313
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.159176018054715e-08,
+      "loss": 0.4168,
+      "step": 4314
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.067366185445236e-08,
+      "loss": 0.3472,
+      "step": 4315
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.97607372539566e-08,
+      "loss": 0.332,
+      "step": 4316
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.3013,
+      "step": 4317
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.795041113171308e-08,
+      "loss": 0.3516,
+      "step": 4318
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.705301055419667e-08,
+      "loss": 0.2779,
+      "step": 4319
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.61607855907387e-08,
+      "loss": 0.3294,
+      "step": 4320
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.52737367067069e-08,
+      "loss": 0.3809,
+      "step": 4321
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.439186436476675e-08,
+      "loss": 0.3784,
+      "step": 4322
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.351516902488698e-08,
+      "loss": 0.3769,
+      "step": 4323
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.264365114433402e-08,
+      "loss": 0.3723,
+      "step": 4324
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.177731117767539e-08,
+      "loss": 0.3799,
+      "step": 4325
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.091614957677517e-08,
+      "loss": 0.3056,
+      "step": 4326
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.006016679080075e-08,
+      "loss": 0.3233,
+      "step": 4327
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.920936326621497e-08,
+      "loss": 0.372,
+      "step": 4328
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.836373944677954e-08,
+      "loss": 0.3043,
+      "step": 4329
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.752329577355498e-08,
+      "loss": 0.2963,
+      "step": 4330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.668803268490066e-08,
+      "loss": 0.3917,
+      "step": 4331
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.58579506164736e-08,
+      "loss": 0.3862,
+      "step": 4332
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.503305000122861e-08,
+      "loss": 0.2551,
+      "step": 4333
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.421333126941708e-08,
+      "loss": 0.378,
+      "step": 4334
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.339879484858924e-08,
+      "loss": 0.3322,
+      "step": 4335
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.258944116358856e-08,
+      "loss": 0.3282,
+      "step": 4336
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.17852706365607e-08,
+      "loss": 0.337,
+      "step": 4337
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.098628368694459e-08,
+      "loss": 0.269,
+      "step": 4338
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.019248073147577e-08,
+      "loss": 0.3413,
+      "step": 4339
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9403862184186364e-08,
+      "loss": 0.3215,
+      "step": 4340
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 0.3238,
+      "step": 4341
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7842179956753006e-08,
+      "loss": 0.3056,
+      "step": 4342
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.706911709115193e-08,
+      "loss": 0.3668,
+      "step": 4343
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6301240262814914e-08,
+      "loss": 0.375,
+      "step": 4344
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.5538549872251603e-08,
+      "loss": 0.3142,
+      "step": 4345
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.3245,
+      "step": 4346
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.4028729992958715e-08,
+      "loss": 0.2864,
+      "step": 4347
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.328160129172255e-08,
+      "loss": 0.3262,
+      "step": 4348
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.253966060324356e-08,
+      "loss": 0.3099,
+      "step": 4349
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1802908314504407e-08,
+      "loss": 0.3461,
+      "step": 4350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1071344809781045e-08,
+      "loss": 0.3474,
+      "step": 4351
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.034497047064158e-08,
+      "loss": 0.3008,
+      "step": 4352
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.962378567594961e-08,
+      "loss": 0.3242,
+      "step": 4353
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.890779080186092e-08,
+      "loss": 0.3185,
+      "step": 4354
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.819698622182456e-08,
+      "loss": 0.3403,
+      "step": 4355
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.749137230658063e-08,
+      "loss": 0.3591,
+      "step": 4356
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.679094942416473e-08,
+      "loss": 0.2643,
+      "step": 4357
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.609571793990242e-08,
+      "loss": 0.306,
+      "step": 4358
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.540567821641584e-08,
+      "loss": 0.3667,
+      "step": 4359
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.4720830613612656e-08,
+      "loss": 0.3,
+      "step": 4360
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.404117548869824e-08,
+      "loss": 0.3698,
+      "step": 4361
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.33667131961657e-08,
+      "loss": 0.3029,
+      "step": 4362
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.269744408780252e-08,
+      "loss": 0.3762,
+      "step": 4363
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.203336851268502e-08,
+      "loss": 0.3687,
+      "step": 4364
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.137448681718392e-08,
+      "loss": 0.2886,
+      "step": 4365
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.072079934495765e-08,
+      "loss": 0.3211,
+      "step": 4366
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.007230643695681e-08,
+      "loss": 0.3177,
+      "step": 4367
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9429008431423054e-08,
+      "loss": 0.315,
+      "step": 4368
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.879090566388688e-08,
+      "loss": 0.3597,
+      "step": 4369
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.815799846717094e-08,
+      "loss": 0.3278,
+      "step": 4370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.753028717138785e-08,
+      "loss": 0.3492,
+      "step": 4371
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.690777210393792e-08,
+      "loss": 0.3624,
+      "step": 4372
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.629045358951366e-08,
+      "loss": 0.3128,
+      "step": 4373
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.56783319500964e-08,
+      "loss": 0.3555,
+      "step": 4374
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 0.3401,
+      "step": 4375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.446968057065459e-08,
+      "loss": 0.25,
+      "step": 4376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3873151461037975e-08,
+      "loss": 0.2841,
+      "step": 4377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3281820487245375e-08,
+      "loss": 0.3881,
+      "step": 4378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2695687957703393e-08,
+      "loss": 0.3441,
+      "step": 4379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.211475417812748e-08,
+      "loss": 0.2992,
+      "step": 4380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1539019451521935e-08,
+      "loss": 0.3661,
+      "step": 4381
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.096848407817765e-08,
+      "loss": 0.3166,
+      "step": 4382
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.040314835567326e-08,
+      "loss": 0.373,
+      "step": 4383
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.984301257887956e-08,
+      "loss": 0.2995,
+      "step": 4384
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9288077039951734e-08,
+      "loss": 0.396,
+      "step": 4385
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.873834202833159e-08,
+      "loss": 0.3756,
+      "step": 4386
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8193807830750875e-08,
+      "loss": 0.2928,
+      "step": 4387
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7654474731229064e-08,
+      "loss": 0.3119,
+      "step": 4388
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7120343011071138e-08,
+      "loss": 0.3126,
+      "step": 4389
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.659141294886869e-08,
+      "loss": 0.3456,
+      "step": 4390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.606768482050215e-08,
+      "loss": 0.3526,
+      "step": 4391
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5549158899137448e-08,
+      "loss": 0.3757,
+      "step": 4392
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.503583545522714e-08,
+      "loss": 0.3492,
+      "step": 4393
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4527714756512612e-08,
+      "loss": 0.3083,
+      "step": 4394
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4024797068017414e-08,
+      "loss": 0.3744,
+      "step": 4395
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3527082652053946e-08,
+      "loss": 0.3627,
+      "step": 4396
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3034571768221213e-08,
+      "loss": 0.4325,
+      "step": 4397
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2547264673401514e-08,
+      "loss": 0.3192,
+      "step": 4398
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2065161621767084e-08,
+      "loss": 0.2994,
+      "step": 4399
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1588262864771225e-08,
+      "loss": 0.3079,
+      "step": 4400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1116568651156076e-08,
+      "loss": 0.3335,
+      "step": 4401
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0650079226948173e-08,
+      "loss": 0.4658,
+      "step": 4402
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0188794835458437e-08,
+      "loss": 0.2734,
+      "step": 4403
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.973271571728441e-08,
+      "loss": 0.3796,
+      "step": 4404
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9281842110308035e-08,
+      "loss": 0.2684,
+      "step": 4405
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8836174249694526e-08,
+      "loss": 0.3318,
+      "step": 4406
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8395712367897943e-08,
+      "loss": 0.3438,
+      "step": 4407
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7960456694653404e-08,
+      "loss": 0.353,
+      "step": 4408
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7530407456980425e-08,
+      "loss": 0.3374,
+      "step": 4409
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7105564879186243e-08,
+      "loss": 0.2922,
+      "step": 4410
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6685929182860274e-08,
+      "loss": 0.3373,
+      "step": 4411
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.62715005868741e-08,
+      "loss": 0.3096,
+      "step": 4412
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.586227930738815e-08,
+      "loss": 0.2676,
+      "step": 4413
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.54582655578428e-08,
+      "loss": 0.3185,
+      "step": 4414
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.505945954896393e-08,
+      "loss": 0.3252,
+      "step": 4415
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4665861488761813e-08,
+      "loss": 0.2891,
+      "step": 4416
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.427747158252779e-08,
+      "loss": 0.2943,
+      "step": 4417
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3894290032838708e-08,
+      "loss": 0.3554,
+      "step": 4418
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3516317039555804e-08,
+      "loss": 0.3414,
+      "step": 4419
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.314355279982138e-08,
+      "loss": 0.3638,
+      "step": 4420
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2775997508061023e-08,
+      "loss": 0.4364,
+      "step": 4421
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2413651355986933e-08,
+      "loss": 0.325,
+      "step": 4422
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2056514532590158e-08,
+      "loss": 0.2983,
+      "step": 4423
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1704587224145025e-08,
+      "loss": 0.2757,
+      "step": 4424
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.135786961421248e-08,
+      "loss": 0.329,
+      "step": 4425
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.101636188363231e-08,
+      "loss": 0.3697,
+      "step": 4426
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0680064210527585e-08,
+      "loss": 0.3818,
+      "step": 4427
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0348976770305774e-08,
+      "loss": 0.3338,
+      "step": 4428
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0023099735654295e-08,
+      "loss": 0.3113,
+      "step": 4429
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.702433276546075e-09,
+      "loss": 0.3187,
+      "step": 4430
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.38697756023288e-09,
+      "loss": 0.3704,
+      "step": 4431
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.076732751249762e-09,
+      "loss": 0.278,
+      "step": 4432
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 0.3727,
+      "step": 4433
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.47187649983039e-09,
+      "loss": 0.2887,
+      "step": 4434
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.17726537287511e-09,
+      "loss": 0.3263,
+      "step": 4435
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.887865784211856e-09,
+      "loss": 0.3351,
+      "step": 4436
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.603677884787663e-09,
+      "loss": 0.2761,
+      "step": 4437
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.324701822827296e-09,
+      "loss": 0.3353,
+      "step": 4438
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.050937743841024e-09,
+      "loss": 0.3624,
+      "step": 4439
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.7823857906179625e-09,
+      "loss": 0.3885,
+      "step": 4440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.5190461032305085e-09,
+      "loss": 0.292,
+      "step": 4441
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.260918819029904e-09,
+      "loss": 0.351,
+      "step": 4442
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.008004072650675e-09,
+      "loss": 0.3506,
+      "step": 4443
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.7603019960095205e-09,
+      "loss": 0.3039,
+      "step": 4444
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.517812718301985e-09,
+      "loss": 0.3735,
+      "step": 4445
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.280536366004674e-09,
+      "loss": 0.304,
+      "step": 4446
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.048473062878589e-09,
+      "loss": 0.3088,
+      "step": 4447
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.821622929961356e-09,
+      "loss": 0.4012,
+      "step": 4448
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.599986085573882e-09,
+      "loss": 0.3194,
+      "step": 4449
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3835626453181405e-09,
+      "loss": 0.3768,
+      "step": 4450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.172352722077167e-09,
+      "loss": 0.3486,
+      "step": 4451
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.96635642601173e-09,
+      "loss": 0.2671,
+      "step": 4452
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.765573864568106e-09,
+      "loss": 0.3593,
+      "step": 4453
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5700051424680803e-09,
+      "loss": 0.3076,
+      "step": 4454
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3796503617167244e-09,
+      "loss": 0.2583,
+      "step": 4455
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1945096216012826e-09,
+      "loss": 0.3779,
+      "step": 4456
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.014583018686734e-09,
+      "loss": 0.3196,
+      "step": 4457
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8398706468180105e-09,
+      "loss": 0.3909,
+      "step": 4458
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6703725971222173e-09,
+      "loss": 0.3335,
+      "step": 4459
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5060889580075245e-09,
+      "loss": 0.3591,
+      "step": 4460
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.347019815158724e-09,
+      "loss": 0.3125,
+      "step": 4461
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.193165251545004e-09,
+      "loss": 0.3285,
+      "step": 4462
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.044525347413284e-09,
+      "loss": 0.3268,
+      "step": 4463
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9011001802915484e-09,
+      "loss": 0.3003,
+      "step": 4464
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7628898249866243e-09,
+      "loss": 0.3448,
+      "step": 4465
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6298943535875133e-09,
+      "loss": 0.2768,
+      "step": 4466
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5021138354609498e-09,
+      "loss": 0.3538,
+      "step": 4467
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3795483372558427e-09,
+      "loss": 0.3453,
+      "step": 4468
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2621979228988334e-09,
+      "loss": 0.3143,
+      "step": 4469
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.150062653598738e-09,
+      "loss": 0.2817,
+      "step": 4470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0431425878421053e-09,
+      "loss": 0.3834,
+      "step": 4471
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.41437781396548e-10,
+      "loss": 0.3828,
+      "step": 4472
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.449482873096326e-10,
+      "loss": 0.3188,
+      "step": 4473
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.536741559077687e-10,
+      "loss": 0.3319,
+      "step": 4474
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.676154347995401e-10,
+      "loss": 0.3161,
+      "step": 4475
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.867721688690431e-10,
+      "loss": 0.3819,
+      "step": 4476
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.111444002836585e-10,
+      "loss": 0.3328,
+      "step": 4477
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.407321684885002e-10,
+      "loss": 0.2699,
+      "step": 4478
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7553551021085597e-10,
+      "loss": 0.3805,
+      "step": 4479
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1555445945574693e-10,
+      "loss": 0.2577,
+      "step": 4480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.607890475070374e-10,
+      "loss": 0.3435,
+      "step": 4481
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1123930292965556e-10,
+      "loss": 0.3339,
+      "step": 4482
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6690525156737303e-10,
+      "loss": 0.3501,
+      "step": 4483
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.277869165461354e-10,
+      "loss": 0.344,
+      "step": 4484
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.388431826629074e-11,
+      "loss": 0.3729,
+      "step": 4485
+    },
+    {
+      "epoch": 3.0,
+      "step": 4485,
+      "total_flos": 4797273293914112.0,
+      "train_loss": 0.4469948373244891,
+      "train_runtime": 160134.8088,
+      "train_samples_per_second": 3.586,
+      "train_steps_per_second": 0.028
+    }
+  ],
+  "max_steps": 4485,
+  "num_train_epochs": 3,
+  "total_flos": 4797273293914112.0,
+  "trial_name": null,
+  "trial_params": null
+}