diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21450 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 3570,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8518518518518518e-07,
+      "loss": 2.278,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7037037037037036e-07,
+      "loss": 2.1144,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 2.1585,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 2.4237,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 2.2728,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 2.3594,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2962962962962962e-06,
+      "loss": 2.1611,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 2.1716,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.7164,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 1.6317,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.037037037037037e-06,
+      "loss": 1.3521,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.3667,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4074074074074075e-06,
+      "loss": 1.3928,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 1.2801,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 1.05,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 1.0592,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1481481481481483e-06,
+      "loss": 0.9234,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.7906,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5185185185185187e-06,
+      "loss": 0.8761,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.8887,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 0.8328,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.074074074074074e-06,
+      "loss": 0.7349,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2592592592592596e-06,
+      "loss": 0.8369,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.7268,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.62962962962963e-06,
+      "loss": 0.6251,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 0.6531,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6658,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.7322,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.370370370370371e-06,
+      "loss": 0.6607,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.5925,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.740740740740741e-06,
+      "loss": 0.6824,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 0.6413,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.6223,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.296296296296297e-06,
+      "loss": 0.5512,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.481481481481482e-06,
+      "loss": 0.6371,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7426,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.851851851851853e-06,
+      "loss": 0.533,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 0.5827,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 0.6313,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.5335,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.592592592592594e-06,
+      "loss": 0.5249,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.4915,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.962962962962963e-06,
+      "loss": 0.636,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 0.6301,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.5327,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.518518518518519e-06,
+      "loss": 0.5583,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.703703703703705e-06,
+      "loss": 0.5557,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.4679,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.074074074074075e-06,
+      "loss": 0.4286,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.25925925925926e-06,
+      "loss": 0.5706,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 0.6021,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.5559,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.814814814814815e-06,
+      "loss": 0.5264,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5647,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0185185185185186e-05,
+      "loss": 0.5643,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.4496,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 0.4617,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0740740740740742e-05,
+      "loss": 0.533,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0925925925925926e-05,
+      "loss": 0.6141,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.4338,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1296296296296297e-05,
+      "loss": 0.5985,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1481481481481482e-05,
+      "loss": 0.5873,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.4725,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.436,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2037037037037039e-05,
+      "loss": 0.5032,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 0.5653,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2407407407407408e-05,
+      "loss": 0.5298,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.4717,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 0.5635,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2962962962962964e-05,
+      "loss": 0.5043,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.314814814814815e-05,
+      "loss": 0.4106,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5027,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.351851851851852e-05,
+      "loss": 0.4894,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3703703703703706e-05,
+      "loss": 0.5795,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.4508,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.5163,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4259259259259259e-05,
+      "loss": 0.5221,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.4799,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4629629629629631e-05,
+      "loss": 0.4468,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.4464,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.496,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5185185185185187e-05,
+      "loss": 0.5856,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.537037037037037e-05,
+      "loss": 0.4404,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.5088,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5740740740740744e-05,
+      "loss": 0.507,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5925925925925926e-05,
+      "loss": 0.4214,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 0.3976,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.451,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6481481481481482e-05,
+      "loss": 0.4927,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.5385,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6851851851851853e-05,
+      "loss": 0.4591,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.5165,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 0.4811,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 0.4365,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7592592592592595e-05,
+      "loss": 0.4767,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.4375,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7962962962962965e-05,
+      "loss": 0.5424,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 0.4658,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.5508,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.4921,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8703703703703707e-05,
+      "loss": 0.5064,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.3769,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9074074074074075e-05,
+      "loss": 0.4408,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.5336,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.4633,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.962962962962963e-05,
+      "loss": 0.5015,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814814814814816e-05,
+      "loss": 0.4465,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.5144,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999995882670796e-05,
+      "loss": 0.4201,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999998353068657e-05,
+      "loss": 0.4162,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999962944057496e-05,
+      "loss": 0.4836,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999934122800527e-05,
+      "loss": 0.4027,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999897066939394e-05,
+      "loss": 0.4812,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999851776504614e-05,
+      "loss": 0.4779,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999798251533477e-05,
+      "loss": 0.5086,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999736492070065e-05,
+      "loss": 0.4572,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999966649816523e-05,
+      "loss": 0.4283,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999588269876614e-05,
+      "loss": 0.4682,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999501807268636e-05,
+      "loss": 0.4781,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999940711041249e-05,
+      "loss": 0.4655,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999304179386153e-05,
+      "loss": 0.4817,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999193014274393e-05,
+      "loss": 0.518,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999073615168753e-05,
+      "loss": 0.4466,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998945982167542e-05,
+      "loss": 0.4307,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999881011537587e-05,
+      "loss": 0.4029,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998666014905617e-05,
+      "loss": 0.4254,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998513680875445e-05,
+      "loss": 0.4538,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998353113410796e-05,
+      "loss": 0.5091,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999818431264389e-05,
+      "loss": 0.4368,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999800727871373e-05,
+      "loss": 0.4909,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997822011766098e-05,
+      "loss": 0.4414,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997628511953554e-05,
+      "loss": 0.4045,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999742677943544e-05,
+      "loss": 0.4515,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997216814377875e-05,
+      "loss": 0.3735,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996998616953757e-05,
+      "loss": 0.4531,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996772187342764e-05,
+      "loss": 0.5374,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996537525731357e-05,
+      "loss": 0.4816,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996294632312766e-05,
+      "loss": 0.4303,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999604350728701e-05,
+      "loss": 0.4696,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999578415086088e-05,
+      "loss": 0.4242,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995516563247944e-05,
+      "loss": 0.373,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995240744668555e-05,
+      "loss": 0.4957,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994956695349845e-05,
+      "loss": 0.4516,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999466441552571e-05,
+      "loss": 0.5066,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994363905436836e-05,
+      "loss": 0.4399,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994055165330683e-05,
+      "loss": 0.4453,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993738195461488e-05,
+      "loss": 0.4701,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993412996090265e-05,
+      "loss": 0.3712,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99930795674848e-05,
+      "loss": 0.4016,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992737909919672e-05,
+      "loss": 0.4274,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999238802367621e-05,
+      "loss": 0.4716,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9992029909042542e-05,
+      "loss": 0.4674,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991663566313563e-05,
+      "loss": 0.4731,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999128899579094e-05,
+      "loss": 0.4427,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990906197783123e-05,
+      "loss": 0.4417,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999051517260533e-05,
+      "loss": 0.3991,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990115920579556e-05,
+      "loss": 0.4362,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989708442034573e-05,
+      "loss": 0.3986,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998929273730593e-05,
+      "loss": 0.5312,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988868806735938e-05,
+      "loss": 0.4708,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988436650673693e-05,
+      "loss": 0.3981,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987996269475063e-05,
+      "loss": 0.4282,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987547663502683e-05,
+      "loss": 0.4517,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987090833125967e-05,
+      "loss": 0.4804,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99866257787211e-05,
+      "loss": 0.3926,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998615250067103e-05,
+      "loss": 0.4351,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99856709993655e-05,
+      "loss": 0.4968,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985181275201e-05,
+      "loss": 0.4711,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984683328580803e-05,
+      "loss": 0.4394,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998417715991495e-05,
+      "loss": 0.4331,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983662769620254e-05,
+      "loss": 0.3908,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983140158120304e-05,
+      "loss": 0.387,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982609325845442e-05,
+      "loss": 0.4264,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982070273232796e-05,
+      "loss": 0.4698,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981523000726258e-05,
+      "loss": 0.5196,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980967508776486e-05,
+      "loss": 0.4002,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998040379784091e-05,
+      "loss": 0.4679,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997983186838373e-05,
+      "loss": 0.4006,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979251720875904e-05,
+      "loss": 0.4247,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978663355795167e-05,
+      "loss": 0.3948,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997806677362602e-05,
+      "loss": 0.4608,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997746197485972e-05,
+      "loss": 0.4895,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976848959994308e-05,
+      "loss": 0.4122,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997622772953457e-05,
+      "loss": 0.4318,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975598283992082e-05,
+      "loss": 0.4206,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974960623885158e-05,
+      "loss": 0.4053,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974314749738895e-05,
+      "loss": 0.3595,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973660662085152e-05,
+      "loss": 0.4052,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997299836146254e-05,
+      "loss": 0.4766,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972327848416442e-05,
+      "loss": 0.5164,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971649123499007e-05,
+      "loss": 0.3658,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970962187269144e-05,
+      "loss": 0.4463,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997026704029251e-05,
+      "loss": 0.4211,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996956368314155e-05,
+      "loss": 0.3708,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968852116395442e-05,
+      "loss": 0.3776,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996813234064014e-05,
+      "loss": 0.407,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996740435646836e-05,
+      "loss": 0.4901,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966668164479567e-05,
+      "loss": 0.434,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965923765279995e-05,
+      "loss": 0.4553,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965171159482623e-05,
+      "loss": 0.436,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964410347707203e-05,
+      "loss": 0.3932,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963641330580237e-05,
+      "loss": 0.386,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996286410873498e-05,
+      "loss": 0.4144,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996207868281145e-05,
+      "loss": 0.4924,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996128505345642e-05,
+      "loss": 0.4523,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960483221323417e-05,
+      "loss": 0.4337,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959673187072723e-05,
+      "loss": 0.4302,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958854951371364e-05,
+      "loss": 0.4504,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958028514893142e-05,
+      "loss": 0.3974,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957193878318594e-05,
+      "loss": 0.3504,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956351042335012e-05,
+      "loss": 0.4547,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955500007636445e-05,
+      "loss": 0.4144,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954640774923694e-05,
+      "loss": 0.4692,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953773344904303e-05,
+      "loss": 0.4213,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995289771829257e-05,
+      "loss": 0.483,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995201389580955e-05,
+      "loss": 0.4141,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951121878183038e-05,
+      "loss": 0.3865,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995022166614758e-05,
+      "loss": 0.4042,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949313260444465e-05,
+      "loss": 0.4452,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948396661821742e-05,
+      "loss": 0.4573,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994747187103419e-05,
+      "loss": 0.3573,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994653888884335e-05,
+      "loss": 0.5066,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99455977160175e-05,
+      "loss": 0.4134,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944648353331663e-05,
+      "loss": 0.3814,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943690801567603e-05,
+      "loss": 0.3658,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942725061513833e-05,
+      "loss": 0.3989,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994175113396561e-05,
+      "loss": 0.5208,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940769019724926e-05,
+      "loss": 0.3696,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993977871960052e-05,
+      "loss": 0.4651,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9938780234407876e-05,
+      "loss": 0.4065,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937773564969204e-05,
+      "loss": 0.4734,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936758712113464e-05,
+      "loss": 0.3827,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935735676676355e-05,
+      "loss": 0.3365,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934704459500312e-05,
+      "loss": 0.4288,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933665061434503e-05,
+      "loss": 0.4864,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993261748333484e-05,
+      "loss": 0.3256,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931561726063968e-05,
+      "loss": 0.5209,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930497790491265e-05,
+      "loss": 0.3961,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992942567749285e-05,
+      "loss": 0.4252,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928345387951563e-05,
+      "loss": 0.3556,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9927256922756996e-05,
+      "loss": 0.3554,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9926160282805456e-05,
+      "loss": 0.4678,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925055468999992e-05,
+      "loss": 0.4145,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923942482250374e-05,
+      "loss": 0.4414,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922821323473117e-05,
+      "loss": 0.4528,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921691993591455e-05,
+      "loss": 0.4533,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9920554493535345e-05,
+      "loss": 0.3522,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991940882424149e-05,
+      "loss": 0.331,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918254986653304e-05,
+      "loss": 0.427,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917092981720934e-05,
+      "loss": 0.4034,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991592281040125e-05,
+      "loss": 0.403,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991474447365785e-05,
+      "loss": 0.4546,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9913557972461052e-05,
+      "loss": 0.4171,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912363307787902e-05,
+      "loss": 0.3737,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991116048062216e-05,
+      "loss": 0.3495,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909949491954325e-05,
+      "loss": 0.4024,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990873034278159e-05,
+      "loss": 0.3857,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907503034107893e-05,
+      "loss": 0.4404,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9906267566943874e-05,
+      "loss": 0.3491,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99050239423069e-05,
+      "loss": 0.467,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903772161221064e-05,
+      "loss": 0.4392,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902512224717147e-05,
+      "loss": 0.3827,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901244133832675e-05,
+      "loss": 0.3837,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899967889611876e-05,
+      "loss": 0.3819,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898683493105692e-05,
+      "loss": 0.4193,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897390945371778e-05,
+      "loss": 0.3874,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896090247474506e-05,
+      "loss": 0.4681,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894781400484956e-05,
+      "loss": 0.4525,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893464405480916e-05,
+      "loss": 0.3783,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989213926354689e-05,
+      "loss": 0.3807,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890805975774084e-05,
+      "loss": 0.3888,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988946454326042e-05,
+      "loss": 0.3954,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888114967110518e-05,
+      "loss": 0.4345,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9886757248435705e-05,
+      "loss": 0.3677,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988539138835402e-05,
+      "loss": 0.4599,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884017387990205e-05,
+      "loss": 0.3967,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9882635248475696e-05,
+      "loss": 0.3855,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988124497094864e-05,
+      "loss": 0.3742,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879846556553882e-05,
+      "loss": 0.386,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878440006442972e-05,
+      "loss": 0.4327,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877025321774154e-05,
+      "loss": 0.3653,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987560250371237e-05,
+      "loss": 0.3863,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874171553429265e-05,
+      "loss": 0.4828,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987273247210318e-05,
+      "loss": 0.4186,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871285260919144e-05,
+      "loss": 0.3725,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869829921068892e-05,
+      "loss": 0.3908,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868366453750835e-05,
+      "loss": 0.3771,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866894860170104e-05,
+      "loss": 0.4446,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865415141538493e-05,
+      "loss": 0.3873,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986392729907451e-05,
+      "loss": 0.4295,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862431334003336e-05,
+      "loss": 0.4471,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9860927247556846e-05,
+      "loss": 0.3666,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985941504097361e-05,
+      "loss": 0.4367,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857894715498873e-05,
+      "loss": 0.3437,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985636627238457e-05,
+      "loss": 0.4201,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854829712889333e-05,
+      "loss": 0.3752,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853285038278453e-05,
+      "loss": 0.4297,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851732249823924e-05,
+      "loss": 0.4274,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985017134880441e-05,
+      "loss": 0.4322,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848602336505262e-05,
+      "loss": 0.3705,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984702521421851e-05,
+      "loss": 0.3715,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845439983242857e-05,
+      "loss": 0.3774,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843846644883683e-05,
+      "loss": 0.3925,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984224520045306e-05,
+      "loss": 0.4235,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984063565126971e-05,
+      "loss": 0.4504,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983901799865905e-05,
+      "loss": 0.4599,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983739224395316e-05,
+      "loss": 0.3933,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835758388490783e-05,
+      "loss": 0.3437,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983411643361736e-05,
+      "loss": 0.4113,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832466380684976e-05,
+      "loss": 0.3754,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9830808231052396e-05,
+      "loss": 0.441,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9829141986085046e-05,
+      "loss": 0.4356,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9827467647155024e-05,
+      "loss": 0.4222,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9825785215641088e-05,
+      "loss": 0.4192,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824094692928664e-05,
+      "loss": 0.3911,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9822396080409848e-05,
+      "loss": 0.355,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982068937948338e-05,
+      "loss": 0.3687,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818974591554668e-05,
+      "loss": 0.4259,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981725171803579e-05,
+      "loss": 0.3498,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981552076034547e-05,
+      "loss": 0.4527,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981378171990909e-05,
+      "loss": 0.4281,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981203459815869e-05,
+      "loss": 0.4281,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810279396532974e-05,
+      "loss": 0.3562,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808516116477278e-05,
+      "loss": 0.3921,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980674475944361e-05,
+      "loss": 0.3798,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980496532689062e-05,
+      "loss": 0.3353,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980317782028361e-05,
+      "loss": 0.4321,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801382241094532e-05,
+      "loss": 0.4306,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799578590801983e-05,
+      "loss": 0.4583,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979776687089121e-05,
+      "loss": 0.3667,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97959470828541e-05,
+      "loss": 0.3767,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794119228189185e-05,
+      "loss": 0.3692,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9792283308401644e-05,
+      "loss": 0.3409,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979043932500329e-05,
+      "loss": 0.4199,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9788587279512582e-05,
+      "loss": 0.4192,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9786727173454623e-05,
+      "loss": 0.4148,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784859008361136e-05,
+      "loss": 0.3133,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9782982785770497e-05,
+      "loss": 0.3976,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978109850722771e-05,
+      "loss": 0.3789,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9779206174284417e-05,
+      "loss": 0.3537,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9777305788498886e-05,
+      "loss": 0.3928,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977539735143602e-05,
+      "loss": 0.4246,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9773480864667355e-05,
+      "loss": 0.4517,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9771556329771043e-05,
+      "loss": 0.3679,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769623748331885e-05,
+      "loss": 0.4052,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976768312194129e-05,
+      "loss": 0.3691,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97657344521973e-05,
+      "loss": 0.3822,
+      "step": 347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9763777740704572e-05,
+      "loss": 0.3503,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9761812989074393e-05,
+      "loss": 0.3852,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759840198924674e-05,
+      "loss": 0.4198,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9757859371879936e-05,
+      "loss": 0.4325,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9755870509571324e-05,
+      "loss": 0.3265,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9753873613636597e-05,
+      "loss": 0.4428,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9751868685720136e-05,
+      "loss": 0.3609,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974985572747292e-05,
+      "loss": 0.3354,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974783474055256e-05,
+      "loss": 0.3611,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745805726623267e-05,
+      "loss": 0.4271,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974376868735586e-05,
+      "loss": 0.4342,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9741723624427783e-05,
+      "loss": 0.4053,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9739670539523065e-05,
+      "loss": 0.4272,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9737609434332353e-05,
+      "loss": 0.3703,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97355403105529e-05,
+      "loss": 0.3721,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733463169888557e-05,
+      "loss": 0.3178,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973137801404978e-05,
+      "loss": 0.3861,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.972928484475362e-05,
+      "loss": 0.4553,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972718366372373e-05,
+      "loss": 0.4149,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972507447269037e-05,
+      "loss": 0.3414,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9722957273390377e-05,
+      "loss": 0.3803,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9720832067567195e-05,
+      "loss": 0.3998,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9718698856970855e-05,
+      "loss": 0.365,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9716557643357993e-05,
+      "loss": 0.3415,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714408428491817e-05,
+      "loss": 0.4139,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712251214142133e-05,
+      "loss": 0.4186,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710086002085335e-05,
+      "loss": 0.361,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.97079127941044e-05,
+      "loss": 0.4242,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9705731591988892e-05,
+      "loss": 0.3816,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9703542397534958e-05,
+      "loss": 0.3742,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970134521254532e-05,
+      "loss": 0.3089,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9699140038829287e-05,
+      "loss": 0.3609,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9696926878202746e-05,
+      "loss": 0.4385,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9694705732488157e-05,
+      "loss": 0.3946,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9692476603514555e-05,
+      "loss": 0.3447,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9690239493117557e-05,
+      "loss": 0.4016,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9687994403139347e-05,
+      "loss": 0.3733,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9685741335428675e-05,
+      "loss": 0.3686,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968348029184087e-05,
+      "loss": 0.3489,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968121127423782e-05,
+      "loss": 0.4269,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9678934284487988e-05,
+      "loss": 0.4059,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9676649324466394e-05,
+      "loss": 0.3663,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9674356396054624e-05,
+      "loss": 0.3857,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9672055501140827e-05,
+      "loss": 0.4161,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9669746641619717e-05,
+      "loss": 0.3876,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9667429819392555e-05,
+      "loss": 0.3207,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966510503636716e-05,
+      "loss": 0.3603,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9662772294457923e-05,
+      "loss": 0.3937,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660431595585773e-05,
+      "loss": 0.4273,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658082941678188e-05,
+      "loss": 0.3079,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9655726334669216e-05,
+      "loss": 0.4088,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9653361776499433e-05,
+      "loss": 0.4115,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965098926911598e-05,
+      "loss": 0.3498,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964860881447253e-05,
+      "loss": 0.3522,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9646220414529305e-05,
+      "loss": 0.3705,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643824071253074e-05,
+      "loss": 0.4268,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9641419786617143e-05,
+      "loss": 0.3738,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9639007562601363e-05,
+      "loss": 0.3881,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963658740119211e-05,
+      "loss": 0.4248,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9634159304382306e-05,
+      "loss": 0.3916,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9631723274171412e-05,
+      "loss": 0.3213,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9629279312565408e-05,
+      "loss": 0.3758,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9626827421576816e-05,
+      "loss": 0.3613,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962436760322469e-05,
+      "loss": 0.4316,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9621899859534594e-05,
+      "loss": 0.3513,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9619424192538642e-05,
+      "loss": 0.4152,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9616940604275454e-05,
+      "loss": 0.392,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9614449096790188e-05,
+      "loss": 0.3421,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611949672134506e-05,
+      "loss": 0.3771,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9609442332366605e-05,
+      "loss": 0.356,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9606927079551187e-05,
+      "loss": 0.4039,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9604403915759486e-05,
+      "loss": 0.3748,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601872843069236e-05,
+      "loss": 0.3898,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959933386356469e-05,
+      "loss": 0.3945,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959678697933661e-05,
+      "loss": 0.3803,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959423219248227e-05,
+      "loss": 0.3436,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591669505105448e-05,
+      "loss": 0.3425,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9589098919316428e-05,
+      "loss": 0.3909,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9586520437232e-05,
+      "loss": 0.4082,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958393406097546e-05,
+      "loss": 0.3453,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9581339792676602e-05,
+      "loss": 0.405,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9578737634471707e-05,
+      "loss": 0.4561,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957612758850357e-05,
+      "loss": 0.3166,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9573509656921477e-05,
+      "loss": 0.3414,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9570883841881204e-05,
+      "loss": 0.38,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9568250145545014e-05,
+      "loss": 0.3837,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9565608570081674e-05,
+      "loss": 0.4274,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9562959117666426e-05,
+      "loss": 0.3427,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9560301790481005e-05,
+      "loss": 0.4428,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955763659071363e-05,
+      "loss": 0.3832,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9554963520559003e-05,
+      "loss": 0.3882,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95522825822183e-05,
+      "loss": 0.3375,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9549593777899193e-05,
+      "loss": 0.3671,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954689710981581e-05,
+      "loss": 0.4396,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9544192580188766e-05,
+      "loss": 0.3746,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9541480191245157e-05,
+      "loss": 0.4219,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9538759945218534e-05,
+      "loss": 0.4299,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9536031844348936e-05,
+      "loss": 0.4202,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533295890882848e-05,
+      "loss": 0.3604,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530552087073243e-05,
+      "loss": 0.3703,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527800435179548e-05,
+      "loss": 0.3723,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952504093746766e-05,
+      "loss": 0.3843,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952227359620992e-05,
+      "loss": 0.4015,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9519498413685145e-05,
+      "loss": 0.3958,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95167153921786e-05,
+      "loss": 0.4309,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513924533982013e-05,
+      "loss": 0.3153,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951112584139356e-05,
+      "loss": 0.3571,
+      "step": 454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508319316717863e-05,
+      "loss": 0.356,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9505504962266005e-05,
+      "loss": 0.3469,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950268278035551e-05,
+      "loss": 0.3742,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9499852773310345e-05,
+      "loss": 0.391,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9497014943460924e-05,
+      "loss": 0.3748,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9494169293144106e-05,
+      "loss": 0.4029,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949131582470319e-05,
+      "loss": 0.3547,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94884545404879e-05,
+      "loss": 0.354,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485585442854416e-05,
+      "loss": 0.3743,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9482708534165337e-05,
+      "loss": 0.3553,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94798238167897e-05,
+      "loss": 0.4202,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947693129310297e-05,
+      "loss": 0.4155,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947403096548704e-05,
+      "loss": 0.4192,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9471122836330236e-05,
+      "loss": 0.3461,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9468206908027296e-05,
+      "loss": 0.3437,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465283182979396e-05,
+      "loss": 0.3682,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9462351663594113e-05,
+      "loss": 0.3527,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459412352285467e-05,
+      "loss": 0.427,
+      "step": 472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9456465251473867e-05,
+      "loss": 0.4119,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9453510363586155e-05,
+      "loss": 0.4222,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9450547691055582e-05,
+      "loss": 0.3503,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9447577236321806e-05,
+      "loss": 0.3798,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9444599001830896e-05,
+      "loss": 0.3719,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441612990035324e-05,
+      "loss": 0.3358,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438619203393972e-05,
+      "loss": 0.3768,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943561764437212e-05,
+      "loss": 0.396,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9432608315441442e-05,
+      "loss": 0.3906,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9429591219080033e-05,
+      "loss": 0.4184,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9426566357772356e-05,
+      "loss": 0.357,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9423533734009287e-05,
+      "loss": 0.3831,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9420493350288082e-05,
+      "loss": 0.3316,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94174452091124e-05,
+      "loss": 0.34,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941438931299228e-05,
+      "loss": 0.3734,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9411325664444143e-05,
+      "loss": 0.4196,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408254265990805e-05,
+      "loss": 0.4103,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405175120161453e-05,
+      "loss": 0.397,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.940208822949166e-05,
+      "loss": 0.3746,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939899359652338e-05,
+      "loss": 0.315,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939589122380493e-05,
+      "loss": 0.3416,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939278111389101e-05,
+      "loss": 0.3574,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9389663269342688e-05,
+      "loss": 0.3935,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938653769272741e-05,
+      "loss": 0.4447,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9383404386618972e-05,
+      "loss": 0.4006,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9380263353597553e-05,
+      "loss": 0.3872,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9377114596249676e-05,
+      "loss": 0.3599,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937395811716825e-05,
+      "loss": 0.3165,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9370793918952514e-05,
+      "loss": 0.3574,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936762200420808e-05,
+      "loss": 0.3821,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936444237554692e-05,
+      "loss": 0.4288,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9361255035587338e-05,
+      "loss": 0.3952,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9358059986954006e-05,
+      "loss": 0.3717,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9354857232277938e-05,
+      "loss": 0.3557,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9351646774196488e-05,
+      "loss": 0.351,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934842861535337e-05,
+      "loss": 0.3307,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9345202758398608e-05,
+      "loss": 0.337,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9341969205988605e-05,
+      "loss": 0.3749,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933872796078607e-05,
+      "loss": 0.4081,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933547902546006e-05,
+      "loss": 0.4222,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9332222402685966e-05,
+      "loss": 0.3291,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93289580951455e-05,
+      "loss": 0.3859,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9325686105526707e-05,
+      "loss": 0.3371,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9322406436523965e-05,
+      "loss": 0.3445,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9319119090837966e-05,
+      "loss": 0.3817,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931582407117573e-05,
+      "loss": 0.3929,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9312521380250586e-05,
+      "loss": 0.4222,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9309211020782192e-05,
+      "loss": 0.3957,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9305892995496514e-05,
+      "loss": 0.3872,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9302567307125836e-05,
+      "loss": 0.3421,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299233958408744e-05,
+      "loss": 0.3582,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9295892952090143e-05,
+      "loss": 0.2832,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929254429092123e-05,
+      "loss": 0.3965,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928918797765952e-05,
+      "loss": 0.4041,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9285824015068818e-05,
+      "loss": 0.4288,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9282452405919235e-05,
+      "loss": 0.3122,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9279073152987173e-05,
+      "loss": 0.4183,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9275686259055334e-05,
+      "loss": 0.3875,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9272291726912705e-05,
+      "loss": 0.3182,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9268889559354576e-05,
+      "loss": 0.3588,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265479759182505e-05,
+      "loss": 0.3834,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9262062329204355e-05,
+      "loss": 0.3817,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9258637272234258e-05,
+      "loss": 0.3851,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9255204591092634e-05,
+      "loss": 0.435,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9251764288606174e-05,
+      "loss": 0.3839,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9248316367607858e-05,
+      "loss": 0.3336,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9244860830936922e-05,
+      "loss": 0.3315,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9241397681438887e-05,
+      "loss": 0.3632,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9237926921965536e-05,
+      "loss": 0.4004,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923444855537493e-05,
+      "loss": 0.4331,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923096258453137e-05,
+      "loss": 0.3684,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9227469012305448e-05,
+      "loss": 0.4,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9223967841573987e-05,
+      "loss": 0.4114,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9220459075220095e-05,
+      "loss": 0.2894,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9216942716133116e-05,
+      "loss": 0.3437,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9213418767208648e-05,
+      "loss": 0.42,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920988723134855e-05,
+      "loss": 0.3957,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9206348111460914e-05,
+      "loss": 0.3945,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920280141046009e-05,
+      "loss": 0.3616,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919924713126666e-05,
+      "loss": 0.4121,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9195685276807456e-05,
+      "loss": 0.3468,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919211585001554e-05,
+      "loss": 0.3174,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9188538853830216e-05,
+      "loss": 0.379,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9184954291197013e-05,
+      "loss": 0.3575,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9181362165067706e-05,
+      "loss": 0.4788,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9177762478400276e-05,
+      "loss": 0.3478,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917415523415895e-05,
+      "loss": 0.4393,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9170540435314167e-05,
+      "loss": 0.3539,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9166918084842595e-05,
+      "loss": 0.3636,
+      "step": 561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916328818572711e-05,
+      "loss": 0.3318,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915965074095681e-05,
+      "loss": 0.365,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9156005753527013e-05,
+      "loss": 0.4124,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9152353226439237e-05,
+      "loss": 0.3766,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914869316270121e-05,
+      "loss": 0.3847,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9145025565326878e-05,
+      "loss": 0.3969,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9141350437336374e-05,
+      "loss": 0.3863,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9137667781756043e-05,
+      "loss": 0.3046,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9133977601618433e-05,
+      "loss": 0.3535,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913027989996227e-05,
+      "loss": 0.3867,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912657467983249e-05,
+      "loss": 0.3776,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9122861944280217e-05,
+      "loss": 0.3446,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911914169636276e-05,
+      "loss": 0.4352,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9115413939143615e-05,
+      "loss": 0.3842,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9111678675692468e-05,
+      "loss": 0.3872,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9107935909085174e-05,
+      "loss": 0.3309,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9104185642403775e-05,
+      "loss": 0.3748,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910042787873649e-05,
+      "loss": 0.4084,
+      "step": 579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.909666262117771e-05,
+      "loss": 0.3996,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9092889872827992e-05,
+      "loss": 0.411,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908910963679407e-05,
+      "loss": 0.4179,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9085321916188833e-05,
+      "loss": 0.3916,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081526714131345e-05,
+      "loss": 0.2984,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907772403374682e-05,
+      "loss": 0.3258,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9073913878166642e-05,
+      "loss": 0.4083,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907009625052834e-05,
+      "loss": 0.3585,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9066271153975602e-05,
+      "loss": 0.4098,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906243859165826e-05,
+      "loss": 0.3777,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90585985667323e-05,
+      "loss": 0.3876,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9054751082359858e-05,
+      "loss": 0.3664,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9050896141709198e-05,
+      "loss": 0.3312,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9047033747954734e-05,
+      "loss": 0.3654,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904316390427702e-05,
+      "loss": 0.3625,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903928661386273e-05,
+      "loss": 0.3471,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9035401879904687e-05,
+      "loss": 0.3971,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9031509705601837e-05,
+      "loss": 0.3914,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902761009415925e-05,
+      "loss": 0.3635,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9023703048788126e-05,
+      "loss": 0.2889,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901978857270578e-05,
+      "loss": 0.3573,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9015866669135654e-05,
+      "loss": 0.4242,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9011937341307292e-05,
+      "loss": 0.3492,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9008000592456373e-05,
+      "loss": 0.3867,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9004056425824672e-05,
+      "loss": 0.3633,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9000104844660068e-05,
+      "loss": 0.435,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899614585221656e-05,
+      "loss": 0.322,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8992179451754245e-05,
+      "loss": 0.3247,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898820564653931e-05,
+      "loss": 0.3728,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8984224439844056e-05,
+      "loss": 0.3461,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898023583494687e-05,
+      "loss": 0.3596,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8976239835132222e-05,
+      "loss": 0.4196,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8972236443690694e-05,
+      "loss": 0.4117,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8968225663918933e-05,
+      "loss": 0.3564,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8964207499119686e-05,
+      "loss": 0.331,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896018195260177e-05,
+      "loss": 0.3241,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8956149027680085e-05,
+      "loss": 0.3786,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8952108727675613e-05,
+      "loss": 0.3899,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8948061055915395e-05,
+      "loss": 0.3699,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8944006015732556e-05,
+      "loss": 0.3981,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893994361046628e-05,
+      "loss": 0.4283,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8935873843461818e-05,
+      "loss": 0.3675,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8931796718070487e-05,
+      "loss": 0.3654,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892771223764966e-05,
+      "loss": 0.3275,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8923620405562768e-05,
+      "loss": 0.3725,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8919521225179292e-05,
+      "loss": 0.3712,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891541469987477e-05,
+      "loss": 0.3782,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8911300833030777e-05,
+      "loss": 0.4255,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8907179628034957e-05,
+      "loss": 0.3904,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890305108828097e-05,
+      "loss": 0.3507,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8898915217168528e-05,
+      "loss": 0.3622,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8894772018103385e-05,
+      "loss": 0.3053,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8890621494497322e-05,
+      "loss": 0.3892,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8886463649768154e-05,
+      "loss": 0.3737,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.888229848733972e-05,
+      "loss": 0.4047,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8878126010641894e-05,
+      "loss": 0.38,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8873946223110564e-05,
+      "loss": 0.3753,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8869759128187643e-05,
+      "loss": 0.3493,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8865564729321062e-05,
+      "loss": 0.3224,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8861363029964764e-05,
+      "loss": 0.3375,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885715403357871e-05,
+      "loss": 0.3542,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8852937743628857e-05,
+      "loss": 0.3827,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8848714163587178e-05,
+      "loss": 0.402,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884448329693165e-05,
+      "loss": 0.4388,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8840245147146245e-05,
+      "loss": 0.3578,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8835999717720935e-05,
+      "loss": 0.3552,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8831747012151687e-05,
+      "loss": 0.3443,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8827487033940454e-05,
+      "loss": 0.3279,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882321978659519e-05,
+      "loss": 0.4111,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8818945273629823e-05,
+      "loss": 0.3864,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8814663498564267e-05,
+      "loss": 0.4004,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8810374464924422e-05,
+      "loss": 0.3847,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8806078176242158e-05,
+      "loss": 0.3917,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.880177463605532e-05,
+      "loss": 0.3405,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8797463847907732e-05,
+      "loss": 0.3588,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8793145815349173e-05,
+      "loss": 0.3551,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8788820541935403e-05,
+      "loss": 0.3802,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8784488031228133e-05,
+      "loss": 0.4096,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8780148286795037e-05,
+      "loss": 0.4351,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8775801312209748e-05,
+      "loss": 0.3884,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8771447111051854e-05,
+      "loss": 0.3206,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8767085686906882e-05,
+      "loss": 0.3606,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8762717043366324e-05,
+      "loss": 0.3059,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8758341184027603e-05,
+      "loss": 0.3623,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87539581124941e-05,
+      "loss": 0.4077,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874956783237511e-05,
+      "loss": 0.3133,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874517034728589e-05,
+      "loss": 0.4317,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8740765660847614e-05,
+      "loss": 0.3351,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873635377668739e-05,
+      "loss": 0.3389,
+      "step": 668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8731934698438258e-05,
+      "loss": 0.316,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8727508429739172e-05,
+      "loss": 0.3588,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872307497423502e-05,
+      "loss": 0.4134,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.87186343355766e-05,
+      "loss": 0.3878,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871418651742062e-05,
+      "loss": 0.4305,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870973152342971e-05,
+      "loss": 0.347,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8705269357272406e-05,
+      "loss": 0.3883,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870080002262315e-05,
+      "loss": 0.3655,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869632352316229e-05,
+      "loss": 0.3297,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869183986257606e-05,
+      "loss": 0.3587,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868734904455661e-05,
+      "loss": 0.3623,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868285107280197e-05,
+      "loss": 0.3559,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867834595101607e-05,
+      "loss": 0.4084,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8673833682908724e-05,
+      "loss": 0.3447,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8669314272195626e-05,
+      "loss": 0.3486,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866478772259836e-05,
+      "loss": 0.309,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.2819,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865571322166704e-05,
+      "loss": 0.3628,
+      "step": 686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8651165277805522e-05,
+      "loss": 0.3882,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8646610210004912e-05,
+      "loss": 0.3428,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8642048022016155e-05,
+      "loss": 0.4124,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8637478717596056e-05,
+      "loss": 0.3995,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8632902300507277e-05,
+      "loss": 0.3762,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8628318774518347e-05,
+      "loss": 0.2993,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862372814340364e-05,
+      "loss": 0.375,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8619130410943386e-05,
+      "loss": 0.3927,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8614525580923657e-05,
+      "loss": 0.4088,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8609913657136375e-05,
+      "loss": 0.3589,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8605294643379304e-05,
+      "loss": 0.4527,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860066854345604e-05,
+      "loss": 0.3658,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859603536117602e-05,
+      "loss": 0.3386,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859139510035451e-05,
+      "loss": 0.2991,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8586747764812612e-05,
+      "loss": 0.3599,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8582093358377245e-05,
+      "loss": 0.4021,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8577431884881148e-05,
+      "loss": 0.3148,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8572763348162893e-05,
+      "loss": 0.3937,
+      "step": 704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8568087752066856e-05,
+      "loss": 0.3991,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8563405100443233e-05,
+      "loss": 0.3777,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8558715397148023e-05,
+      "loss": 0.3521,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8554018646043045e-05,
+      "loss": 0.3148,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8549314850995906e-05,
+      "loss": 0.3655,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8544604015880024e-05,
+      "loss": 0.3858,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853988614457461e-05,
+      "loss": 0.3485,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8535161240964666e-05,
+      "loss": 0.4077,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8530429308940997e-05,
+      "loss": 0.3488,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852569035240018e-05,
+      "loss": 0.3424,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8520944375244586e-05,
+      "loss": 0.2961,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8516191381382367e-05,
+      "loss": 0.3551,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851143137472745e-05,
+      "loss": 0.4264,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8506664359199536e-05,
+      "loss": 0.3612,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85018903387241e-05,
+      "loss": 0.3345,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8497109317232386e-05,
+      "loss": 0.36,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8492321298661405e-05,
+      "loss": 0.3579,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8487526286953922e-05,
+      "loss": 0.3142,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8482724286058463e-05,
+      "loss": 0.2877,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8477915299929317e-05,
+      "loss": 0.4163,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847309933252652e-05,
+      "loss": 0.3637,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8468276387815853e-05,
+      "loss": 0.3235,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846344646976885e-05,
+      "loss": 0.41,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8458609582362783e-05,
+      "loss": 0.371,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845376572958066e-05,
+      "loss": 0.3909,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8448914915411227e-05,
+      "loss": 0.2897,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.844405714384897e-05,
+      "loss": 0.3468,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8439192418894096e-05,
+      "loss": 0.3733,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8434320744552537e-05,
+      "loss": 0.3818,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8429442124835955e-05,
+      "loss": 0.3438,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8424556563761724e-05,
+      "loss": 0.4347,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8419664065352933e-05,
+      "loss": 0.3919,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8414764633638395e-05,
+      "loss": 0.3587,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840985827265262e-05,
+      "loss": 0.3017,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840494498643583e-05,
+      "loss": 0.3538,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8400024779033944e-05,
+      "loss": 0.3803,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8395097654498592e-05,
+      "loss": 0.3082,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.839016361688709e-05,
+      "loss": 0.3929,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8385222670262445e-05,
+      "loss": 0.3857,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8380274818693366e-05,
+      "loss": 0.3767,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8375320066254234e-05,
+      "loss": 0.2799,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8370358417025116e-05,
+      "loss": 0.3516,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836538987509177e-05,
+      "loss": 0.3461,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8360414444545608e-05,
+      "loss": 0.3475,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8355432129483736e-05,
+      "loss": 0.3054,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835044293400892e-05,
+      "loss": 0.3809,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8345446862229594e-05,
+      "loss": 0.3865,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8340443918259842e-05,
+      "loss": 0.3569,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8335434106219426e-05,
+      "loss": 0.3092,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8330417430233756e-05,
+      "loss": 0.2997,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8325393894433883e-05,
+      "loss": 0.3917,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8320363502956533e-05,
+      "loss": 0.2921,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831532625994405e-05,
+      "loss": 0.3485,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8310282169544436e-05,
+      "loss": 0.4145,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8305231235911324e-05,
+      "loss": 0.3549,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8300173463203986e-05,
+      "loss": 0.2912,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829510885558733e-05,
+      "loss": 0.3218,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829003741723188e-05,
+      "loss": 0.3632,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.82849591523138e-05,
+      "loss": 0.3534,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827987406501486e-05,
+      "loss": 0.3533,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827478215952246e-05,
+      "loss": 0.3218,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826968344002961e-05,
+      "loss": 0.4183,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826457791073493e-05,
+      "loss": 0.3315,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.825946557584265e-05,
+      "loss": 0.3201,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8254346439562603e-05,
+      "loss": 0.2847,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8249220506110225e-05,
+      "loss": 0.376,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8244087779706545e-05,
+      "loss": 0.3858,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8238948264578185e-05,
+      "loss": 0.3655,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8233801964957366e-05,
+      "loss": 0.3776,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8228648885081886e-05,
+      "loss": 0.3874,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822348902919513e-05,
+      "loss": 0.342,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8218322401546063e-05,
+      "loss": 0.3226,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8213149006389228e-05,
+      "loss": 0.3259,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820796884798474e-05,
+      "loss": 0.3725,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820278193059828e-05,
+      "loss": 0.3537,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.81975882585011e-05,
+      "loss": 0.388,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8192387835970013e-05,
+      "loss": 0.3745,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818718066728738e-05,
+      "loss": 0.3765,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8181966756741136e-05,
+      "loss": 0.3493,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8176746108624756e-05,
+      "loss": 0.3334,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8171518727237264e-05,
+      "loss": 0.3339,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816628461688323e-05,
+      "loss": 0.321,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8161043781872768e-05,
+      "loss": 0.4075,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8155796226521522e-05,
+      "loss": 0.3506,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815054195515068e-05,
+      "loss": 0.3933,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8145280972086946e-05,
+      "loss": 0.3027,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814001328166257e-05,
+      "loss": 0.3124,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8134738888215308e-05,
+      "loss": 0.3051,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812945779608845e-05,
+      "loss": 0.3335,
+      "step": 793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8124170009630788e-05,
+      "loss": 0.3544,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8118875533196637e-05,
+      "loss": 0.3944,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8113574371145817e-05,
+      "loss": 0.3534,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8108266527843653e-05,
+      "loss": 0.3663,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.810295200766097e-05,
+      "loss": 0.323,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8097630814974102e-05,
+      "loss": 0.3175,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8092302954164863e-05,
+      "loss": 0.3474,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8086968429620563e-05,
+      "loss": 0.3333,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.808162724573401e-05,
+      "loss": 0.3903,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807627940690347e-05,
+      "loss": 0.3438,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8070924917532725e-05,
+      "loss": 0.3737,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8065563782031003e-05,
+      "loss": 0.3324,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8060196004813013e-05,
+      "loss": 0.3029,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8054821590298944e-05,
+      "loss": 0.3455,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8049440542914438e-05,
+      "loss": 0.3052,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8044052867090607e-05,
+      "loss": 0.3805,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803865856726402e-05,
+      "loss": 0.3736,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803325764787669e-05,
+      "loss": 0.3666,
+      "step": 811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8027850113376092e-05,
+      "loss": 0.3561,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8022435968215154e-05,
+      "loss": 0.3334,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801701521685223e-05,
+      "loss": 0.3141,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801158786375113e-05,
+      "loss": 0.2856,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8006153913381098e-05,
+      "loss": 0.3209,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800071337021679e-05,
+      "loss": 0.3917,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7995266238738324e-05,
+      "loss": 0.3504,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7989812523431218e-05,
+      "loss": 0.3237,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798435222878642e-05,
+      "loss": 0.3635,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79788853593003e-05,
+      "loss": 0.3668,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797341191947464e-05,
+      "loss": 0.3178,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796793191381663e-05,
+      "loss": 0.2907,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7962445346838856e-05,
+      "loss": 0.338,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7956952223059333e-05,
+      "loss": 0.3738,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7951452547001454e-05,
+      "loss": 0.3818,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794594632319402e-05,
+      "loss": 0.3308,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794043355617121e-05,
+      "loss": 0.3641,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7934914250472604e-05,
+      "loss": 0.3112,
+      "step": 829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.792938841064316e-05,
+      "loss": 0.3317,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792385604123322e-05,
+      "loss": 0.3503,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.79183171467985e-05,
+      "loss": 0.3599,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7912771731900095e-05,
+      "loss": 0.3363,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7907219801104458e-05,
+      "loss": 0.3134,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7901661358983417e-05,
+      "loss": 0.3719,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789609641011416e-05,
+      "loss": 0.3301,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789052495907923e-05,
+      "loss": 0.3234,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7884947010466533e-05,
+      "loss": 0.3131,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787936256886931e-05,
+      "loss": 0.3444,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7873771638886163e-05,
+      "loss": 0.4267,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786817422512103e-05,
+      "loss": 0.3332,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7862570332183193e-05,
+      "loss": 0.286,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7856959964687263e-05,
+      "loss": 0.3924,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785134312725319e-05,
+      "loss": 0.3437,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7845719824506244e-05,
+      "loss": 0.311,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7840090061077023e-05,
+      "loss": 0.3207,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7834453841601448e-05,
+      "loss": 0.364,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782881117072075e-05,
+      "loss": 0.3737,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7823162053081476e-05,
+      "loss": 0.2982,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7817506493335485e-05,
+      "loss": 0.368,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7811844496139935e-05,
+      "loss": 0.3839,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7806176066157285e-05,
+      "loss": 0.3358,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7800501208055293e-05,
+      "loss": 0.3171,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7794819926507013e-05,
+      "loss": 0.3039,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778913222619079e-05,
+      "loss": 0.3763,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7783438111790247e-05,
+      "loss": 0.3388,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777773758799429e-05,
+      "loss": 0.3042,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7772030659497112e-05,
+      "loss": 0.3875,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7766317330998168e-05,
+      "loss": 0.3304,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7760597607202195e-05,
+      "loss": 0.3212,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775487149281918e-05,
+      "loss": 0.2837,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7749138992564395e-05,
+      "loss": 0.3486,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7743400111158346e-05,
+      "loss": 0.4106,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7737654853326818e-05,
+      "loss": 0.3105,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7731903223800823e-05,
+      "loss": 0.3026,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7726145227316642e-05,
+      "loss": 0.4024,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772038086861578e-05,
+      "loss": 0.3551,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7714610152444994e-05,
+      "loss": 0.2762,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.770883308355627e-05,
+      "loss": 0.303,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7703049666706826e-05,
+      "loss": 0.3658,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7697259906659112e-05,
+      "loss": 0.4253,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7691463808180797e-05,
+      "loss": 0.288,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7685661376044762e-05,
+      "loss": 0.3874,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7679852615029124e-05,
+      "loss": 0.3656,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767403752991719e-05,
+      "loss": 0.344,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7668216125497487e-05,
+      "loss": 0.2904,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7662388406563745e-05,
+      "loss": 0.3382,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7656554377914887e-05,
+      "loss": 0.4085,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.765071404435504e-05,
+      "loss": 0.3077,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7644867410693517e-05,
+      "loss": 0.3106,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7639014481744822e-05,
+      "loss": 0.3889,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7633155262328644e-05,
+      "loss": 0.3677,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7627289757269847e-05,
+      "loss": 0.2889,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7621417971398472e-05,
+      "loss": 0.3118,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7615539909549743e-05,
+      "loss": 0.3278,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760965557656403e-05,
+      "loss": 0.3743,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7603764977286896e-05,
+      "loss": 0.3164,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7597868116569036e-05,
+      "loss": 0.3564,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7591964999266315e-05,
+      "loss": 0.3939,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7586055630239755e-05,
+      "loss": 0.3415,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7580140014355512e-05,
+      "loss": 0.3036,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7574218156484896e-05,
+      "loss": 0.335,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7568290061504355e-05,
+      "loss": 0.3341,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7562355734295478e-05,
+      "loss": 0.368,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.755641517974497e-05,
+      "loss": 0.3062,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7550468402744685e-05,
+      "loss": 0.4012,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7544515408191583e-05,
+      "loss": 0.3699,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753855620098776e-05,
+      "loss": 0.3276,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753259078604041e-05,
+      "loss": 0.3131,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7526619168261856e-05,
+      "loss": 0.3512,
+      "step": 900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7520641352569518e-05,
+      "loss": 0.3507,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7514657343885923e-05,
+      "loss": 0.289,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7508667147138697e-05,
+      "loss": 0.3479,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7502670767260562e-05,
+      "loss": 0.3379,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749666820918934e-05,
+      "loss": 0.3823,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749065947786792e-05,
+      "loss": 0.2663,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7484644578244293e-05,
+      "loss": 0.3365,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7478623515271526e-05,
+      "loss": 0.3147,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7472596293907756e-05,
+      "loss": 0.3868,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7466562919116192e-05,
+      "loss": 0.2991,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7460523395865116e-05,
+      "loss": 0.3724,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.745447772912787e-05,
+      "loss": 0.3782,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7448425923882846e-05,
+      "loss": 0.2801,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7442367985113504e-05,
+      "loss": 0.2679,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7436303917808353e-05,
+      "loss": 0.2996,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743023372696094e-05,
+      "loss": 0.3552,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7424157417569866e-05,
+      "loss": 0.3561,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7418074994638752e-05,
+      "loss": 0.3402,
+      "step": 918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7411986463176278e-05,
+      "loss": 0.3462,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7405891828196138e-05,
+      "loss": 0.3609,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7399791094717055e-05,
+      "loss": 0.2734,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7393684267762773e-05,
+      "loss": 0.3093,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738757135236206e-05,
+      "loss": 0.3134,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7381452353548687e-05,
+      "loss": 0.3355,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7375327276361443e-05,
+      "loss": 0.3674,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7369196125844124e-05,
+      "loss": 0.3129,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7363058907045518e-05,
+      "loss": 0.3721,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7356915625019416e-05,
+      "loss": 0.3466,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7350766284824597e-05,
+      "loss": 0.2761,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734461089152484e-05,
+      "loss": 0.3273,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7338449450188905e-05,
+      "loss": 0.2944,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7332281965890513e-05,
+      "loss": 0.3375,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7326108443708386e-05,
+      "loss": 0.3797,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731992888872621e-05,
+      "loss": 0.319,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7313743306032632e-05,
+      "loss": 0.3873,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7307551700721273e-05,
+      "loss": 0.3101,
+      "step": 936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730135407789071e-05,
+      "loss": 0.3105,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.729515044264447e-05,
+      "loss": 0.3192,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728894080009103e-05,
+      "loss": 0.3159,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7282725155343828e-05,
+      "loss": 0.3474,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727650351352123e-05,
+      "loss": 0.3445,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7270275879746547e-05,
+      "loss": 0.3586,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726404225914802e-05,
+      "loss": 0.3616,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7257802656858826e-05,
+      "loss": 0.2696,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7251557078017062e-05,
+      "loss": 0.3274,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724530552776575e-05,
+      "loss": 0.3323,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.723904801125283e-05,
+      "loss": 0.3335,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7232784533631148e-05,
+      "loss": 0.3747,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.722651510005847e-05,
+      "loss": 0.3312,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7220239715697455e-05,
+      "loss": 0.3493,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.721395838571567e-05,
+      "loss": 0.3635,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7207671115285577e-05,
+      "loss": 0.2935,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7201377909584526e-05,
+      "loss": 0.3136,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7195078773794756e-05,
+      "loss": 0.3219,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7188773713103394e-05,
+      "loss": 0.37,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7182462732702442e-05,
+      "loss": 0.3378,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7176145837788775e-05,
+      "loss": 0.3717,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7169823033564137e-05,
+      "loss": 0.3476,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7163494325235145e-05,
+      "loss": 0.3486,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7157159718013275e-05,
+      "loss": 0.3225,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7150819217114857e-05,
+      "loss": 0.312,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714447282776108e-05,
+      "loss": 0.3535,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7138120555177976e-05,
+      "loss": 0.3901,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7131762404596424e-05,
+      "loss": 0.327,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712539838125215e-05,
+      "loss": 0.3224,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7119028490385703e-05,
+      "loss": 0.3364,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.711265273724247e-05,
+      "loss": 0.3446,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7106271127072672e-05,
+      "loss": 0.3003,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7099883665131343e-05,
+      "loss": 0.3405,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7093490356678338e-05,
+      "loss": 0.3315,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708709120697833e-05,
+      "loss": 0.3762,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7080686221300804e-05,
+      "loss": 0.3331,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7074275404920043e-05,
+      "loss": 0.3606,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706785876311513e-05,
+      "loss": 0.3756,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7061436301169968e-05,
+      "loss": 0.2922,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7055008024373214e-05,
+      "loss": 0.3063,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7048573938018347e-05,
+      "loss": 0.3158,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7042134047403613e-05,
+      "loss": 0.3828,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7035688357832044e-05,
+      "loss": 0.3729,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7029236874611444e-05,
+      "loss": 0.3509,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7022779603054385e-05,
+      "loss": 0.3429,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7016316548478217e-05,
+      "loss": 0.3228,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7009847716205042e-05,
+      "loss": 0.3249,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700337311156172e-05,
+      "loss": 0.3241,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699689273987987e-05,
+      "loss": 0.3585,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6990406606495854e-05,
+      "loss": 0.3857,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6983914716750783e-05,
+      "loss": 0.359,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6977417075990508e-05,
+      "loss": 0.2894,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697091368956561e-05,
+      "loss": 0.3853,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696440456283141e-05,
+      "loss": 0.3118,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6957889701147943e-05,
+      "loss": 0.3045,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6951369109879987e-05,
+      "loss": 0.3289,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6944842794397018e-05,
+      "loss": 0.3513,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693831076007324e-05,
+      "loss": 0.4244,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693177301228755e-05,
+      "loss": 0.3299,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692522955642357e-05,
+      "loss": 0.3471,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691868039786961e-05,
+      "loss": 0.3397,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6912125542018675e-05,
+      "loss": 0.3148,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.690556499426847e-05,
+      "loss": 0.3021,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6898998760021376e-05,
+      "loss": 0.3016,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689242684468447e-05,
+      "loss": 0.4026,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6885849253669492e-05,
+      "loss": 0.3809,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6879265992392868e-05,
+      "loss": 0.2836,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6872677066275685e-05,
+      "loss": 0.3324,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6866082480743705e-05,
+      "loss": 0.338,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6859482241227337e-05,
+      "loss": 0.3233,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685287635316166e-05,
+      "loss": 0.2765,
+      "step": 1007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.684626482198639e-05,
+      "loss": 0.3837,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6839647653145902e-05,
+      "loss": 0.3399,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6833024852089207e-05,
+      "loss": 0.4039,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6826396424269955e-05,
+      "loss": 0.3396,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6819762375146427e-05,
+      "loss": 0.3291,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6813122710181544e-05,
+      "loss": 0.3328,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6806477434842835e-05,
+      "loss": 0.2927,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6799826554602465e-05,
+      "loss": 0.3345,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6793170074937196e-05,
+      "loss": 0.3358,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678650800132842e-05,
+      "loss": 0.3616,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6779840339262128e-05,
+      "loss": 0.2977,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6773167094228906e-05,
+      "loss": 0.3742,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6766488271723948e-05,
+      "loss": 0.335,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6759803877247033e-05,
+      "loss": 0.3046,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675311391630253e-05,
+      "loss": 0.2663,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6746418394399403e-05,
+      "loss": 0.3199,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6739717317051173e-05,
+      "loss": 0.3671,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6733010689775956e-05,
+      "loss": 0.358,
+      "step": 1025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6726298518096427e-05,
+      "loss": 0.2978,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6719580807539833e-05,
+      "loss": 0.3826,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671285756363798e-05,
+      "loss": 0.3457,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6706128791927226e-05,
+      "loss": 0.2722,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6699394497948486e-05,
+      "loss": 0.3394,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669265468724722e-05,
+      "loss": 0.3272,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6685909365373435e-05,
+      "loss": 0.3646,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.667915853788167e-05,
+      "loss": 0.2887,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6672402210331002e-05,
+      "loss": 0.3819,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6665640388285034e-05,
+      "loss": 0.3642,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6658873077311902e-05,
+      "loss": 0.2949,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665210028298425e-05,
+      "loss": 0.2759,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6645322010879242e-05,
+      "loss": 0.3436,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6638538266578555e-05,
+      "loss": 0.3668,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663174905566837e-05,
+      "loss": 0.3235,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6624954383739375e-05,
+      "loss": 0.3416,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6618154256386746e-05,
+      "loss": 0.3566,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6611348679210153e-05,
+      "loss": 0.3736,
+      "step": 1043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660453765781376e-05,
+      "loss": 0.2972,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659772119780621e-05,
+      "loss": 0.2905,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6590899304800626e-05,
+      "loss": 0.3055,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6584071984414603e-05,
+      "loss": 0.3382,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6577239242270208e-05,
+      "loss": 0.3407,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6570401083993965e-05,
+      "loss": 0.3811,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6563557515216868e-05,
+      "loss": 0.3275,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.655670854157436e-05,
+      "loss": 0.3444,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654985416870634e-05,
+      "loss": 0.2832,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654299440225715e-05,
+      "loss": 0.3427,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6536129247875568e-05,
+      "loss": 0.3341,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.652925871121482e-05,
+      "loss": 0.3344,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6522382797932555e-05,
+      "loss": 0.3421,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6515501513690853e-05,
+      "loss": 0.351,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6508614864156218e-05,
+      "loss": 0.3446,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6501722854999572e-05,
+      "loss": 0.3174,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6494825491896246e-05,
+      "loss": 0.2966,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648792278052598e-05,
+      "loss": 0.3592,
+      "step": 1061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6481014726572924e-05,
+      "loss": 0.3515,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474101335725633e-05,
+      "loss": 0.2959,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6467182613677035e-05,
+      "loss": 0.3939,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6460258566124466e-05,
+      "loss": 0.3139,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645332919876965e-05,
+      "loss": 0.3614,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6446394517318673e-05,
+      "loss": 0.2873,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6439454527482014e-05,
+      "loss": 0.3191,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6432509234974516e-05,
+      "loss": 0.3465,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642555864551539e-05,
+      "loss": 0.3585,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641860276482821e-05,
+      "loss": 0.3069,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6411641598640908e-05,
+      "loss": 0.3641,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.640467515268576e-05,
+      "loss": 0.3406,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6397703432699404e-05,
+      "loss": 0.3103,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6390726444422812e-05,
+      "loss": 0.288,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6383744193601293e-05,
+      "loss": 0.3555,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6376756685984488e-05,
+      "loss": 0.3184,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6369763927326378e-05,
+      "loss": 0.3082,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.636276592338526e-05,
+      "loss": 0.3709,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6355762679923744e-05,
+      "loss": 0.357,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6348754202708773e-05,
+      "loss": 0.3326,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6341740497511584e-05,
+      "loss": 0.2826,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6334721570107718e-05,
+      "loss": 0.312,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632769742627703e-05,
+      "loss": 0.3126,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6320668071803654e-05,
+      "loss": 0.3104,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6313633512476034e-05,
+      "loss": 0.3423,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.630659375408688e-05,
+      "loss": 0.3354,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62995488024332e-05,
+      "loss": 0.3287,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629249866331626e-05,
+      "loss": 0.3164,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628544334254162e-05,
+      "loss": 0.297,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6278382845919093e-05,
+      "loss": 0.3281,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6271317179262754e-05,
+      "loss": 0.388,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6264246348390938e-05,
+      "loss": 0.3033,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6257170359126237e-05,
+      "loss": 0.382,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6250089217295483e-05,
+      "loss": 0.3424,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6243002928729752e-05,
+      "loss": 0.3206,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.623591149926437e-05,
+      "loss": 0.2824,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6228814934738873e-05,
+      "loss": 0.2782,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622171324099705e-05,
+      "loss": 0.315,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6214606423886903e-05,
+      "loss": 0.3348,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6207494489260647e-05,
+      "loss": 0.3552,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6200377442974723e-05,
+      "loss": 0.3869,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6193255290889772e-05,
+      "loss": 0.3759,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6186128038870644e-05,
+      "loss": 0.294,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6178995692786386e-05,
+      "loss": 0.3245,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6171858258510246e-05,
+      "loss": 0.2842,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616471574191965e-05,
+      "loss": 0.3325,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6157568148896227e-05,
+      "loss": 0.3526,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6150415485325764e-05,
+      "loss": 0.3691,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6143257757098242e-05,
+      "loss": 0.3784,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6136094970107807e-05,
+      "loss": 0.3065,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6128927130252764e-05,
+      "loss": 0.3321,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6121754243435588e-05,
+      "loss": 0.2895,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61145763155629e-05,
+      "loss": 0.2826,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.610739335254549e-05,
+      "loss": 0.3368,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6100205360298277e-05,
+      "loss": 0.341,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6093012344740322e-05,
+      "loss": 0.3773,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6085814311794833e-05,
+      "loss": 0.3414,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6078611267389147e-05,
+      "loss": 0.3324,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607140321745472e-05,
+      "loss": 0.3528,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6064190167927135e-05,
+      "loss": 0.2827,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605697212474609e-05,
+      "loss": 0.33,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60497490938554e-05,
+      "loss": 0.2815,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604252108120299e-05,
+      "loss": 0.3945,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6035288092740874e-05,
+      "loss": 0.3192,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6028050134425172e-05,
+      "loss": 0.3327,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6020807212216095e-05,
+      "loss": 0.3393,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6013559332077945e-05,
+      "loss": 0.3212,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60063064999791e-05,
+      "loss": 0.341,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5999048721892024e-05,
+      "loss": 0.309,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5991786003793244e-05,
+      "loss": 0.3129,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5984518351663365e-05,
+      "loss": 0.3469,
+      "step": 1132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5977245771487048e-05,
+      "loss": 0.3361,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5969968269253014e-05,
+      "loss": 0.3031,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5962685850954037e-05,
+      "loss": 0.3197,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5955398522586942e-05,
+      "loss": 0.34,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5948106290152593e-05,
+      "loss": 0.3276,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5940809159655898e-05,
+      "loss": 0.3001,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593350713710579e-05,
+      "loss": 0.3464,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5926200228515235e-05,
+      "loss": 0.3281,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5918888439901226e-05,
+      "loss": 0.3646,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591157177728477e-05,
+      "loss": 0.2773,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5904250246690888e-05,
+      "loss": 0.2928,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5896923854148613e-05,
+      "loss": 0.3384,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5889592605690976e-05,
+      "loss": 0.2897,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588225650735501e-05,
+      "loss": 0.337,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5874915565181742e-05,
+      "loss": 0.3756,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5867569785216187e-05,
+      "loss": 0.312,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5860219173507343e-05,
+      "loss": 0.3211,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585286373610819e-05,
+      "loss": 0.3261,
+      "step": 1150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5845503479075676e-05,
+      "loss": 0.3139,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5838138408470723e-05,
+      "loss": 0.2912,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5830768530358218e-05,
+      "loss": 0.2562,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5823393850807e-05,
+      "loss": 0.329,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5816014375889865e-05,
+      "loss": 0.3855,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5808630111683566e-05,
+      "loss": 0.3641,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580124106426878e-05,
+      "loss": 0.2785,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5793847239730148e-05,
+      "loss": 0.3153,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578644864415622e-05,
+      "loss": 0.2986,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5779045283639495e-05,
+      "loss": 0.2919,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5771637164276386e-05,
+      "loss": 0.3347,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5764224292167223e-05,
+      "loss": 0.3284,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575680667341626e-05,
+      "loss": 0.3772,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5749384314131643e-05,
+      "loss": 0.3188,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5741957220425435e-05,
+      "loss": 0.3466,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5734525398413597e-05,
+      "loss": 0.3167,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5727088854215984e-05,
+      "loss": 0.2603,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5719647593956328e-05,
+      "loss": 0.3148,
+      "step": 1168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5712201623762253e-05,
+      "loss": 0.3099,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5704750949765263e-05,
+      "loss": 0.3133,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5697295578100735e-05,
+      "loss": 0.3346,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5689835514907916e-05,
+      "loss": 0.3143,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5682370766329903e-05,
+      "loss": 0.2836,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5674901338513672e-05,
+      "loss": 0.3056,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5667427237610037e-05,
+      "loss": 0.2467,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5659948469773665e-05,
+      "loss": 0.2908,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5652465041163068e-05,
+      "loss": 0.3526,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5644976957940586e-05,
+      "loss": 0.364,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563748422627241e-05,
+      "loss": 0.3081,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5629986852328546e-05,
+      "loss": 0.3667,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5622484842282818e-05,
+      "loss": 0.2985,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5614978202312886e-05,
+      "loss": 0.3134,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5607466938600204e-05,
+      "loss": 0.2816,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559995105733004e-05,
+      "loss": 0.3476,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5592430564691474e-05,
+      "loss": 0.3409,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558490546687737e-05,
+      "loss": 0.3124,
+      "step": 1186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5577375770084384e-05,
+      "loss": 0.365,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5569841480512972e-05,
+      "loss": 0.2944,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5562302604367356e-05,
+      "loss": 0.3022,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5554759147855554e-05,
+      "loss": 0.2763,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5547211117189335e-05,
+      "loss": 0.2991,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5539658518584248e-05,
+      "loss": 0.366,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5532101358259598e-05,
+      "loss": 0.3067,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552453964243845e-05,
+      "loss": 0.2992,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5516973377347625e-05,
+      "loss": 0.3691,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5509402569217677e-05,
+      "loss": 0.339,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.55018272242829e-05,
+      "loss": 0.2811,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.549424734878135e-05,
+      "loss": 0.2673,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5486662948954783e-05,
+      "loss": 0.3439,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5479074031048695e-05,
+      "loss": 0.333,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5471480601312302e-05,
+      "loss": 0.3021,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5463882665998534e-05,
+      "loss": 0.2937,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.545628023136403e-05,
+      "loss": 0.3264,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.544867330366914e-05,
+      "loss": 0.3341,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5441061889177902e-05,
+      "loss": 0.2695,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5433445994158057e-05,
+      "loss": 0.3033,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5425825624881043e-05,
+      "loss": 0.3321,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5418200787621966e-05,
+      "loss": 0.3465,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5410571488659614e-05,
+      "loss": 0.2805,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5402937734276466e-05,
+      "loss": 0.4038,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5395299530758653e-05,
+      "loss": 0.3058,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5387656884395972e-05,
+      "loss": 0.3296,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5380009801481883e-05,
+      "loss": 0.2769,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.53723582883135e-05,
+      "loss": 0.2942,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536470235119158e-05,
+      "loss": 0.3934,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5357041996420527e-05,
+      "loss": 0.3165,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5349377230308376e-05,
+      "loss": 0.2871,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534170805916681e-05,
+      "loss": 0.3149,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5334034489311125e-05,
+      "loss": 0.3569,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5326356527060243e-05,
+      "loss": 0.276,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.53186741787367e-05,
+      "loss": 0.2765,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5310987450666655e-05,
+      "loss": 0.3226,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5303296349179858e-05,
+      "loss": 0.3023,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529560088060968e-05,
+      "loss": 0.2458,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5287901051293062e-05,
+      "loss": 0.3226,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5280196867570558e-05,
+      "loss": 0.3594,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.52724883357863e-05,
+      "loss": 0.2821,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5264775462288002e-05,
+      "loss": 0.301,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.525705825342695e-05,
+      "loss": 0.3063,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5249336715558003e-05,
+      "loss": 0.2923,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5241610855039578e-05,
+      "loss": 0.3461,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5233880678233665e-05,
+      "loss": 0.2769,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5226146191505797e-05,
+      "loss": 0.372,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5218407401225057e-05,
+      "loss": 0.3086,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5210664313764083e-05,
+      "loss": 0.2646,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5202916935499032e-05,
+      "loss": 0.2785,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5195165272809616e-05,
+      "loss": 0.2925,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5187409332079058e-05,
+      "loss": 0.3225,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5179649119694106e-05,
+      "loss": 0.2824,
+      "step": 1239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.517188464204504e-05,
+      "loss": 0.3359,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.516411590552564e-05,
+      "loss": 0.3393,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5156342916533187e-05,
+      "loss": 0.3302,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5148565681468479e-05,
+      "loss": 0.2844,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5140784206735803e-05,
+      "loss": 0.3006,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5132998498742937e-05,
+      "loss": 0.2893,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5125208563901143e-05,
+      "loss": 0.3342,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.511741440862517e-05,
+      "loss": 0.2702,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.510961603933324e-05,
+      "loss": 0.3213,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5101813462447035e-05,
+      "loss": 0.3257,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.509400668439172e-05,
+      "loss": 0.278,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5086195711595904e-05,
+      "loss": 0.2957,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.507838055049166e-05,
+      "loss": 0.2981,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5070561207514504e-05,
+      "loss": 0.3413,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5062737689103399e-05,
+      "loss": 0.2995,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5054910001700745e-05,
+      "loss": 0.3481,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5047078151752375e-05,
+      "loss": 0.3626,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5039242145707551e-05,
+      "loss": 0.3311,
+      "step": 1257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5031401990018955e-05,
+      "loss": 0.2603,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5023557691142685e-05,
+      "loss": 0.2788,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5015709255538258e-05,
+      "loss": 0.2939,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.500785668966859e-05,
+      "loss": 0.3179,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2933,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4992139193002207e-05,
+      "loss": 0.2959,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4984274275148316e-05,
+      "loss": 0.3147,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4976405252914814e-05,
+      "loss": 0.2431,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4968532132781573e-05,
+      "loss": 0.2826,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4960654921231846e-05,
+      "loss": 0.3049,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4952773624752237e-05,
+      "loss": 0.3093,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4944888249832732e-05,
+      "loss": 0.272,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4936998802966667e-05,
+      "loss": 0.3646,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4929105290650727e-05,
+      "loss": 0.3131,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4921207719384956e-05,
+      "loss": 0.3319,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4913306095672732e-05,
+      "loss": 0.2731,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4905400426020769e-05,
+      "loss": 0.3089,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.489749071693912e-05,
+      "loss": 0.2785,
+      "step": 1275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4889576974941158e-05,
+      "loss": 0.3147,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.488165920654358e-05,
+      "loss": 0.2991,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4873737418266398e-05,
+      "loss": 0.3097,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4865811616632934e-05,
+      "loss": 0.3337,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4857881808169814e-05,
+      "loss": 0.2617,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4849947999406966e-05,
+      "loss": 0.2564,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.484201019687761e-05,
+      "loss": 0.3071,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4834068407118255e-05,
+      "loss": 0.2616,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4826122636668692e-05,
+      "loss": 0.288,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4818172892071993e-05,
+      "loss": 0.301,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4810219179874502e-05,
+      "loss": 0.2791,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4802261506625827e-05,
+      "loss": 0.3178,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4794299878878844e-05,
+      "loss": 0.2691,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4786334303189676e-05,
+      "loss": 0.2915,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4778364786117708e-05,
+      "loss": 0.2822,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4770391334225563e-05,
+      "loss": 0.2833,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.47624139540791e-05,
+      "loss": 0.3516,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4754432652247433e-05,
+      "loss": 0.3233,
+      "step": 1293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.474644743530288e-05,
+      "loss": 0.3084,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4738458309820996e-05,
+      "loss": 0.275,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4730465282380557e-05,
+      "loss": 0.3029,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4722468359563542e-05,
+      "loss": 0.3065,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4714467547955152e-05,
+      "loss": 0.2741,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4706462854143775e-05,
+      "loss": 0.3211,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4698454284721003e-05,
+      "loss": 0.3606,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4690441846281629e-05,
+      "loss": 0.3547,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.468242554542361e-05,
+      "loss": 0.295,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.46744053887481e-05,
+      "loss": 0.3419,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.466638138285943e-05,
+      "loss": 0.2883,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4658353534365087e-05,
+      "loss": 0.2732,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4650321849875732e-05,
+      "loss": 0.2889,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4642286336005187e-05,
+      "loss": 0.2842,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4634246999370415e-05,
+      "loss": 0.3245,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4626203846591542e-05,
+      "loss": 0.314,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4618156884291832e-05,
+      "loss": 0.3091,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4610106119097673e-05,
+      "loss": 0.3051,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4602051557638604e-05,
+      "loss": 0.2982,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4593993206547279e-05,
+      "loss": 0.3068,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4585931072459474e-05,
+      "loss": 0.2963,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.457786516201408e-05,
+      "loss": 0.3505,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4569795481853102e-05,
+      "loss": 0.3186,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4561722038621645e-05,
+      "loss": 0.348,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.455364483896791e-05,
+      "loss": 0.2997,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4545563889543199e-05,
+      "loss": 0.2783,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4537479197001898e-05,
+      "loss": 0.2744,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.452939076800147e-05,
+      "loss": 0.2472,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4521298609202468e-05,
+      "loss": 0.3511,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4513202727268503e-05,
+      "loss": 0.3087,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4505103128866255e-05,
+      "loss": 0.3379,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4496999820665477e-05,
+      "loss": 0.278,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4488892809338952e-05,
+      "loss": 0.3485,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4480782101562538e-05,
+      "loss": 0.2621,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4472667704015125e-05,
+      "loss": 0.2739,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4464549623378632e-05,
+      "loss": 0.2877,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4456427866338034e-05,
+      "loss": 0.3277,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4448302439581315e-05,
+      "loss": 0.2966,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4440173349799483e-05,
+      "loss": 0.2842,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4432040603686573e-05,
+      "loss": 0.353,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4423904207939616e-05,
+      "loss": 0.2794,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4415764169258662e-05,
+      "loss": 0.2941,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4407620494346755e-05,
+      "loss": 0.2695,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4399473189909924e-05,
+      "loss": 0.3211,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4391322262657206e-05,
+      "loss": 0.3419,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.438316771930061e-05,
+      "loss": 0.3023,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4375009566555119e-05,
+      "loss": 0.2567,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4366847811138692e-05,
+      "loss": 0.2868,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4358682459772266e-05,
+      "loss": 0.3109,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4350513519179719e-05,
+      "loss": 0.2992,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.43423409960879e-05,
+      "loss": 0.275,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4334164897226599e-05,
+      "loss": 0.3381,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4325985229328557e-05,
+      "loss": 0.4178,
+      "step": 1346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4317801999129447e-05,
+      "loss": 0.4407,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4309615213367884e-05,
+      "loss": 0.5813,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4301424878785406e-05,
+      "loss": 0.5177,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4293231002126472e-05,
+      "loss": 0.355,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4285033590138457e-05,
+      "loss": 0.3416,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4276832649571657e-05,
+      "loss": 0.3534,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.426862818717926e-05,
+      "loss": 0.353,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4260420209717363e-05,
+      "loss": 0.3781,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4252208723944954e-05,
+      "loss": 0.3084,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424399373662391e-05,
+      "loss": 0.3591,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4235775254519e-05,
+      "loss": 0.3566,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4227553284397853e-05,
+      "loss": 0.2715,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4219327833030986e-05,
+      "loss": 0.3025,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4211098907191778e-05,
+      "loss": 0.3308,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4202866513656464e-05,
+      "loss": 0.3516,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4194630659204145e-05,
+      "loss": 0.3084,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.418639135061676e-05,
+      "loss": 0.3216,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4178148594679098e-05,
+      "loss": 0.3517,
+      "step": 1364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4169902398178794e-05,
+      "loss": 0.3241,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.41616527679063e-05,
+      "loss": 0.2405,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4153399710654911e-05,
+      "loss": 0.3156,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4145143233220741e-05,
+      "loss": 0.2987,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4136883342402707e-05,
+      "loss": 0.3652,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4128620045002552e-05,
+      "loss": 0.255,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4120353347824823e-05,
+      "loss": 0.3414,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4112083257676857e-05,
+      "loss": 0.3352,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4103809781368792e-05,
+      "loss": 0.2934,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4095532925713553e-05,
+      "loss": 0.3021,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408725269752685e-05,
+      "loss": 0.3342,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4078969103627167e-05,
+      "loss": 0.3082,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4070682150835761e-05,
+      "loss": 0.3095,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4062391845976653e-05,
+      "loss": 0.2862,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4054098195876626e-05,
+      "loss": 0.3385,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4045801207365222e-05,
+      "loss": 0.3155,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4037500887274723e-05,
+      "loss": 0.2537,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4029197242440158e-05,
+      "loss": 0.3136,
+      "step": 1382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4020890279699298e-05,
+      "loss": 0.3173,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.401258000589264e-05,
+      "loss": 0.3306,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4004266427863416e-05,
+      "loss": 0.2483,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3995949552457568e-05,
+      "loss": 0.3416,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3987629386523762e-05,
+      "loss": 0.317,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.397930593691337e-05,
+      "loss": 0.2799,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3970979210480467e-05,
+      "loss": 0.2659,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3962649214081828e-05,
+      "loss": 0.3338,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3954315954576923e-05,
+      "loss": 0.3244,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3945979438827903e-05,
+      "loss": 0.3169,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3937639673699603e-05,
+      "loss": 0.2872,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3929296666059545e-05,
+      "loss": 0.3408,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3920950422777894e-05,
+      "loss": 0.2885,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3912600950727511e-05,
+      "loss": 0.2435,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3904248256783897e-05,
+      "loss": 0.2839,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3895892347825205e-05,
+      "loss": 0.2998,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3887533230732245e-05,
+      "loss": 0.3232,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3879170912388467e-05,
+      "loss": 0.266,
+      "step": 1400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3870805399679945e-05,
+      "loss": 0.2817,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3862436699495401e-05,
+      "loss": 0.3716,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3854064818726172e-05,
+      "loss": 0.2855,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3845689764266217e-05,
+      "loss": 0.2687,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3837311543012106e-05,
+      "loss": 0.3097,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3828930161863015e-05,
+      "loss": 0.3069,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3820545627720732e-05,
+      "loss": 0.3156,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3812157947489629e-05,
+      "loss": 0.2962,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3803767128076673e-05,
+      "loss": 0.3331,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3795373176391423e-05,
+      "loss": 0.3294,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3786976099346005e-05,
+      "loss": 0.2804,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3778575903855131e-05,
+      "loss": 0.2744,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3770172596836072e-05,
+      "loss": 0.3622,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3761766185208663e-05,
+      "loss": 0.2803,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.37533566758953e-05,
+      "loss": 0.3112,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3744944075820926e-05,
+      "loss": 0.351,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3736528391913027e-05,
+      "loss": 0.3603,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3728109631101631e-05,
+      "loss": 0.294,
+      "step": 1418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3719687800319305e-05,
+      "loss": 0.2686,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3711262906501136e-05,
+      "loss": 0.3331,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3702834956584736e-05,
+      "loss": 0.2996,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3694403957510235e-05,
+      "loss": 0.3008,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3685969916220271e-05,
+      "loss": 0.3195,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3677532839659993e-05,
+      "loss": 0.3359,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3669092734777039e-05,
+      "loss": 0.2884,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3660649608521551e-05,
+      "loss": 0.2941,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.365220346784615e-05,
+      "loss": 0.268,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3643754319705956e-05,
+      "loss": 0.2946,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3635302171058543e-05,
+      "loss": 0.3107,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3626847028863967e-05,
+      "loss": 0.3346,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3618388900084757e-05,
+      "loss": 0.3083,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3609927791685889e-05,
+      "loss": 0.3609,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.360146371063479e-05,
+      "loss": 0.3548,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3592996663901354e-05,
+      "loss": 0.2513,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3584526658457893e-05,
+      "loss": 0.2915,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3576053701279175e-05,
+      "loss": 0.2928,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3567577799342391e-05,
+      "loss": 0.2645,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3559098959627148e-05,
+      "loss": 0.3549,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.355061718911549e-05,
+      "loss": 0.3112,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3542132494791865e-05,
+      "loss": 0.3413,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.353364488364312e-05,
+      "loss": 0.2638,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3525154362658521e-05,
+      "loss": 0.292,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3516660938829722e-05,
+      "loss": 0.2862,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3508164619150763e-05,
+      "loss": 0.2905,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3499665410618078e-05,
+      "loss": 0.3005,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.349116332023047e-05,
+      "loss": 0.3143,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3482658354989122e-05,
+      "loss": 0.3137,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3474150521897584e-05,
+      "loss": 0.3007,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3465639827961765e-05,
+      "loss": 0.2982,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.345712628018993e-05,
+      "loss": 0.2793,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3448609885592694e-05,
+      "loss": 0.3001,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3440090651183017e-05,
+      "loss": 0.3172,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3431568583976202e-05,
+      "loss": 0.3071,
+      "step": 1453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3423043690989874e-05,
+      "loss": 0.3421,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3414515979243997e-05,
+      "loss": 0.2723,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3405985455760843e-05,
+      "loss": 0.3132,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3397452127565016e-05,
+      "loss": 0.2611,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3388916001683412e-05,
+      "loss": 0.247,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3380377085145241e-05,
+      "loss": 0.2926,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3371835384982014e-05,
+      "loss": 0.2668,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3363290908227526e-05,
+      "loss": 0.3241,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.335474366191786e-05,
+      "loss": 0.3318,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3346193653091385e-05,
+      "loss": 0.3051,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3337640888788734e-05,
+      "loss": 0.311,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3329085376052823e-05,
+      "loss": 0.2823,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3320527121928829e-05,
+      "loss": 0.2603,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3311966133464169e-05,
+      "loss": 0.2681,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3303402417708532e-05,
+      "loss": 0.291,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3294835981713852e-05,
+      "loss": 0.3503,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.328626683253428e-05,
+      "loss": 0.3342,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3277694977226231e-05,
+      "loss": 0.2968,
+      "step": 1471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3269120422848332e-05,
+      "loss": 0.2987,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.326054317646143e-05,
+      "loss": 0.2636,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3251963245128602e-05,
+      "loss": 0.301,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3243380635915126e-05,
+      "loss": 0.2936,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3234795355888488e-05,
+      "loss": 0.353,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3226207412118368e-05,
+      "loss": 0.3272,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.321761681167665e-05,
+      "loss": 0.3259,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3209023561637398e-05,
+      "loss": 0.2789,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3200427669076858e-05,
+      "loss": 0.3061,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.319182914107346e-05,
+      "loss": 0.2887,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3183227984707788e-05,
+      "loss": 0.2428,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3174624207062611e-05,
+      "loss": 0.2938,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.316601781522284e-05,
+      "loss": 0.3162,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3157408816275545e-05,
+      "loss": 0.3366,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3148797217309945e-05,
+      "loss": 0.2986,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3140183025417397e-05,
+      "loss": 0.2958,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3131566247691387e-05,
+      "loss": 0.2867,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.312294689122755e-05,
+      "loss": 0.2566,
+      "step": 1489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3114324963123624e-05,
+      "loss": 0.2939,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105700470479467e-05,
+      "loss": 0.321,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3097073420397068e-05,
+      "loss": 0.3098,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3088443819980495e-05,
+      "loss": 0.3353,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3079811676335933e-05,
+      "loss": 0.3087,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071176996571665e-05,
+      "loss": 0.2953,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3062539787798042e-05,
+      "loss": 0.2743,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3053900057127518e-05,
+      "loss": 0.2465,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3045257811674618e-05,
+      "loss": 0.2811,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036613058555932e-05,
+      "loss": 0.3132,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3027965804890118e-05,
+      "loss": 0.3287,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30193160577979e-05,
+      "loss": 0.2435,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3010663824402039e-05,
+      "loss": 0.3318,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3002009111827363e-05,
+      "loss": 0.2514,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2993351927200729e-05,
+      "loss": 0.2807,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984692277651031e-05,
+      "loss": 0.2761,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976030170309199e-05,
+      "loss": 0.2951,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2967365612308178e-05,
+      "loss": 0.3252,
+      "step": 1507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958698610782939e-05,
+      "loss": 0.3505,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2950029172870457e-05,
+      "loss": 0.3231,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2941357305709724e-05,
+      "loss": 0.3019,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2932683016441724e-05,
+      "loss": 0.3058,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2924006312209434e-05,
+      "loss": 0.233,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2915327200157828e-05,
+      "loss": 0.3302,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2906645687433858e-05,
+      "loss": 0.297,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2897961781186452e-05,
+      "loss": 0.3299,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.288927548856651e-05,
+      "loss": 0.2704,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2880586816726896e-05,
+      "loss": 0.2967,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2871895772822442e-05,
+      "loss": 0.3059,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2863202364009917e-05,
+      "loss": 0.253,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2854506597448049e-05,
+      "loss": 0.2774,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2845808480297504e-05,
+      "loss": 0.3265,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2837108019720885e-05,
+      "loss": 0.3224,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2828405222882726e-05,
+      "loss": 0.2917,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2819700096949484e-05,
+      "loss": 0.3481,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2810992649089524e-05,
+      "loss": 0.3111,
+      "step": 1525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2802282886473143e-05,
+      "loss": 0.308,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2793570816272524e-05,
+      "loss": 0.2444,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2784856445661764e-05,
+      "loss": 0.2837,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2776139781816851e-05,
+      "loss": 0.3103,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2767420831915658e-05,
+      "loss": 0.2995,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2758699603137941e-05,
+      "loss": 0.3317,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2749976102665336e-05,
+      "loss": 0.3089,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2741250337681344e-05,
+      "loss": 0.3156,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2732522315371338e-05,
+      "loss": 0.2716,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2723792042922549e-05,
+      "loss": 0.2391,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.271505952752405e-05,
+      "loss": 0.3102,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2706324776366774e-05,
+      "loss": 0.3164,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2697587796643496e-05,
+      "loss": 0.2849,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2688848595548807e-05,
+      "loss": 0.3138,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2680107180279148e-05,
+      "loss": 0.3186,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2671363558032779e-05,
+      "loss": 0.2968,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.266261773600977e-05,
+      "loss": 0.2299,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.265386972141201e-05,
+      "loss": 0.2757,
+      "step": 1543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2645119521443183e-05,
+      "loss": 0.3024,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2636367143308787e-05,
+      "loss": 0.2758,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2627612594216105e-05,
+      "loss": 0.2594,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2618855881374208e-05,
+      "loss": 0.3351,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.261009701199395e-05,
+      "loss": 0.2963,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2601335993287958e-05,
+      "loss": 0.2848,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2592572832470637e-05,
+      "loss": 0.2216,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2583807536758149e-05,
+      "loss": 0.277,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2575040113368413e-05,
+      "loss": 0.3205,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2566270569521105e-05,
+      "loss": 0.2709,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2557498912437642e-05,
+      "loss": 0.3049,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2548725149341188e-05,
+      "loss": 0.3061,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2539949287456634e-05,
+      "loss": 0.3228,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2531171334010604e-05,
+      "loss": 0.2585,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2522391296231444e-05,
+      "loss": 0.2668,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2513609181349215e-05,
+      "loss": 0.3155,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2504824996595684e-05,
+      "loss": 0.2724,
+      "step": 1560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2496038749204334e-05,
+      "loss": 0.3214,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.248725044641033e-05,
+      "loss": 0.3195,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.247846009545055e-05,
+      "loss": 0.2917,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2469667703563543e-05,
+      "loss": 0.2484,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2460873277989543e-05,
+      "loss": 0.2492,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2452076825970458e-05,
+      "loss": 0.2792,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.244327835474987e-05,
+      "loss": 0.3046,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2434477871573014e-05,
+      "loss": 0.2732,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2425675383686791e-05,
+      "loss": 0.2749,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2416870898339748e-05,
+      "loss": 0.2611,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2408064422782078e-05,
+      "loss": 0.2711,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2399255964265613e-05,
+      "loss": 0.2642,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2390445530043817e-05,
+      "loss": 0.2599,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2381633127371784e-05,
+      "loss": 0.2721,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2372818763506222e-05,
+      "loss": 0.2713,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2364002445705463e-05,
+      "loss": 0.2825,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2355184181229441e-05,
+      "loss": 0.2973,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2346363977339698e-05,
+      "loss": 0.3332,
+      "step": 1578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2337541841299365e-05,
+      "loss": 0.284,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2328717780373178e-05,
+      "loss": 0.246,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2319891801827443e-05,
+      "loss": 0.276,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2311063912930054e-05,
+      "loss": 0.3179,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2302234120950478e-05,
+      "loss": 0.2925,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2293402433159744e-05,
+      "loss": 0.3001,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2284568856830447e-05,
+      "loss": 0.3454,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2275733399236734e-05,
+      "loss": 0.2932,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2266896067654305e-05,
+      "loss": 0.254,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.22580568693604e-05,
+      "loss": 0.2504,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2249215811633793e-05,
+      "loss": 0.2924,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2240372901754798e-05,
+      "loss": 0.2959,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2231528147005248e-05,
+      "loss": 0.2947,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2222681554668489e-05,
+      "loss": 0.2874,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2213833132029402e-05,
+      "loss": 0.3043,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2204982886374348e-05,
+      "loss": 0.2759,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.219613082499121e-05,
+      "loss": 0.2464,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2187276955169356e-05,
+      "loss": 0.2776,
+      "step": 1596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2178421284199641e-05,
+      "loss": 0.2831,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2169563819374413e-05,
+      "loss": 0.2888,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2160704567987492e-05,
+      "loss": 0.2998,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2151843537334163e-05,
+      "loss": 0.2779,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2142980734711185e-05,
+      "loss": 0.3429,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2134116167416777e-05,
+      "loss": 0.2449,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2125249842750604e-05,
+      "loss": 0.2677,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.211638176801378e-05,
+      "loss": 0.2575,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2107511950508866e-05,
+      "loss": 0.2679,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.209864039753985e-05,
+      "loss": 0.2953,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2089767116412151e-05,
+      "loss": 0.3132,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.208089211443262e-05,
+      "loss": 0.2908,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2072015398909511e-05,
+      "loss": 0.3197,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.20631369771525e-05,
+      "loss": 0.2241,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2054256856472664e-05,
+      "loss": 0.2603,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2045375044182477e-05,
+      "loss": 0.2778,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.203649154759581e-05,
+      "loss": 0.2437,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2027606374027916e-05,
+      "loss": 0.2755,
+      "step": 1614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2018719530795435e-05,
+      "loss": 0.2952,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2009831025216379e-05,
+      "loss": 0.3338,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2000940864610124e-05,
+      "loss": 0.2702,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1992049056297418e-05,
+      "loss": 0.2532,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1983155607600363e-05,
+      "loss": 0.253,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1974260525842408e-05,
+      "loss": 0.2731,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1965363818348344e-05,
+      "loss": 0.2538,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1956465492444316e-05,
+      "loss": 0.3082,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1947565555457781e-05,
+      "loss": 0.3153,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1938664014717538e-05,
+      "loss": 0.3333,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1929760877553707e-05,
+      "loss": 0.2853,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1920856151297705e-05,
+      "loss": 0.2697,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1911949843282283e-05,
+      "loss": 0.2425,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1903041960841473e-05,
+      "loss": 0.277,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1894132511310611e-05,
+      "loss": 0.3157,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1885221502026331e-05,
+      "loss": 0.293,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.187630894032654e-05,
+      "loss": 0.2979,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.186739483355043e-05,
+      "loss": 0.2703,
+      "step": 1632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1858479189038466e-05,
+      "loss": 0.2441,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1849562014132367e-05,
+      "loss": 0.2537,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1840643316175131e-05,
+      "loss": 0.2337,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1831723102511002e-05,
+      "loss": 0.3018,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1822801380485463e-05,
+      "loss": 0.2944,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1813878157445253e-05,
+      "loss": 0.2784,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1804953440738342e-05,
+      "loss": 0.2842,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1796027237713927e-05,
+      "loss": 0.2875,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1787099555722435e-05,
+      "loss": 0.2555,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1778170402115502e-05,
+      "loss": 0.2502,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1769239784245982e-05,
+      "loss": 0.2461,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1760307709467938e-05,
+      "loss": 0.3179,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1751374185136624e-05,
+      "loss": 0.283,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1742439218608492e-05,
+      "loss": 0.2603,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1733502817241185e-05,
+      "loss": 0.2843,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1724564988393523e-05,
+      "loss": 0.262,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1715625739425502e-05,
+      "loss": 0.264,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1706685077698288e-05,
+      "loss": 0.2242,
+      "step": 1650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1697743010574211e-05,
+      "loss": 0.2583,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.168879954541676e-05,
+      "loss": 0.3096,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1679854689590568e-05,
+      "loss": 0.295,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1670908450461422e-05,
+      "loss": 0.3059,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1661960835396244e-05,
+      "loss": 0.2781,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.165301185176309e-05,
+      "loss": 0.2797,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.164406150693114e-05,
+      "loss": 0.2681,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1635109808270696e-05,
+      "loss": 0.1965,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1626156763153184e-05,
+      "loss": 0.2743,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1617202378951124e-05,
+      "loss": 0.3067,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1608246663038142e-05,
+      "loss": 0.2664,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1599289622788976e-05,
+      "loss": 0.303,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1590331265579428e-05,
+      "loss": 0.2967,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1581371598786413e-05,
+      "loss": 0.2622,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1572410629787904e-05,
+      "loss": 0.2507,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1563448365962951e-05,
+      "loss": 0.2163,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1554484814691673e-05,
+      "loss": 0.2879,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1545519983355255e-05,
+      "loss": 0.3113,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.153655387933592e-05,
+      "loss": 0.223,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.152758651001695e-05,
+      "loss": 0.316,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1518617882782674e-05,
+      "loss": 0.271,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1509648005018445e-05,
+      "loss": 0.2676,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1500676884110652e-05,
+      "loss": 0.1956,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1491704527446706e-05,
+      "loss": 0.3107,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1482730942415037e-05,
+      "loss": 0.3471,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1473756136405086e-05,
+      "loss": 0.2409,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1464780116807296e-05,
+      "loss": 0.2875,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1455802891013119e-05,
+      "loss": 0.2979,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1446824466414987e-05,
+      "loss": 0.3221,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.143784485040633e-05,
+      "loss": 0.2523,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1428864050381553e-05,
+      "loss": 0.2501,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1419882073736042e-05,
+      "loss": 0.2912,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1410898927866142e-05,
+      "loss": 0.2721,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.140191462016917e-05,
+      "loss": 0.2783,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1392929158043397e-05,
+      "loss": 0.3207,
+      "step": 1685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1383942548888046e-05,
+      "loss": 0.2805,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1374954800103274e-05,
+      "loss": 0.2865,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1365965919090196e-05,
+      "loss": 0.2482,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1356975913250842e-05,
+      "loss": 0.2573,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1347984789988175e-05,
+      "loss": 0.3365,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1338992556706083e-05,
+      "loss": 0.2638,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1329999220809354e-05,
+      "loss": 0.2833,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.13210047897037e-05,
+      "loss": 0.2872,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.131200927079573e-05,
+      "loss": 0.3003,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1303012671492936e-05,
+      "loss": 0.2574,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.129401499920372e-05,
+      "loss": 0.2426,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1285016261337355e-05,
+      "loss": 0.2728,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1276016465303989e-05,
+      "loss": 0.308,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1267015618514656e-05,
+      "loss": 0.2286,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1258013728381241e-05,
+      "loss": 0.2956,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1249010802316492e-05,
+      "loss": 0.2937,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1240006847734012e-05,
+      "loss": 0.2906,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1231001872048248e-05,
+      "loss": 0.2149,
+      "step": 1703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1221995882674494e-05,
+      "loss": 0.2697,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1212988887028872e-05,
+      "loss": 0.2882,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1203980892528339e-05,
+      "loss": 0.256,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1194971906590666e-05,
+      "loss": 0.2427,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1185961936634446e-05,
+      "loss": 0.2962,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1176950990079083e-05,
+      "loss": 0.2665,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1167939074344782e-05,
+      "loss": 0.2373,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1158926196852551e-05,
+      "loss": 0.2418,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1149912365024182e-05,
+      "loss": 0.2511,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1140897586282263e-05,
+      "loss": 0.2946,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1131881868050151e-05,
+      "loss": 0.2416,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1122865217751988e-05,
+      "loss": 0.2659,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1113847642812669e-05,
+      "loss": 0.3176,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.110482915065787e-05,
+      "loss": 0.2976,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1095809748714002e-05,
+      "loss": 0.2429,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1086789444408235e-05,
+      "loss": 0.2621,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.107776824516849e-05,
+      "loss": 0.2629,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1068746158423407e-05,
+      "loss": 0.3109,
+      "step": 1721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1059723191602368e-05,
+      "loss": 0.2767,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1050699352135484e-05,
+      "loss": 0.305,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1041674647453569e-05,
+      "loss": 0.285,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1032649084988168e-05,
+      "loss": 0.2643,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.102362267217152e-05,
+      "loss": 0.2402,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1014595416436567e-05,
+      "loss": 0.2619,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1005567325216946e-05,
+      "loss": 0.2769,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0996538405946982e-05,
+      "loss": 0.2694,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.098750866606168e-05,
+      "loss": 0.2689,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0978478112996726e-05,
+      "loss": 0.2842,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0969446754188466e-05,
+      "loss": 0.2895,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0960414597073924e-05,
+      "loss": 0.2436,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0951381649090766e-05,
+      "loss": 0.231,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0942347917677317e-05,
+      "loss": 0.2711,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0933313410272547e-05,
+      "loss": 0.2575,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0924278134316067e-05,
+      "loss": 0.275,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0915242097248114e-05,
+      "loss": 0.303,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.090620530650956e-05,
+      "loss": 0.318,
+      "step": 1739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0897167769541888e-05,
+      "loss": 0.2466,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0888129493787205e-05,
+      "loss": 0.2298,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0879090486688221e-05,
+      "loss": 0.2142,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0870050755688253e-05,
+      "loss": 0.2995,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0861010308231206e-05,
+      "loss": 0.2688,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0851969151761581e-05,
+      "loss": 0.2744,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0842927293724461e-05,
+      "loss": 0.2854,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.083388474156551e-05,
+      "loss": 0.2722,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0824841502730958e-05,
+      "loss": 0.2405,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0815797584667605e-05,
+      "loss": 0.2355,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0806752994822807e-05,
+      "loss": 0.2477,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0797707740644475e-05,
+      "loss": 0.2564,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0788661829581069e-05,
+      "loss": 0.2473,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0779615269081583e-05,
+      "loss": 0.2702,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0770568066595557e-05,
+      "loss": 0.3105,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076152022957305e-05,
+      "loss": 0.2501,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0752471765464643e-05,
+      "loss": 0.2414,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0743422681721445e-05,
+      "loss": 0.2487,
+      "step": 1757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0734372985795062e-05,
+      "loss": 0.2587,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.072532268513761e-05,
+      "loss": 0.2393,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0716271787201704e-05,
+      "loss": 0.2784,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0707220299440446e-05,
+      "loss": 0.2623,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.069816822930743e-05,
+      "loss": 0.2652,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0689115584256729e-05,
+      "loss": 0.2771,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0680062371742878e-05,
+      "loss": 0.2227,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06710085992209e-05,
+      "loss": 0.2366,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.066195427414626e-05,
+      "loss": 0.2789,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0652899403974886e-05,
+      "loss": 0.2842,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0643843996163154e-05,
+      "loss": 0.2781,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0634788058167886e-05,
+      "loss": 0.2869,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0625731597446333e-05,
+      "loss": 0.2957,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.061667462145619e-05,
+      "loss": 0.2366,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0607617137655554e-05,
+      "loss": 0.2569,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0598559153502965e-05,
+      "loss": 0.2802,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0589500676457358e-05,
+      "loss": 0.2572,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0580441713978083e-05,
+      "loss": 0.2989,
+      "step": 1775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.057138227352488e-05,
+      "loss": 0.2882,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0562322362557895e-05,
+      "loss": 0.2802,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.055326198853765e-05,
+      "loss": 0.2752,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.054420115892506e-05,
+      "loss": 0.2367,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0535139881181404e-05,
+      "loss": 0.2381,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0526078162768334e-05,
+      "loss": 0.2545,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.051701601114787e-05,
+      "loss": 0.2783,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0507953433782379e-05,
+      "loss": 0.2826,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0498890438134587e-05,
+      "loss": 0.2835,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.048982703166756e-05,
+      "loss": 0.2156,
+      "step": 1785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0480763221844705e-05,
+      "loss": 0.2808,
+      "step": 1786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0471699016129761e-05,
+      "loss": 0.2399,
+      "step": 1787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0462634421986786e-05,
+      "loss": 0.2251,
+      "step": 1788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0453569446880171e-05,
+      "loss": 0.3004,
+      "step": 1789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0444504098274608e-05,
+      "loss": 0.2696,
+      "step": 1790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0435438383635105e-05,
+      "loss": 0.3102,
+      "step": 1791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.042637231042697e-05,
+      "loss": 0.2387,
+      "step": 1792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0417305886115796e-05,
+      "loss": 0.2498,
+      "step": 1793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0408239118167483e-05,
+      "loss": 0.2538,
+      "step": 1794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0399172014048203e-05,
+      "loss": 0.2442,
+      "step": 1795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0390104581224401e-05,
+      "loss": 0.2155,
+      "step": 1796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0381036827162802e-05,
+      "loss": 0.2691,
+      "step": 1797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0371968759330393e-05,
+      "loss": 0.3054,
+      "step": 1798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0362900385194416e-05,
+      "loss": 0.2816,
+      "step": 1799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0353831712222369e-05,
+      "loss": 0.2394,
+      "step": 1800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0344762747881993e-05,
+      "loss": 0.2737,
+      "step": 1801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0335693499641272e-05,
+      "loss": 0.2605,
+      "step": 1802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0326623974968419e-05,
+      "loss": 0.2473,
+      "step": 1803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0317554181331882e-05,
+      "loss": 0.2563,
+      "step": 1804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0308484126200322e-05,
+      "loss": 0.2883,
+      "step": 1805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0299413817042623e-05,
+      "loss": 0.2961,
+      "step": 1806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0290343261327873e-05,
+      "loss": 0.2441,
+      "step": 1807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0281272466525365e-05,
+      "loss": 0.3033,
+      "step": 1808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.027220144010459e-05,
+      "loss": 0.2545,
+      "step": 1809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0263130189535225e-05,
+      "loss": 0.2315,
+      "step": 1810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.025405872228714e-05,
+      "loss": 0.2448,
+      "step": 1811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0244987045830373e-05,
+      "loss": 0.2753,
+      "step": 1812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0235915167635143e-05,
+      "loss": 0.2958,
+      "step": 1813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0226843095171828e-05,
+      "loss": 0.269,
+      "step": 1814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0217770835910976e-05,
+      "loss": 0.2584,
+      "step": 1815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.020869839732328e-05,
+      "loss": 0.2684,
+      "step": 1816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0199625786879581e-05,
+      "loss": 0.2647,
+      "step": 1817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0190553012050868e-05,
+      "loss": 0.2606,
+      "step": 1818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0181480080308257e-05,
+      "loss": 0.2506,
+      "step": 1819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0172406999122998e-05,
+      "loss": 0.2688,
+      "step": 1820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0163333775966466e-05,
+      "loss": 0.3138,
+      "step": 1821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0154260418310148e-05,
+      "loss": 0.2164,
+      "step": 1822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0145186933625643e-05,
+      "loss": 0.2946,
+      "step": 1823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0136113329384662e-05,
+      "loss": 0.2788,
+      "step": 1824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0127039613059003e-05,
+      "loss": 0.2608,
+      "step": 1825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0117965792120558e-05,
+      "loss": 0.2518,
+      "step": 1826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0108891874041317e-05,
+      "loss": 0.2435,
+      "step": 1827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0099817866293336e-05,
+      "loss": 0.3033,
+      "step": 1828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.009074377634875e-05,
+      "loss": 0.3039,
+      "step": 1829
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0081669611679768e-05,
+      "loss": 0.2423,
+      "step": 1830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0072595379758647e-05,
+      "loss": 0.2653,
+      "step": 1831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0063521088057713e-05,
+      "loss": 0.2534,
+      "step": 1832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.005444674404933e-05,
+      "loss": 0.2142,
+      "step": 1833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0045372355205916e-05,
+      "loss": 0.211,
+      "step": 1834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0036297928999918e-05,
+      "loss": 0.2776,
+      "step": 1835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0027223472903811e-05,
+      "loss": 0.2802,
+      "step": 1836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0018148994390106e-05,
+      "loss": 0.2428,
+      "step": 1837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0009074500931324e-05,
+      "loss": 0.2865,
+      "step": 1838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.2965,
+      "step": 1839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.990925499068676e-06,
+      "loss": 0.2722,
+      "step": 1840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.981851005609899e-06,
+      "loss": 0.2251,
+      "step": 1841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.972776527096192e-06,
+      "loss": 0.2338,
+      "step": 1842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.963702071000085e-06,
+      "loss": 0.2542,
+      "step": 1843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.95462764479409e-06,
+      "loss": 0.3064,
+      "step": 1844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.945553255950671e-06,
+      "loss": 0.25,
+      "step": 1845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.936478911942292e-06,
+      "loss": 0.2798,
+      "step": 1846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.927404620241353e-06,
+      "loss": 0.2591,
+      "step": 1847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.918330388320235e-06,
+      "loss": 0.2583,
+      "step": 1848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.909256223651252e-06,
+      "loss": 0.2232,
+      "step": 1849
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.900182133706664e-06,
+      "loss": 0.2679,
+      "step": 1850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.891108125958686e-06,
+      "loss": 0.2675,
+      "step": 1851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.882034207879443e-06,
+      "loss": 0.2459,
+      "step": 1852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.872960386940999e-06,
+      "loss": 0.2411,
+      "step": 1853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.863886670615341e-06,
+      "loss": 0.2827,
+      "step": 1854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.854813066374359e-06,
+      "loss": 0.2576,
+      "step": 1855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.845739581689854e-06,
+      "loss": 0.2279,
+      "step": 1856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.836666224033539e-06,
+      "loss": 0.27,
+      "step": 1857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.827593000877006e-06,
+      "loss": 0.2704,
+      "step": 1858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.818519919691746e-06,
+      "loss": 0.3127,
+      "step": 1859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.809446987949137e-06,
+      "loss": 0.2449,
+      "step": 1860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.800374213120422e-06,
+      "loss": 0.2875,
+      "step": 1861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791301602676722e-06,
+      "loss": 0.276,
+      "step": 1862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.782229164089025e-06,
+      "loss": 0.2492,
+      "step": 1863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.773156904828177e-06,
+      "loss": 0.2408,
+      "step": 1864
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.764084832364862e-06,
+      "loss": 0.2454,
+      "step": 1865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.755012954169632e-06,
+      "loss": 0.2638,
+      "step": 1866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.745941277712867e-06,
+      "loss": 0.3212,
+      "step": 1867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.736869810464777e-06,
+      "loss": 0.2306,
+      "step": 1868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.727798559895414e-06,
+      "loss": 0.2656,
+      "step": 1869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.71872753347464e-06,
+      "loss": 0.2496,
+      "step": 1870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.70965673867213e-06,
+      "loss": 0.2045,
+      "step": 1871
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.700586182957378e-06,
+      "loss": 0.2379,
+      "step": 1872
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.691515873799681e-06,
+      "loss": 0.2471,
+      "step": 1873
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.68244581866812e-06,
+      "loss": 0.2658,
+      "step": 1874
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.673376025031583e-06,
+      "loss": 0.2847,
+      "step": 1875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.664306500358733e-06,
+      "loss": 0.2488,
+      "step": 1876
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.65523725211801e-06,
+      "loss": 0.2774,
+      "step": 1877
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.646168287777633e-06,
+      "loss": 0.2812,
+      "step": 1878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.637099614805583e-06,
+      "loss": 0.2489,
+      "step": 1879
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.628031240669609e-06,
+      "loss": 0.2332,
+      "step": 1880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.618963172837201e-06,
+      "loss": 0.2588,
+      "step": 1881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.6098954187756e-06,
+      "loss": 0.2843,
+      "step": 1882
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.600827985951802e-06,
+      "loss": 0.2451,
+      "step": 1883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.59176088183252e-06,
+      "loss": 0.2865,
+      "step": 1884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.582694113884204e-06,
+      "loss": 0.3084,
+      "step": 1885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.573627689573036e-06,
+      "loss": 0.2917,
+      "step": 1886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.564561616364898e-06,
+      "loss": 0.2181,
+      "step": 1887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.555495901725392e-06,
+      "loss": 0.2439,
+      "step": 1888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.546430553119834e-06,
+      "loss": 0.2675,
+      "step": 1889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.537365578013217e-06,
+      "loss": 0.25,
+      "step": 1890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.528300983870242e-06,
+      "loss": 0.3085,
+      "step": 1891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.519236778155298e-06,
+      "loss": 0.2414,
+      "step": 1892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.510172968332444e-06,
+      "loss": 0.2632,
+      "step": 1893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501109561865415e-06,
+      "loss": 0.2291,
+      "step": 1894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.492046566217624e-06,
+      "loss": 0.2293,
+      "step": 1895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.482983988852137e-06,
+      "loss": 0.2604,
+      "step": 1896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.473921837231668e-06,
+      "loss": 0.237,
+      "step": 1897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.464860118818599e-06,
+      "loss": 0.2693,
+      "step": 1898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.455798841074944e-06,
+      "loss": 0.2636,
+      "step": 1899
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.446738011462351e-06,
+      "loss": 0.2677,
+      "step": 1900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.437677637442108e-06,
+      "loss": 0.2446,
+      "step": 1901
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.428617726475124e-06,
+      "loss": 0.2423,
+      "step": 1902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.419558286021922e-06,
+      "loss": 0.2643,
+      "step": 1903
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.410499323542645e-06,
+      "loss": 0.2747,
+      "step": 1904
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.40144084649704e-06,
+      "loss": 0.2513,
+      "step": 1905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.392382862344448e-06,
+      "loss": 0.2681,
+      "step": 1906
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.383325378543815e-06,
+      "loss": 0.303,
+      "step": 1907
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.374268402553665e-06,
+      "loss": 0.2408,
+      "step": 1908
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.365211941832117e-06,
+      "loss": 0.2278,
+      "step": 1909
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.356156003836849e-06,
+      "loss": 0.2417,
+      "step": 1910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.347100596025115e-06,
+      "loss": 0.2417,
+      "step": 1911
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.338045725853743e-06,
+      "loss": 0.2603,
+      "step": 1912
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.328991400779104e-06,
+      "loss": 0.2673,
+      "step": 1913
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.31993762825712e-06,
+      "loss": 0.2536,
+      "step": 1914
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.310884415743275e-06,
+      "loss": 0.2688,
+      "step": 1915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.301831770692573e-06,
+      "loss": 0.2424,
+      "step": 1916
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.292779700559556e-06,
+      "loss": 0.2365,
+      "step": 1917
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.283728212798302e-06,
+      "loss": 0.2267,
+      "step": 1918
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.274677314862396e-06,
+      "loss": 0.2311,
+      "step": 1919
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.265627014204941e-06,
+      "loss": 0.2369,
+      "step": 1920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.256577318278558e-06,
+      "loss": 0.2867,
+      "step": 1921
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.24752823453536e-06,
+      "loss": 0.2828,
+      "step": 1922
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.238479770426953e-06,
+      "loss": 0.2355,
+      "step": 1923
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.229431933404446e-06,
+      "loss": 0.2584,
+      "step": 1924
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.220384730918422e-06,
+      "loss": 0.2452,
+      "step": 1925
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.211338170418936e-06,
+      "loss": 0.2201,
+      "step": 1926
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.202292259355528e-06,
+      "loss": 0.2893,
+      "step": 1927
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.1932470051772e-06,
+      "loss": 0.2585,
+      "step": 1928
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.184202415332399e-06,
+      "loss": 0.2928,
+      "step": 1929
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.175158497269045e-06,
+      "loss": 0.2395,
+      "step": 1930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.166115258434495e-06,
+      "loss": 0.2355,
+      "step": 1931
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.157072706275542e-06,
+      "loss": 0.2605,
+      "step": 1932
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.148030848238422e-06,
+      "loss": 0.2116,
+      "step": 1933
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.138989691768801e-06,
+      "loss": 0.2521,
+      "step": 1934
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12994924431175e-06,
+      "loss": 0.2283,
+      "step": 1935
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12090951331178e-06,
+      "loss": 0.2865,
+      "step": 1936
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.1118705062128e-06,
+      "loss": 0.278,
+      "step": 1937
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.102832230458115e-06,
+      "loss": 0.2674,
+      "step": 1938
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.093794693490444e-06,
+      "loss": 0.2516,
+      "step": 1939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084757902751886e-06,
+      "loss": 0.2318,
+      "step": 1940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.075721865683936e-06,
+      "loss": 0.2367,
+      "step": 1941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.066686589727456e-06,
+      "loss": 0.2622,
+      "step": 1942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.057652082322685e-06,
+      "loss": 0.2435,
+      "step": 1943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.048618350909239e-06,
+      "loss": 0.303,
+      "step": 1944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.039585402926081e-06,
+      "loss": 0.2761,
+      "step": 1945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.030553245811534e-06,
+      "loss": 0.2327,
+      "step": 1946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.021521887003279e-06,
+      "loss": 0.2392,
+      "step": 1947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.012491333938324e-06,
+      "loss": 0.2431,
+      "step": 1948
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00346159405302e-06,
+      "loss": 0.2412,
+      "step": 1949
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.994432674783059e-06,
+      "loss": 0.2234,
+      "step": 1950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.985404583563437e-06,
+      "loss": 0.2571,
+      "step": 1951
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.976377327828481e-06,
+      "loss": 0.285,
+      "step": 1952
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.967350915011832e-06,
+      "loss": 0.2827,
+      "step": 1953
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.958325352546433e-06,
+      "loss": 0.2219,
+      "step": 1954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.94930064786452e-06,
+      "loss": 0.2299,
+      "step": 1955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.940276808397634e-06,
+      "loss": 0.2494,
+      "step": 1956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.9312538415766e-06,
+      "loss": 0.2279,
+      "step": 1957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.922231754831514e-06,
+      "loss": 0.2556,
+      "step": 1958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.913210555591765e-06,
+      "loss": 0.2713,
+      "step": 1959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.904190251286003e-06,
+      "loss": 0.3029,
+      "step": 1960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.895170849342134e-06,
+      "loss": 0.2331,
+      "step": 1961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.886152357187333e-06,
+      "loss": 0.2469,
+      "step": 1962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.877134782248018e-06,
+      "loss": 0.2226,
+      "step": 1963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.86811813194985e-06,
+      "loss": 0.2297,
+      "step": 1964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.859102413717739e-06,
+      "loss": 0.2222,
+      "step": 1965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.850087634975821e-06,
+      "loss": 0.2364,
+      "step": 1966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.841073803147452e-06,
+      "loss": 0.2735,
+      "step": 1967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.83206092565522e-06,
+      "loss": 0.2489,
+      "step": 1968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.823049009920918e-06,
+      "loss": 0.2417,
+      "step": 1969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.814038063365557e-06,
+      "loss": 0.2559,
+      "step": 1970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.805028093409338e-06,
+      "loss": 0.2255,
+      "step": 1971
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.796019107471661e-06,
+      "loss": 0.2213,
+      "step": 1972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.78701111297113e-06,
+      "loss": 0.2462,
+      "step": 1973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.778004117325509e-06,
+      "loss": 0.2695,
+      "step": 1974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.768998127951752e-06,
+      "loss": 0.3024,
+      "step": 1975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.759993152265992e-06,
+      "loss": 0.217,
+      "step": 1976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.750989197683512e-06,
+      "loss": 0.2575,
+      "step": 1977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.74198627161876e-06,
+      "loss": 0.2464,
+      "step": 1978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.732984381485346e-06,
+      "loss": 0.2033,
+      "step": 1979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.723983534696012e-06,
+      "loss": 0.2134,
+      "step": 1980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.714983738662648e-06,
+      "loss": 0.2212,
+      "step": 1981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.705985000796284e-06,
+      "loss": 0.2772,
+      "step": 1982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.696987328507067e-06,
+      "loss": 0.254,
+      "step": 1983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.687990729204274e-06,
+      "loss": 0.2672,
+      "step": 1984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.678995210296301e-06,
+      "loss": 0.2425,
+      "step": 1985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.67000077919065e-06,
+      "loss": 0.228,
+      "step": 1986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.66100744329392e-06,
+      "loss": 0.2247,
+      "step": 1987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.652015210011827e-06,
+      "loss": 0.2311,
+      "step": 1988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.643024086749163e-06,
+      "loss": 0.2671,
+      "step": 1989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.634034080909807e-06,
+      "loss": 0.2419,
+      "step": 1990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.625045199896727e-06,
+      "loss": 0.2377,
+      "step": 1991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.61605745111196e-06,
+      "loss": 0.2744,
+      "step": 1992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.607070841956605e-06,
+      "loss": 0.2608,
+      "step": 1993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.598085379830831e-06,
+      "loss": 0.237,
+      "step": 1994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.589101072133863e-06,
+      "loss": 0.2214,
+      "step": 1995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.580117926263961e-06,
+      "loss": 0.2689,
+      "step": 1996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.571135949618448e-06,
+      "loss": 0.231,
+      "step": 1997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.562155149593673e-06,
+      "loss": 0.2327,
+      "step": 1998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.553175533585015e-06,
+      "loss": 0.2386,
+      "step": 1999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.544197108986884e-06,
+      "loss": 0.2854,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.535219883192704e-06,
+      "loss": 0.2259,
+      "step": 2001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.526243863594917e-06,
+      "loss": 0.2239,
+      "step": 2002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.517269057584967e-06,
+      "loss": 0.2286,
+      "step": 2003
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.508295472553294e-06,
+      "loss": 0.268,
+      "step": 2004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.49932311588935e-06,
+      "loss": 0.2588,
+      "step": 2005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.490351994981558e-06,
+      "loss": 0.205,
+      "step": 2006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.481382117217326e-06,
+      "loss": 0.3053,
+      "step": 2007
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.472413489983052e-06,
+      "loss": 0.2512,
+      "step": 2008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.463446120664085e-06,
+      "loss": 0.2223,
+      "step": 2009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.454480016644748e-06,
+      "loss": 0.2065,
+      "step": 2010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.445515185308329e-06,
+      "loss": 0.2396,
+      "step": 2011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.436551634037054e-06,
+      "loss": 0.2798,
+      "step": 2012
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.4275893702121e-06,
+      "loss": 0.2195,
+      "step": 2013
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.418628401213589e-06,
+      "loss": 0.2715,
+      "step": 2014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.409668734420573e-06,
+      "loss": 0.2348,
+      "step": 2015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.400710377211027e-06,
+      "loss": 0.2649,
+      "step": 2016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.39175333696186e-06,
+      "loss": 0.2211,
+      "step": 2017
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.38279762104888e-06,
+      "loss": 0.2047,
+      "step": 2018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.373843236846818e-06,
+      "loss": 0.2595,
+      "step": 2019
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.364890191729304e-06,
+      "loss": 0.2809,
+      "step": 2020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.355938493068864e-06,
+      "loss": 0.1913,
+      "step": 2021
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.346988148236912e-06,
+      "loss": 0.2777,
+      "step": 2022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.338039164603757e-06,
+      "loss": 0.238,
+      "step": 2023
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.329091549538581e-06,
+      "loss": 0.2322,
+      "step": 2024
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.320145310409435e-06,
+      "loss": 0.2003,
+      "step": 2025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.311200454583244e-06,
+      "loss": 0.1902,
+      "step": 2026
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.302256989425794e-06,
+      "loss": 0.2328,
+      "step": 2027
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.293314922301715e-06,
+      "loss": 0.2522,
+      "step": 2028
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.284374260574501e-06,
+      "loss": 0.2753,
+      "step": 2029
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.275435011606479e-06,
+      "loss": 0.2471,
+      "step": 2030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.266497182758818e-06,
+      "loss": 0.2647,
+      "step": 2031
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.25756078139151e-06,
+      "loss": 0.1872,
+      "step": 2032
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.248625814863377e-06,
+      "loss": 0.1923,
+      "step": 2033
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.239692290532067e-06,
+      "loss": 0.2419,
+      "step": 2034
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.230760215754021e-06,
+      "loss": 0.2305,
+      "step": 2035
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.2218295978845e-06,
+      "loss": 0.2146,
+      "step": 2036
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.21290044427757e-06,
+      "loss": 0.2787,
+      "step": 2037
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.203972762286074e-06,
+      "loss": 0.2351,
+      "step": 2038
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.195046559261658e-06,
+      "loss": 0.2299,
+      "step": 2039
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.186121842554749e-06,
+      "loss": 0.1994,
+      "step": 2040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.17719861951454e-06,
+      "loss": 0.2329,
+      "step": 2041
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.168276897489001e-06,
+      "loss": 0.2029,
+      "step": 2042
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.159356683824874e-06,
+      "loss": 0.2532,
+      "step": 2043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.150437985867636e-06,
+      "loss": 0.2139,
+      "step": 2044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.141520810961539e-06,
+      "loss": 0.2659,
+      "step": 2045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.132605166449571e-06,
+      "loss": 0.2506,
+      "step": 2046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.123691059673463e-06,
+      "loss": 0.2005,
+      "step": 2047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11477849797367e-06,
+      "loss": 0.2356,
+      "step": 2048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.10586748868939e-06,
+      "loss": 0.2413,
+      "step": 2049
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.096958039158532e-06,
+      "loss": 0.2349,
+      "step": 2050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.088050156717719e-06,
+      "loss": 0.2341,
+      "step": 2051
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.079143848702296e-06,
+      "loss": 0.2674,
+      "step": 2052
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.070239122446298e-06,
+      "loss": 0.2622,
+      "step": 2053
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.061335985282464e-06,
+      "loss": 0.2267,
+      "step": 2054
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.05243444454222e-06,
+      "loss": 0.2128,
+      "step": 2055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.043534507555691e-06,
+      "loss": 0.221,
+      "step": 2056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.03463618165166e-06,
+      "loss": 0.2207,
+      "step": 2057
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.025739474157595e-06,
+      "loss": 0.2311,
+      "step": 2058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.01684439239964e-06,
+      "loss": 0.193,
+      "step": 2059
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.007950943702585e-06,
+      "loss": 0.263,
+      "step": 2060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.999059135389877e-06,
+      "loss": 0.2435,
+      "step": 2061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.990168974783623e-06,
+      "loss": 0.2195,
+      "step": 2062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.98128046920457e-06,
+      "loss": 0.2157,
+      "step": 2063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.972393625972086e-06,
+      "loss": 0.2214,
+      "step": 2064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.963508452404191e-06,
+      "loss": 0.259,
+      "step": 2065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.954624955817528e-06,
+      "loss": 0.2092,
+      "step": 2066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.945743143527338e-06,
+      "loss": 0.2179,
+      "step": 2067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.936863022847501e-06,
+      "loss": 0.2974,
+      "step": 2068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.927984601090492e-06,
+      "loss": 0.2275,
+      "step": 2069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.919107885567382e-06,
+      "loss": 0.2142,
+      "step": 2070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91023288358785e-06,
+      "loss": 0.2333,
+      "step": 2071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.901359602460157e-06,
+      "loss": 0.2114,
+      "step": 2072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.892488049491139e-06,
+      "loss": 0.2794,
+      "step": 2073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.883618231986223e-06,
+      "loss": 0.2388,
+      "step": 2074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.874750157249398e-06,
+      "loss": 0.2443,
+      "step": 2075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.865883832583225e-06,
+      "loss": 0.2461,
+      "step": 2076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.857019265288816e-06,
+      "loss": 0.2037,
+      "step": 2077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.848156462665839e-06,
+      "loss": 0.259,
+      "step": 2078
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.839295432012513e-06,
+      "loss": 0.2233,
+      "step": 2079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.83043618062559e-06,
+      "loss": 0.2385,
+      "step": 2080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.82157871580036e-06,
+      "loss": 0.2129,
+      "step": 2081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.812723044830649e-06,
+      "loss": 0.2516,
+      "step": 2082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.803869175008794e-06,
+      "loss": 0.2431,
+      "step": 2083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.795017113625652e-06,
+      "loss": 0.239,
+      "step": 2084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.786166867970603e-06,
+      "loss": 0.1844,
+      "step": 2085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.777318445331513e-06,
+      "loss": 0.2102,
+      "step": 2086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.768471852994757e-06,
+      "loss": 0.1969,
+      "step": 2087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.759627098245207e-06,
+      "loss": 0.2227,
+      "step": 2088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.750784188366212e-06,
+      "loss": 0.2313,
+      "step": 2089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.741943130639603e-06,
+      "loss": 0.2577,
+      "step": 2090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.733103932345696e-06,
+      "loss": 0.2812,
+      "step": 2091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.72426660076327e-06,
+      "loss": 0.2177,
+      "step": 2092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.715431143169555e-06,
+      "loss": 0.1998,
+      "step": 2093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.706597566840257e-06,
+      "loss": 0.2285,
+      "step": 2094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.697765879049528e-06,
+      "loss": 0.2202,
+      "step": 2095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.688936087069947e-06,
+      "loss": 0.2547,
+      "step": 2096
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.68010819817256e-06,
+      "loss": 0.2368,
+      "step": 2097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.671282219626827e-06,
+      "loss": 0.2476,
+      "step": 2098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.662458158700637e-06,
+      "loss": 0.2465,
+      "step": 2099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.653636022660306e-06,
+      "loss": 0.2037,
+      "step": 2100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.644815818770564e-06,
+      "loss": 0.1982,
+      "step": 2101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.63599755429454e-06,
+      "loss": 0.2143,
+      "step": 2102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.62718123649378e-06,
+      "loss": 0.24,
+      "step": 2103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.618366872628222e-06,
+      "loss": 0.2055,
+      "step": 2104
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.609554469956185e-06,
+      "loss": 0.2522,
+      "step": 2105
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.60074403573439e-06,
+      "loss": 0.2549,
+      "step": 2106
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.591935577217921e-06,
+      "loss": 0.2126,
+      "step": 2107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5831291016602544e-06,
+      "loss": 0.221,
+      "step": 2108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.574324616313211e-06,
+      "loss": 0.2191,
+      "step": 2109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.565522128426986e-06,
+      "loss": 0.2074,
+      "step": 2110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5567216452501336e-06,
+      "loss": 0.1985,
+      "step": 2111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.547923174029544e-06,
+      "loss": 0.2276,
+      "step": 2112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.539126722010458e-06,
+      "loss": 0.2251,
+      "step": 2113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.53033229643646e-06,
+      "loss": 0.2694,
+      "step": 2114
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.521539904549453e-06,
+      "loss": 0.2077,
+      "step": 2115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.51274955358967e-06,
+      "loss": 0.2185,
+      "step": 2116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.503961250795673e-06,
+      "loss": 0.2153,
+      "step": 2117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.49517500340432e-06,
+      "loss": 0.2026,
+      "step": 2118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4863908186507885e-06,
+      "loss": 0.2462,
+      "step": 2119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.477608703768557e-06,
+      "loss": 0.2315,
+      "step": 2120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.468828665989399e-06,
+      "loss": 0.2373,
+      "step": 2121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.460050712543367e-06,
+      "loss": 0.1919,
+      "step": 2122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.451274850658815e-06,
+      "loss": 0.2275,
+      "step": 2123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.442501087562363e-06,
+      "loss": 0.2126,
+      "step": 2124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4337294304789e-06,
+      "loss": 0.1912,
+      "step": 2125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4249598866315906e-06,
+      "loss": 0.2349,
+      "step": 2126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.416192463241857e-06,
+      "loss": 0.2466,
+      "step": 2127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.407427167529366e-06,
+      "loss": 0.2602,
+      "step": 2128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.398664006712044e-06,
+      "loss": 0.2026,
+      "step": 2129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.389902988006057e-06,
+      "loss": 0.2416,
+      "step": 2130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.381144118625796e-06,
+      "loss": 0.1945,
+      "step": 2131
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.372387405783898e-06,
+      "loss": 0.2216,
+      "step": 2132
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.363632856691218e-06,
+      "loss": 0.196,
+      "step": 2133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3548804785568185e-06,
+      "loss": 0.202,
+      "step": 2134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.346130278587995e-06,
+      "loss": 0.2555,
+      "step": 2135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.337382263990231e-06,
+      "loss": 0.2141,
+      "step": 2136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.328636441967223e-06,
+      "loss": 0.1857,
+      "step": 2137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.319892819720853e-06,
+      "loss": 0.2157,
+      "step": 2138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.311151404451196e-06,
+      "loss": 0.2164,
+      "step": 2139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.30241220335651e-06,
+      "loss": 0.186,
+      "step": 2140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.293675223633227e-06,
+      "loss": 0.2008,
+      "step": 2141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.284940472475952e-06,
+      "loss": 0.2658,
+      "step": 2142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.276207957077457e-06,
+      "loss": 0.2441,
+      "step": 2143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2674776846286656e-06,
+      "loss": 0.2521,
+      "step": 2144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2587496623186585e-06,
+      "loss": 0.2424,
+      "step": 2145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.250023897334669e-06,
+      "loss": 0.2216,
+      "step": 2146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.241300396862063e-06,
+      "loss": 0.2075,
+      "step": 2147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.232579168084344e-06,
+      "loss": 0.1822,
+      "step": 2148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.223860218183152e-06,
+      "loss": 0.2268,
+      "step": 2149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.215143554338237e-06,
+      "loss": 0.2386,
+      "step": 2150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.206429183727477e-06,
+      "loss": 0.2481,
+      "step": 2151
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.197717113526861e-06,
+      "loss": 0.1831,
+      "step": 2152
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.18900735091048e-06,
+      "loss": 0.2308,
+      "step": 2153
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.18029990305052e-06,
+      "loss": 0.2265,
+      "step": 2154
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.171594777117275e-06,
+      "loss": 0.2089,
+      "step": 2155
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.162891980279118e-06,
+      "loss": 0.1855,
+      "step": 2156
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.154191519702498e-06,
+      "loss": 0.2493,
+      "step": 2157
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1454934025519555e-06,
+      "loss": 0.2358,
+      "step": 2158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.136797635990089e-06,
+      "loss": 0.2056,
+      "step": 2159
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.128104227177562e-06,
+      "loss": 0.2738,
+      "step": 2160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.119413183273104e-06,
+      "loss": 0.2182,
+      "step": 2161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1107245114334946e-06,
+      "loss": 0.2286,
+      "step": 2162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.102038218813551e-06,
+      "loss": 0.172,
+      "step": 2163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.093354312566145e-06,
+      "loss": 0.2201,
+      "step": 2164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.084672799842176e-06,
+      "loss": 0.2646,
+      "step": 2165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.07599368779057e-06,
+      "loss": 0.2238,
+      "step": 2166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0673169835582815e-06,
+      "loss": 0.1828,
+      "step": 2167
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.058642694290278e-06,
+      "loss": 0.2307,
+      "step": 2168
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.049970827129546e-06,
+      "loss": 0.2213,
+      "step": 2169
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.041301389217065e-06,
+      "loss": 0.2101,
+      "step": 2170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.032634387691823e-06,
+      "loss": 0.2084,
+      "step": 2171
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.023969829690804e-06,
+      "loss": 0.2353,
+      "step": 2172
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.015307722348972e-06,
+      "loss": 0.2343,
+      "step": 2173
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.006648072799273e-06,
+      "loss": 0.2058,
+      "step": 2174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.997990888172641e-06,
+      "loss": 0.204,
+      "step": 2175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9893361755979635e-06,
+      "loss": 0.243,
+      "step": 2176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.980683942202104e-06,
+      "loss": 0.213,
+      "step": 2177
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.972034195109885e-06,
+      "loss": 0.2002,
+      "step": 2178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.963386941444072e-06,
+      "loss": 0.2102,
+      "step": 2179
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.954742188325383e-06,
+      "loss": 0.2242,
+      "step": 2180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9460999428724815e-06,
+      "loss": 0.253,
+      "step": 2181
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.937460212201961e-06,
+      "loss": 0.1698,
+      "step": 2182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.928823003428339e-06,
+      "loss": 0.2201,
+      "step": 2183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.920188323664067e-06,
+      "loss": 0.2448,
+      "step": 2184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.91155618001951e-06,
+      "loss": 0.1987,
+      "step": 2185
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.902926579602935e-06,
+      "loss": 0.1899,
+      "step": 2186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.894299529520533e-06,
+      "loss": 0.211,
+      "step": 2187
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.885675036876382e-06,
+      "loss": 0.2287,
+      "step": 2188
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.877053108772451e-06,
+      "loss": 0.2221,
+      "step": 2189
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.868433752308612e-06,
+      "loss": 0.2002,
+      "step": 2190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8598169745826095e-06,
+      "loss": 0.2472,
+      "step": 2191
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.851202782690057e-06,
+      "loss": 0.2278,
+      "step": 2192
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8425911837244565e-06,
+      "loss": 0.1859,
+      "step": 2193
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8339821847771645e-06,
+      "loss": 0.2063,
+      "step": 2194
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.825375792937392e-06,
+      "loss": 0.2115,
+      "step": 2195
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.816772015292212e-06,
+      "loss": 0.2356,
+      "step": 2196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.808170858926542e-06,
+      "loss": 0.197,
+      "step": 2197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.799572330923143e-06,
+      "loss": 0.2208,
+      "step": 2198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.790976438362604e-06,
+      "loss": 0.2258,
+      "step": 2199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.782383188323351e-06,
+      "loss": 0.2039,
+      "step": 2200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.773792587881633e-06,
+      "loss": 0.1898,
+      "step": 2201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.765204644111516e-06,
+      "loss": 0.1966,
+      "step": 2202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.756619364084873e-06,
+      "loss": 0.2333,
+      "step": 2203
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.748036754871398e-06,
+      "loss": 0.2012,
+      "step": 2204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.739456823538571e-06,
+      "loss": 0.2253,
+      "step": 2205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.73087957715167e-06,
+      "loss": 0.2156,
+      "step": 2206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.722305022773772e-06,
+      "loss": 0.1934,
+      "step": 2207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.713733167465723e-06,
+      "loss": 0.196,
+      "step": 2208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.705164018286152e-06,
+      "loss": 0.1804,
+      "step": 2209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.696597582291469e-06,
+      "loss": 0.2201,
+      "step": 2210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.688033866535834e-06,
+      "loss": 0.2193,
+      "step": 2211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.679472878071174e-06,
+      "loss": 0.1898,
+      "step": 2212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.670914623947175e-06,
+      "loss": 0.224,
+      "step": 2213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.662359111211268e-06,
+      "loss": 0.2543,
+      "step": 2214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.653806346908619e-06,
+      "loss": 0.1746,
+      "step": 2215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.645256338082142e-06,
+      "loss": 0.1947,
+      "step": 2216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.636709091772478e-06,
+      "loss": 0.2056,
+      "step": 2217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.628164615017987e-06,
+      "loss": 0.1843,
+      "step": 2218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.61962291485476e-06,
+      "loss": 0.2242,
+      "step": 2219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.611083998316592e-06,
+      "loss": 0.1925,
+      "step": 2220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.602547872434988e-06,
+      "loss": 0.2676,
+      "step": 2221
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.594014544239158e-06,
+      "loss": 0.2027,
+      "step": 2222
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.585484020756009e-06,
+      "loss": 0.2077,
+      "step": 2223
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.576956309010128e-06,
+      "loss": 0.2054,
+      "step": 2224
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.568431416023802e-06,
+      "loss": 0.2108,
+      "step": 2225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.559909348816986e-06,
+      "loss": 0.2269,
+      "step": 2226
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.55139011440731e-06,
+      "loss": 0.2176,
+      "step": 2227
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542873719810074e-06,
+      "loss": 0.2174,
+      "step": 2228
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.534360172038236e-06,
+      "loss": 0.245,
+      "step": 2229
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5258494781024185e-06,
+      "loss": 0.2138,
+      "step": 2230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.51734164501088e-06,
+      "loss": 0.2082,
+      "step": 2231
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.50883667976953e-06,
+      "loss": 0.2124,
+      "step": 2232
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.500334589381926e-06,
+      "loss": 0.2179,
+      "step": 2233
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.491835380849239e-06,
+      "loss": 0.2016,
+      "step": 2234
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.483339061170279e-06,
+      "loss": 0.2304,
+      "step": 2235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.474845637341481e-06,
+      "loss": 0.2275,
+      "step": 2236
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.466355116356884e-06,
+      "loss": 0.2324,
+      "step": 2237
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4578675052081395e-06,
+      "loss": 0.1899,
+      "step": 2238
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.449382810884512e-06,
+      "loss": 0.1951,
+      "step": 2239
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4409010403728554e-06,
+      "loss": 0.2036,
+      "step": 2240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.432422200657613e-06,
+      "loss": 0.2084,
+      "step": 2241
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4239462987208255e-06,
+      "loss": 0.2082,
+      "step": 2242
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.415473341542109e-06,
+      "loss": 0.2037,
+      "step": 2243
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.40700333609865e-06,
+      "loss": 0.1943,
+      "step": 2244
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.398536289365209e-06,
+      "loss": 0.2102,
+      "step": 2245
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.390072208314116e-06,
+      "loss": 0.186,
+      "step": 2246
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.381611099915243e-06,
+      "loss": 0.1948,
+      "step": 2247
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.373152971136032e-06,
+      "loss": 0.1876,
+      "step": 2248
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3646978289414614e-06,
+      "loss": 0.2012,
+      "step": 2249
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.356245680294047e-06,
+      "loss": 0.2429,
+      "step": 2250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.347796532153849e-06,
+      "loss": 0.221,
+      "step": 2251
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.339350391478456e-06,
+      "loss": 0.2232,
+      "step": 2252
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3309072652229656e-06,
+      "loss": 0.1866,
+      "step": 2253
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.322467160340011e-06,
+      "loss": 0.1962,
+      "step": 2254
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.314030083779732e-06,
+      "loss": 0.2286,
+      "step": 2255
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.305596042489768e-06,
+      "loss": 0.202,
+      "step": 2256
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.2971650434152655e-06,
+      "loss": 0.2352,
+      "step": 2257
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.288737093498864e-06,
+      "loss": 0.2329,
+      "step": 2258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.280312199680696e-06,
+      "loss": 0.2246,
+      "step": 2259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.27189036889837e-06,
+      "loss": 0.2045,
+      "step": 2260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.263471608086975e-06,
+      "loss": 0.2211,
+      "step": 2261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.255055924179079e-06,
+      "loss": 0.2194,
+      "step": 2262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.246643324104704e-06,
+      "loss": 0.1904,
+      "step": 2263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.238233814791338e-06,
+      "loss": 0.2097,
+      "step": 2264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.229827403163935e-06,
+      "loss": 0.2296,
+      "step": 2265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.221424096144873e-06,
+      "loss": 0.2133,
+      "step": 2266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.213023900653996e-06,
+      "loss": 0.2201,
+      "step": 2267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.204626823608584e-06,
+      "loss": 0.1818,
+      "step": 2268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.19623287192333e-06,
+      "loss": 0.22,
+      "step": 2269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.187842052510373e-06,
+      "loss": 0.1868,
+      "step": 2270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.179454372279274e-06,
+      "loss": 0.1828,
+      "step": 2271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.171069838136987e-06,
+      "loss": 0.1952,
+      "step": 2272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.162688456987895e-06,
+      "loss": 0.2496,
+      "step": 2273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.154310235733785e-06,
+      "loss": 0.249,
+      "step": 2274
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.14593518127383e-06,
+      "loss": 0.227,
+      "step": 2275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.137563300504602e-06,
+      "loss": 0.2126,
+      "step": 2276
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.129194600320056e-06,
+      "loss": 0.1827,
+      "step": 2277
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.120829087611539e-06,
+      "loss": 0.1971,
+      "step": 2278
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1124667692677585e-06,
+      "loss": 0.2055,
+      "step": 2279
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.104107652174796e-06,
+      "loss": 0.213,
+      "step": 2280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.095751743216107e-06,
+      "loss": 0.264,
+      "step": 2281
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.087399049272491e-06,
+      "loss": 0.2167,
+      "step": 2282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.079049577222105e-06,
+      "loss": 0.2085,
+      "step": 2283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.070703333940462e-06,
+      "loss": 0.1981,
+      "step": 2284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.062360326300399e-06,
+      "loss": 0.1762,
+      "step": 2285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0540205611721005e-06,
+      "loss": 0.1895,
+      "step": 2286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0456840454230794e-06,
+      "loss": 0.2137,
+      "step": 2287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.037350785918177e-06,
+      "loss": 0.2272,
+      "step": 2288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.029020789519536e-06,
+      "loss": 0.2188,
+      "step": 2289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.020694063086634e-06,
+      "loss": 0.1963,
+      "step": 2290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.012370613476243e-06,
+      "loss": 0.1884,
+      "step": 2291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.004050447542436e-06,
+      "loss": 0.1929,
+      "step": 2292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9957335721365886e-06,
+      "loss": 0.1889,
+      "step": 2293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.987419994107364e-06,
+      "loss": 0.2029,
+      "step": 2294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.979109720300707e-06,
+      "loss": 0.1981,
+      "step": 2295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.970802757559846e-06,
+      "loss": 0.2253,
+      "step": 2296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.962499112725284e-06,
+      "loss": 0.2234,
+      "step": 2297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.954198792634782e-06,
+      "loss": 0.1961,
+      "step": 2298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.945901804123375e-06,
+      "loss": 0.2238,
+      "step": 2299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.937608154023352e-06,
+      "loss": 0.1831,
+      "step": 2300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.929317849164242e-06,
+      "loss": 0.1793,
+      "step": 2301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.921030896372836e-06,
+      "loss": 0.2061,
+      "step": 2302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.912747302473151e-06,
+      "loss": 0.2146,
+      "step": 2303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.904467074286451e-06,
+      "loss": 0.2248,
+      "step": 2304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.896190218631213e-06,
+      "loss": 0.2143,
+      "step": 2305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8879167423231456e-06,
+      "loss": 0.2038,
+      "step": 2306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8796466521751815e-06,
+      "loss": 0.2048,
+      "step": 2307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.87137995499745e-06,
+      "loss": 0.1893,
+      "step": 2308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.863116657597294e-06,
+      "loss": 0.1663,
+      "step": 2309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.854856766779264e-06,
+      "loss": 0.2068,
+      "step": 2310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.84660028934509e-06,
+      "loss": 0.2421,
+      "step": 2311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8383472320936995e-06,
+      "loss": 0.2313,
+      "step": 2312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.830097601821211e-06,
+      "loss": 0.1555,
+      "step": 2313
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.821851405320903e-06,
+      "loss": 0.2396,
+      "step": 2314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.813608649383245e-06,
+      "loss": 0.2239,
+      "step": 2315
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.805369340795863e-06,
+      "loss": 0.1728,
+      "step": 2316
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.79713348634354e-06,
+      "loss": 0.201,
+      "step": 2317
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.788901092808226e-06,
+      "loss": 0.2055,
+      "step": 2318
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.780672166969015e-06,
+      "loss": 0.2088,
+      "step": 2319
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.772446715602149e-06,
+      "loss": 0.2061,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.764224745481006e-06,
+      "loss": 0.254,
+      "step": 2321
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.75600626337609e-06,
+      "loss": 0.2102,
+      "step": 2322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.74779127605505e-06,
+      "loss": 0.204,
+      "step": 2323
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.739579790282641e-06,
+      "loss": 0.1826,
+      "step": 2324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.731371812820742e-06,
+      "loss": 0.1975,
+      "step": 2325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.723167350428347e-06,
+      "loss": 0.2048,
+      "step": 2326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.714966409861547e-06,
+      "loss": 0.2179,
+      "step": 2327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.706768997873533e-06,
+      "loss": 0.1945,
+      "step": 2328
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6985751212146e-06,
+      "loss": 0.2168,
+      "step": 2329
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.690384786632118e-06,
+      "loss": 0.2117,
+      "step": 2330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.682198000870553e-06,
+      "loss": 0.1682,
+      "step": 2331
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6740147706714475e-06,
+      "loss": 0.1852,
+      "step": 2332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.665835102773406e-06,
+      "loss": 0.2202,
+      "step": 2333
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.657659003912103e-06,
+      "loss": 0.2116,
+      "step": 2334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.649486480820281e-06,
+      "loss": 0.2043,
+      "step": 2335
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.641317540227738e-06,
+      "loss": 0.1925,
+      "step": 2336
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.633152188861306e-06,
+      "loss": 0.2279,
+      "step": 2337
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.624990433444886e-06,
+      "loss": 0.1892,
+      "step": 2338
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.616832280699397e-06,
+      "loss": 0.1762,
+      "step": 2339
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.608677737342797e-06,
+      "loss": 0.1979,
+      "step": 2340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6005268100900765e-06,
+      "loss": 0.204,
+      "step": 2341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.592379505653252e-06,
+      "loss": 0.2543,
+      "step": 2342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.584235830741341e-06,
+      "loss": 0.1903,
+      "step": 2343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.576095792060387e-06,
+      "loss": 0.2605,
+      "step": 2344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.567959396313434e-06,
+      "loss": 0.1949,
+      "step": 2345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.559826650200519e-06,
+      "loss": 0.1981,
+      "step": 2346
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.551697560418688e-06,
+      "loss": 0.1904,
+      "step": 2347
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.543572133661965e-06,
+      "loss": 0.2101,
+      "step": 2348
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.535450376621368e-06,
+      "loss": 0.2281,
+      "step": 2349
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.527332295984882e-06,
+      "loss": 0.1983,
+      "step": 2350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.519217898437464e-06,
+      "loss": 0.2058,
+      "step": 2351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.511107190661051e-06,
+      "loss": 0.2044,
+      "step": 2352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.503000179334528e-06,
+      "loss": 0.2048,
+      "step": 2353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.494896871133744e-06,
+      "loss": 0.1866,
+      "step": 2354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.486797272731501e-06,
+      "loss": 0.2074,
+      "step": 2355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4787013907975365e-06,
+      "loss": 0.2222,
+      "step": 2356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4706092319985315e-06,
+      "loss": 0.2023,
+      "step": 2357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.462520802998108e-06,
+      "loss": 0.1784,
+      "step": 2358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.454436110456805e-06,
+      "loss": 0.231,
+      "step": 2359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.446355161032091e-06,
+      "loss": 0.2162,
+      "step": 2360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4382779613783595e-06,
+      "loss": 0.2226,
+      "step": 2361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.430204518146902e-06,
+      "loss": 0.2028,
+      "step": 2362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.422134837985922e-06,
+      "loss": 0.2064,
+      "step": 2363
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.414068927540528e-06,
+      "loss": 0.2199,
+      "step": 2364
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.406006793452725e-06,
+      "loss": 0.2388,
+      "step": 2365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.397948442361396e-06,
+      "loss": 0.2342,
+      "step": 2366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3898938809023295e-06,
+      "loss": 0.2418,
+      "step": 2367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.381843115708175e-06,
+      "loss": 0.2141,
+      "step": 2368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.37379615340846e-06,
+      "loss": 0.1829,
+      "step": 2369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.365753000629585e-06,
+      "loss": 0.1871,
+      "step": 2370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.357713663994818e-06,
+      "loss": 0.2232,
+      "step": 2371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3496781501242686e-06,
+      "loss": 0.1913,
+      "step": 2372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.341646465634917e-06,
+      "loss": 0.2151,
+      "step": 2373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.333618617140577e-06,
+      "loss": 0.2176,
+      "step": 2374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.325594611251901e-06,
+      "loss": 0.1985,
+      "step": 2375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.317574454576391e-06,
+      "loss": 0.2071,
+      "step": 2376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.309558153718375e-06,
+      "loss": 0.1682,
+      "step": 2377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.301545715278995e-06,
+      "loss": 0.1924,
+      "step": 2378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.293537145856229e-06,
+      "loss": 0.2058,
+      "step": 2379
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2855324520448504e-06,
+      "loss": 0.1862,
+      "step": 2380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2775316404364594e-06,
+      "loss": 0.2307,
+      "step": 2381
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.269534717619447e-06,
+      "loss": 0.214,
+      "step": 2382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.261541690179004e-06,
+      "loss": 0.1972,
+      "step": 2383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.253552564697123e-06,
+      "loss": 0.1602,
+      "step": 2384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2455673477525715e-06,
+      "loss": 0.204,
+      "step": 2385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.237586045920899e-06,
+      "loss": 0.2082,
+      "step": 2386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2296086657744425e-06,
+      "loss": 0.1796,
+      "step": 2387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.221635213882295e-06,
+      "loss": 0.2184,
+      "step": 2388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.213665696810324e-06,
+      "loss": 0.2019,
+      "step": 2389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.20570012112116e-06,
+      "loss": 0.2512,
+      "step": 2390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.197738493374177e-06,
+      "loss": 0.1739,
+      "step": 2391
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.189780820125501e-06,
+      "loss": 0.1761,
+      "step": 2392
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.181827107928012e-06,
+      "loss": 0.1898,
+      "step": 2393
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.173877363331311e-06,
+      "loss": 0.1808,
+      "step": 2394
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.165931592881748e-06,
+      "loss": 0.1857,
+      "step": 2395
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.157989803122393e-06,
+      "loss": 0.2292,
+      "step": 2396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.150052000593038e-06,
+      "loss": 0.2254,
+      "step": 2397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.142118191830188e-06,
+      "loss": 0.2046,
+      "step": 2398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.134188383367067e-06,
+      "loss": 0.173,
+      "step": 2399
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.126262581733605e-06,
+      "loss": 0.1741,
+      "step": 2400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.11834079345642e-06,
+      "loss": 0.1883,
+      "step": 2401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.110423025058845e-06,
+      "loss": 0.209,
+      "step": 2402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.102509283060886e-06,
+      "loss": 0.2125,
+      "step": 2403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.094599573979234e-06,
+      "loss": 0.2238,
+      "step": 2404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.08669390432727e-06,
+      "loss": 0.2178,
+      "step": 2405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.078792280615046e-06,
+      "loss": 0.1924,
+      "step": 2406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0708947093492725e-06,
+      "loss": 0.1962,
+      "step": 2407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.063001197033337e-06,
+      "loss": 0.1817,
+      "step": 2408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0551117501672676e-06,
+      "loss": 0.1848,
+      "step": 2409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.047226375247765e-06,
+      "loss": 0.1975,
+      "step": 2410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.039345078768156e-06,
+      "loss": 0.2009,
+      "step": 2411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.031467867218425e-06,
+      "loss": 0.2269,
+      "step": 2412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.023594747085189e-06,
+      "loss": 0.2125,
+      "step": 2413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.01572572485169e-06,
+      "loss": 0.1964,
+      "step": 2414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.007860806997795e-06,
+      "loss": 0.1921,
+      "step": 2415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.168,
+      "step": 2416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9921433103314124e-06,
+      "loss": 0.2104,
+      "step": 2417
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9842907444617415e-06,
+      "loss": 0.1932,
+      "step": 2418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.976442308857317e-06,
+      "loss": 0.2198,
+      "step": 2419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9685980099810505e-06,
+      "loss": 0.1929,
+      "step": 2420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9607578542924515e-06,
+      "loss": 0.2074,
+      "step": 2421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9529218482476285e-06,
+      "loss": 0.1906,
+      "step": 2422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.945089998299257e-06,
+      "loss": 0.1881,
+      "step": 2423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.937262310896601e-06,
+      "loss": 0.1792,
+      "step": 2424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.929438792485498e-06,
+      "loss": 0.1992,
+      "step": 2425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9216194495083444e-06,
+      "loss": 0.2084,
+      "step": 2426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.913804288404098e-06,
+      "loss": 0.2237,
+      "step": 2427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.905993315608282e-06,
+      "loss": 0.2273,
+      "step": 2428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8981865375529665e-06,
+      "loss": 0.1956,
+      "step": 2429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.890383960666763e-06,
+      "loss": 0.1746,
+      "step": 2430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.882585591374832e-06,
+      "loss": 0.1803,
+      "step": 2431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.87479143609886e-06,
+      "loss": 0.1802,
+      "step": 2432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.867001501257066e-06,
+      "loss": 0.2155,
+      "step": 2433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.859215793264198e-06,
+      "loss": 0.2111,
+      "step": 2434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.851434318531524e-06,
+      "loss": 0.2044,
+      "step": 2435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.843657083466814e-06,
+      "loss": 0.2018,
+      "step": 2436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.835884094474365e-06,
+      "loss": 0.1961,
+      "step": 2437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8281153579549645e-06,
+      "loss": 0.1788,
+      "step": 2438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.820350880305896e-06,
+      "loss": 0.1835,
+      "step": 2439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8125906679209455e-06,
+      "loss": 0.1649,
+      "step": 2440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.804834727190385e-06,
+      "loss": 0.1965,
+      "step": 2441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.797083064500968e-06,
+      "loss": 0.2088,
+      "step": 2442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.789335686235921e-06,
+      "loss": 0.2287,
+      "step": 2443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7815925987749425e-06,
+      "loss": 0.1873,
+      "step": 2444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.773853808494208e-06,
+      "loss": 0.1807,
+      "step": 2445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.766119321766337e-06,
+      "loss": 0.1844,
+      "step": 2446
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7583891449604215e-06,
+      "loss": 0.1725,
+      "step": 2447
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.750663284442001e-06,
+      "loss": 0.179,
+      "step": 2448
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.742941746573055e-06,
+      "loss": 0.1985,
+      "step": 2449
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7352245377119985e-06,
+      "loss": 0.166,
+      "step": 2450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.727511664213702e-06,
+      "loss": 0.2269,
+      "step": 2451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.719803132429443e-06,
+      "loss": 0.1931,
+      "step": 2452
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.712098948706939e-06,
+      "loss": 0.1795,
+      "step": 2453
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.704399119390325e-06,
+      "loss": 0.17,
+      "step": 2454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.696703650820144e-06,
+      "loss": 0.1819,
+      "step": 2455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.689012549333347e-06,
+      "loss": 0.2345,
+      "step": 2456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.681325821263299e-06,
+      "loss": 0.1889,
+      "step": 2457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.673643472939761e-06,
+      "loss": 0.2308,
+      "step": 2458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6659655106888744e-06,
+      "loss": 0.1994,
+      "step": 2459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.658291940833191e-06,
+      "loss": 0.2156,
+      "step": 2460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.650622769691625e-06,
+      "loss": 0.2029,
+      "step": 2461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.642958003579476e-06,
+      "loss": 0.1845,
+      "step": 2462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635297648808421e-06,
+      "loss": 0.1801,
+      "step": 2463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.627641711686502e-06,
+      "loss": 0.1951,
+      "step": 2464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.619990198518117e-06,
+      "loss": 0.1897,
+      "step": 2465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.612343115604031e-06,
+      "loss": 0.2073,
+      "step": 2466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.604700469241354e-06,
+      "loss": 0.1843,
+      "step": 2467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.597062265723537e-06,
+      "loss": 0.1918,
+      "step": 2468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5894285113403855e-06,
+      "loss": 0.1562,
+      "step": 2469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.581799212378037e-06,
+      "loss": 0.1631,
+      "step": 2470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.574174375118959e-06,
+      "loss": 0.1877,
+      "step": 2471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.566554005841944e-06,
+      "loss": 0.2102,
+      "step": 2472
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.558938110822101e-06,
+      "loss": 0.1634,
+      "step": 2473
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.551326696330866e-06,
+      "loss": 0.198,
+      "step": 2474
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.543719768635971e-06,
+      "loss": 0.1997,
+      "step": 2475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536117334001466e-06,
+      "loss": 0.2109,
+      "step": 2476
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5285193986877e-06,
+      "loss": 0.1565,
+      "step": 2477
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.52092596895131e-06,
+      "loss": 0.1793,
+      "step": 2478
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.51333705104522e-06,
+      "loss": 0.2358,
+      "step": 2479
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.505752651218654e-06,
+      "loss": 0.199,
+      "step": 2480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.498172775717099e-06,
+      "loss": 0.1935,
+      "step": 2481
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4905974307823264e-06,
+      "loss": 0.2298,
+      "step": 2482
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.483026622652381e-06,
+      "loss": 0.2166,
+      "step": 2483
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.47546035756155e-06,
+      "loss": 0.1975,
+      "step": 2484
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.467898641740402e-06,
+      "loss": 0.1566,
+      "step": 2485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.460341481415753e-06,
+      "loss": 0.1756,
+      "step": 2486
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.452788882810668e-06,
+      "loss": 0.2241,
+      "step": 2487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.445240852144448e-06,
+      "loss": 0.1895,
+      "step": 2488
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.437697395632645e-06,
+      "loss": 0.1974,
+      "step": 2489
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.430158519487034e-06,
+      "loss": 0.2013,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.422624229915618e-06,
+      "loss": 0.1907,
+      "step": 2491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4150945331226324e-06,
+      "loss": 0.178,
+      "step": 2492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4075694353085285e-06,
+      "loss": 0.1605,
+      "step": 2493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4000489426699625e-06,
+      "loss": 0.1976,
+      "step": 2494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.392533061399798e-06,
+      "loss": 0.2143,
+      "step": 2495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.385021797687119e-06,
+      "loss": 0.187,
+      "step": 2496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.377515157717183e-06,
+      "loss": 0.216,
+      "step": 2497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.370013147671456e-06,
+      "loss": 0.1724,
+      "step": 2498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.362515773727592e-06,
+      "loss": 0.1746,
+      "step": 2499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.355023042059416e-06,
+      "loss": 0.1621,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.347534958836937e-06,
+      "loss": 0.1823,
+      "step": 2501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.340051530226336e-06,
+      "loss": 0.2143,
+      "step": 2502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3325727623899646e-06,
+      "loss": 0.2083,
+      "step": 2503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.325098661486328e-06,
+      "loss": 0.1615,
+      "step": 2504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3176292336700975e-06,
+      "loss": 0.1993,
+      "step": 2505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310164485092091e-06,
+      "loss": 0.182,
+      "step": 2506
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302704421899266e-06,
+      "loss": 0.1619,
+      "step": 2507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.295249050234738e-06,
+      "loss": 0.1504,
+      "step": 2508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.287798376237752e-06,
+      "loss": 0.2169,
+      "step": 2509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2803524060436765e-06,
+      "loss": 0.1721,
+      "step": 2510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.272911145784021e-06,
+      "loss": 0.1662,
+      "step": 2511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.265474601586406e-06,
+      "loss": 0.2154,
+      "step": 2512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.258042779574567e-06,
+      "loss": 0.1858,
+      "step": 2513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.25061568586836e-06,
+      "loss": 0.1959,
+      "step": 2514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.243193326583743e-06,
+      "loss": 0.1564,
+      "step": 2515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2357757078327785e-06,
+      "loss": 0.174,
+      "step": 2516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.228362835723619e-06,
+      "loss": 0.1988,
+      "step": 2517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.220954716360507e-06,
+      "loss": 0.1751,
+      "step": 2518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213551355843785e-06,
+      "loss": 0.1885,
+      "step": 2519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.206152760269857e-06,
+      "loss": 0.2292,
+      "step": 2520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.198758935731221e-06,
+      "loss": 0.1944,
+      "step": 2521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.191369888316439e-06,
+      "loss": 0.1798,
+      "step": 2522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1839856241101385e-06,
+      "loss": 0.1584,
+      "step": 2523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.176606149193003e-06,
+      "loss": 0.171,
+      "step": 2524
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.169231469641787e-06,
+      "loss": 0.1938,
+      "step": 2525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1618615915292796e-06,
+      "loss": 0.1537,
+      "step": 2526
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1544965209243246e-06,
+      "loss": 0.1858,
+      "step": 2527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.147136263891814e-06,
+      "loss": 0.2044,
+      "step": 2528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.139780826492662e-06,
+      "loss": 0.1855,
+      "step": 2529
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1324302147838165e-06,
+      "loss": 0.1432,
+      "step": 2530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.125084434818261e-06,
+      "loss": 0.1739,
+      "step": 2531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.117743492644993e-06,
+      "loss": 0.1927,
+      "step": 2532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.110407394309026e-06,
+      "loss": 0.1848,
+      "step": 2533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.10307614585139e-06,
+      "loss": 0.1524,
+      "step": 2534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.095749753309115e-06,
+      "loss": 0.1954,
+      "step": 2535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.088428222715232e-06,
+      "loss": 0.1877,
+      "step": 2536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0811115600987736e-06,
+      "loss": 0.1867,
+      "step": 2537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.073799771484768e-06,
+      "loss": 0.1538,
+      "step": 2538
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.066492862894212e-06,
+      "loss": 0.1601,
+      "step": 2539
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.059190840344106e-06,
+      "loss": 0.2049,
+      "step": 2540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.051893709847411e-06,
+      "loss": 0.1448,
+      "step": 2541
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.044601477413062e-06,
+      "loss": 0.1731,
+      "step": 2542
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.037314149045964e-06,
+      "loss": 0.2258,
+      "step": 2543
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.030031730746991e-06,
+      "loss": 0.1753,
+      "step": 2544
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.022754228512955e-06,
+      "loss": 0.1445,
+      "step": 2545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.015481648336639e-06,
+      "loss": 0.1574,
+      "step": 2546
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.008213996206758e-06,
+      "loss": 0.1802,
+      "step": 2547
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.000951278107981e-06,
+      "loss": 0.1789,
+      "step": 2548
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.993693500020901e-06,
+      "loss": 0.1878,
+      "step": 2549
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.986440667922055e-06,
+      "loss": 0.1706,
+      "step": 2550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.979192787783906e-06,
+      "loss": 0.2303,
+      "step": 2551
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.971949865574832e-06,
+      "loss": 0.1708,
+      "step": 2552
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.964711907259128e-06,
+      "loss": 0.1558,
+      "step": 2553
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.957478918797013e-06,
+      "loss": 0.1642,
+      "step": 2554
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.950250906144599e-06,
+      "loss": 0.1931,
+      "step": 2555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.94302787525391e-06,
+      "loss": 0.1872,
+      "step": 2556
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.93580983207287e-06,
+      "loss": 0.1846,
+      "step": 2557
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.928596782545286e-06,
+      "loss": 0.2044,
+      "step": 2558
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.921388732610856e-06,
+      "loss": 0.1918,
+      "step": 2559
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9141856882051695e-06,
+      "loss": 0.1782,
+      "step": 2560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.90698765525968e-06,
+      "loss": 0.1662,
+      "step": 2561
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.899794639701725e-06,
+      "loss": 0.1655,
+      "step": 2562
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.892606647454511e-06,
+      "loss": 0.1782,
+      "step": 2563
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.885423684437102e-06,
+      "loss": 0.1682,
+      "step": 2564
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.878245756564417e-06,
+      "loss": 0.1823,
+      "step": 2565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.871072869747239e-06,
+      "loss": 0.1903,
+      "step": 2566
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8639050298921975e-06,
+      "loss": 0.1916,
+      "step": 2567
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8567422429017585e-06,
+      "loss": 0.1829,
+      "step": 2568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.84958451467424e-06,
+      "loss": 0.1634,
+      "step": 2569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.84243185110378e-06,
+      "loss": 0.1799,
+      "step": 2570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.835284258080352e-06,
+      "loss": 0.1647,
+      "step": 2571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.828141741489757e-06,
+      "loss": 0.2128,
+      "step": 2572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8210043072136175e-06,
+      "loss": 0.1985,
+      "step": 2573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8138719611293596e-06,
+      "loss": 0.1992,
+      "step": 2574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8067447091102326e-06,
+      "loss": 0.1618,
+      "step": 2575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7996225570252797e-06,
+      "loss": 0.1631,
+      "step": 2576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.792505510739356e-06,
+      "loss": 0.1598,
+      "step": 2577
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7853935761131e-06,
+      "loss": 0.1837,
+      "step": 2578
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.778286759002949e-06,
+      "loss": 0.1571,
+      "step": 2579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7711850652611283e-06,
+      "loss": 0.2062,
+      "step": 2580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7640885007356366e-06,
+      "loss": 0.181,
+      "step": 2581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7569970712702487e-06,
+      "loss": 0.1769,
+      "step": 2582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7499107827045224e-06,
+      "loss": 0.1519,
+      "step": 2583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7428296408737652e-06,
+      "loss": 0.1547,
+      "step": 2584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.735753651609062e-06,
+      "loss": 0.1871,
+      "step": 2585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7286828207372485e-06,
+      "loss": 0.1608,
+      "step": 2586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7216171540809122e-06,
+      "loss": 0.1852,
+      "step": 2587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7145566574583814e-06,
+      "loss": 0.1719,
+      "step": 2588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.707501336683743e-06,
+      "loss": 0.1837,
+      "step": 2589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7004511975668055e-06,
+      "loss": 0.1588,
+      "step": 2590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6934062459131216e-06,
+      "loss": 0.1583,
+      "step": 2591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.68636648752397e-06,
+      "loss": 0.1847,
+      "step": 2592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6793319281963493e-06,
+      "loss": 0.1656,
+      "step": 2593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6723025737229747e-06,
+      "loss": 0.1755,
+      "step": 2594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6652784298922826e-06,
+      "loss": 0.1905,
+      "step": 2595
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6582595024884204e-06,
+      "loss": 0.1927,
+      "step": 2596
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6512457972912263e-06,
+      "loss": 0.1826,
+      "step": 2597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.644237320076256e-06,
+      "loss": 0.1733,
+      "step": 2598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.637234076614745e-06,
+      "loss": 0.1712,
+      "step": 2599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.630236072673624e-06,
+      "loss": 0.149,
+      "step": 2600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.623243314015513e-06,
+      "loss": 0.1606,
+      "step": 2601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.616255806398713e-06,
+      "loss": 0.2121,
+      "step": 2602
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.609273555577191e-06,
+      "loss": 0.1824,
+      "step": 2603
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.602296567300598e-06,
+      "loss": 0.1673,
+      "step": 2604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5953248473142434e-06,
+      "loss": 0.1806,
+      "step": 2605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5883584013590967e-06,
+      "loss": 0.2003,
+      "step": 2606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5813972351717918e-06,
+      "loss": 0.1579,
+      "step": 2607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5744413544846103e-06,
+      "loss": 0.1615,
+      "step": 2608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5674907650254876e-06,
+      "loss": 0.1739,
+      "step": 2609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.560545472517991e-06,
+      "loss": 0.175,
+      "step": 2610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5536054826813295e-06,
+      "loss": 0.1729,
+      "step": 2611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5466708012303554e-06,
+      "loss": 0.1582,
+      "step": 2612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.539741433875533e-06,
+      "loss": 0.1793,
+      "step": 2613
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5328173863229644e-06,
+      "loss": 0.1568,
+      "step": 2614
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.52589866427437e-06,
+      "loss": 0.1614,
+      "step": 2615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.518985273427077e-06,
+      "loss": 0.1602,
+      "step": 2616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5120772194740228e-06,
+      "loss": 0.1804,
+      "step": 2617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5051745081037605e-06,
+      "loss": 0.1552,
+      "step": 2618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4982771450004316e-06,
+      "loss": 0.1605,
+      "step": 2619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.491385135843782e-06,
+      "loss": 0.1789,
+      "step": 2620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4844984863091493e-06,
+      "loss": 0.1598,
+      "step": 2621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.477617202067449e-06,
+      "loss": 0.16,
+      "step": 2622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4707412887851834e-06,
+      "loss": 0.1682,
+      "step": 2623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.463870752124433e-06,
+      "loss": 0.175,
+      "step": 2624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4570055977428542e-06,
+      "loss": 0.1956,
+      "step": 2625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.450145831293661e-06,
+      "loss": 0.1883,
+      "step": 2626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.443291458425643e-06,
+      "loss": 0.1516,
+      "step": 2627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.436442484783138e-06,
+      "loss": 0.1823,
+      "step": 2628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4295989160060394e-06,
+      "loss": 0.1627,
+      "step": 2629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4227607577297963e-06,
+      "loss": 0.1528,
+      "step": 2630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.415928015585399e-06,
+      "loss": 0.1746,
+      "step": 2631
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4091006951993744e-06,
+      "loss": 0.1784,
+      "step": 2632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4022788021937914e-06,
+      "loss": 0.2001,
+      "step": 2633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3954623421862442e-06,
+      "loss": 0.1519,
+      "step": 2634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.388651320789851e-06,
+      "loss": 0.1888,
+      "step": 2635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3818457436132578e-06,
+      "loss": 0.1945,
+      "step": 2636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.375045616260626e-06,
+      "loss": 0.175,
+      "step": 2637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3682509443316303e-06,
+      "loss": 0.1656,
+      "step": 2638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3614617334214496e-06,
+      "loss": 0.1544,
+      "step": 2639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3546779891207614e-06,
+      "loss": 0.1915,
+      "step": 2640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3478997170157555e-06,
+      "loss": 0.1666,
+      "step": 2641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3411269226881005e-06,
+      "loss": 0.1464,
+      "step": 2642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3343596117149648e-06,
+      "loss": 0.1789,
+      "step": 2643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3275977896690015e-06,
+      "loss": 0.1741,
+      "step": 2644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.320841462118335e-06,
+      "loss": 0.1537,
+      "step": 2645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3140906346265688e-06,
+      "loss": 0.1393,
+      "step": 2646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3073453127527846e-06,
+      "loss": 0.178,
+      "step": 2647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3006055020515183e-06,
+      "loss": 0.1972,
+      "step": 2648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2938712080727755e-06,
+      "loss": 0.159,
+      "step": 2649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.287142436362023e-06,
+      "loss": 0.1577,
+      "step": 2650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.280419192460169e-06,
+      "loss": 0.2057,
+      "step": 2651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.273701481903574e-06,
+      "loss": 0.1839,
+      "step": 2652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2669893102240447e-06,
+      "loss": 0.1563,
+      "step": 2653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.26028268294883e-06,
+      "loss": 0.1328,
+      "step": 2654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2535816056006e-06,
+      "loss": 0.1867,
+      "step": 2655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2468860836974713e-06,
+      "loss": 0.202,
+      "step": 2656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.240196122752972e-06,
+      "loss": 0.1452,
+      "step": 2657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2335117282760563e-06,
+      "loss": 0.1902,
+      "step": 2658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2268329057710943e-06,
+      "loss": 0.1791,
+      "step": 2659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.220159660737875e-06,
+      "loss": 0.1724,
+      "step": 2660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2134919986715796e-06,
+      "loss": 0.1412,
+      "step": 2661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2068299250628053e-06,
+      "loss": 0.163,
+      "step": 2662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.200173445397541e-06,
+      "loss": 0.2013,
+      "step": 2663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1935225651571666e-06,
+      "loss": 0.1545,
+      "step": 2664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.186877289818456e-06,
+      "loss": 0.1478,
+      "step": 2665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1802376248535726e-06,
+      "loss": 0.1943,
+      "step": 2666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1736035757300466e-06,
+      "loss": 0.1812,
+      "step": 2667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1669751479107955e-06,
+      "loss": 0.1563,
+      "step": 2668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1603523468540985e-06,
+      "loss": 0.1671,
+      "step": 2669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1537351780136117e-06,
+      "loss": 0.1574,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1471236468383415e-06,
+      "loss": 0.195,
+      "step": 2671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.140517758772661e-06,
+      "loss": 0.1388,
+      "step": 2672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1339175192562964e-06,
+      "loss": 0.1656,
+      "step": 2673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.127322933724317e-06,
+      "loss": 0.1965,
+      "step": 2674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1207340076071344e-06,
+      "loss": 0.1811,
+      "step": 2675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.114150746330512e-06,
+      "loss": 0.1501,
+      "step": 2676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.107573155315533e-06,
+      "loss": 0.1691,
+      "step": 2677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.101001239978623e-06,
+      "loss": 0.1617,
+      "step": 2678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.094435005731532e-06,
+      "loss": 0.1852,
+      "step": 2679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0878744579813267e-06,
+      "loss": 0.148,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0813196021303916e-06,
+      "loss": 0.2053,
+      "step": 2681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.074770443576429e-06,
+      "loss": 0.1729,
+      "step": 2682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0682269877124516e-06,
+      "loss": 0.1604,
+      "step": 2683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.061689239926764e-06,
+      "loss": 0.1437,
+      "step": 2684
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.055157205602984e-06,
+      "loss": 0.1766,
+      "step": 2685
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0486308901200177e-06,
+      "loss": 0.1747,
+      "step": 2686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.042110298852059e-06,
+      "loss": 0.1545,
+      "step": 2687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0355954371685948e-06,
+      "loss": 0.1795,
+      "step": 2688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0290863104343937e-06,
+      "loss": 0.1793,
+      "step": 2689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.022582924009495e-06,
+      "loss": 0.1825,
+      "step": 2690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0160852832492194e-06,
+      "loss": 0.1326,
+      "step": 2691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.00959339350415e-06,
+      "loss": 0.1639,
+      "step": 2692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0031072601201327e-06,
+      "loss": 0.1577,
+      "step": 2693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.996626888438281e-06,
+      "loss": 0.1807,
+      "step": 2694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9901522837949616e-06,
+      "loss": 0.1508,
+      "step": 2695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.983683451521784e-06,
+      "loss": 0.1835,
+      "step": 2696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.977220396945617e-06,
+      "loss": 0.1654,
+      "step": 2697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9707631253885594e-06,
+      "loss": 0.1485,
+      "step": 2698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9643116421679597e-06,
+      "loss": 0.1318,
+      "step": 2699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.957865952596388e-06,
+      "loss": 0.1399,
+      "step": 2700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9514260619816525e-06,
+      "loss": 0.1544,
+      "step": 2701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9449919756267895e-06,
+      "loss": 0.1817,
+      "step": 2702
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.938563698830037e-06,
+      "loss": 0.1647,
+      "step": 2703
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.932141236884867e-06,
+      "loss": 0.174,
+      "step": 2704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9257245950799605e-06,
+      "loss": 0.1691,
+      "step": 2705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9193137786991964e-06,
+      "loss": 0.1465,
+      "step": 2706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.912908793021668e-06,
+      "loss": 0.1432,
+      "step": 2707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9065096433216667e-06,
+      "loss": 0.1429,
+      "step": 2708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.900116334868662e-06,
+      "loss": 0.1627,
+      "step": 2709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8937288729273304e-06,
+      "loss": 0.1809,
+      "step": 2710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.887347262757533e-06,
+      "loss": 0.1486,
+      "step": 2711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8809715096143e-06,
+      "loss": 0.1731,
+      "step": 2712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8746016187478544e-06,
+      "loss": 0.1734,
+      "step": 2713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8682375954035756e-06,
+      "loss": 0.1416,
+      "step": 2714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8618794448220265e-06,
+      "loss": 0.1611,
+      "step": 2715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8555271722389212e-06,
+      "loss": 0.1417,
+      "step": 2716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8491807828851424e-06,
+      "loss": 0.1584,
+      "step": 2717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.842840281986726e-06,
+      "loss": 0.1886,
+      "step": 2718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.836505674764858e-06,
+      "loss": 0.1566,
+      "step": 2719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.830176966435866e-06,
+      "loss": 0.2058,
+      "step": 2720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8238541622112303e-06,
+      "loss": 0.1446,
+      "step": 2721
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.817537267297561e-06,
+      "loss": 0.1535,
+      "step": 2722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8112262868966057e-06,
+      "loss": 0.1398,
+      "step": 2723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.804921226205245e-06,
+      "loss": 0.1569,
+      "step": 2724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7986220904154795e-06,
+      "loss": 0.1708,
+      "step": 2725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.792328884714427e-06,
+      "loss": 0.1583,
+      "step": 2726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7860416142843338e-06,
+      "loss": 0.1566,
+      "step": 2727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7797602843025483e-06,
+      "loss": 0.1823,
+      "step": 2728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.773484899941532e-06,
+      "loss": 0.1308,
+      "step": 2729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7672154663688544e-06,
+      "loss": 0.1547,
+      "step": 2730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7609519887471748e-06,
+      "loss": 0.1575,
+      "step": 2731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7546944722342515e-06,
+      "loss": 0.1639,
+      "step": 2732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.748442921982939e-06,
+      "loss": 0.1812,
+      "step": 2733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7421973431411775e-06,
+      "loss": 0.153,
+      "step": 2734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.735957740851981e-06,
+      "loss": 0.1544,
+      "step": 2735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7297241202534574e-06,
+      "loss": 0.1675,
+      "step": 2736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.723496486478775e-06,
+      "loss": 0.1396,
+      "step": 2737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.717274844656175e-06,
+      "loss": 0.1576,
+      "step": 2738
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7110591999089698e-06,
+      "loss": 0.1542,
+      "step": 2739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7048495573555345e-06,
+      "loss": 0.1739,
+      "step": 2740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6986459221092908e-06,
+      "loss": 0.1738,
+      "step": 2741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.692448299278727e-06,
+      "loss": 0.1705,
+      "step": 2742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6862566939673675e-06,
+      "loss": 0.1585,
+      "step": 2743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.680071111273793e-06,
+      "loss": 0.1743,
+      "step": 2744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.673891556291616e-06,
+      "loss": 0.159,
+      "step": 2745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6677180341094886e-06,
+      "loss": 0.1413,
+      "step": 2746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6615505498111007e-06,
+      "loss": 0.1722,
+      "step": 2747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6553891084751604e-06,
+      "loss": 0.1798,
+      "step": 2748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.649233715175403e-06,
+      "loss": 0.1561,
+      "step": 2749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.643084374980589e-06,
+      "loss": 0.1575,
+      "step": 2750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.636941092954486e-06,
+      "loss": 0.1741,
+      "step": 2751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.630803874155877e-06,
+      "loss": 0.157,
+      "step": 2752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6246727236385573e-06,
+      "loss": 0.152,
+      "step": 2753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.618547646451317e-06,
+      "loss": 0.1702,
+      "step": 2754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6124286476379433e-06,
+      "loss": 0.1572,
+      "step": 2755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.606315732237229e-06,
+      "loss": 0.19,
+      "step": 2756
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6002089052829472e-06,
+      "loss": 0.1624,
+      "step": 2757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5941081718038618e-06,
+      "loss": 0.1534,
+      "step": 2758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5880135368237226e-06,
+      "loss": 0.1698,
+      "step": 2759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.581925005361251e-06,
+      "loss": 0.1455,
+      "step": 2760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.575842582430139e-06,
+      "loss": 0.1467,
+      "step": 2761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5697662730390604e-06,
+      "loss": 0.1513,
+      "step": 2762
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.56369608219165e-06,
+      "loss": 0.1761,
+      "step": 2763
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5576320148864964e-06,
+      "loss": 0.1817,
+      "step": 2764
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.551574076117157e-06,
+      "loss": 0.1565,
+      "step": 2765
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.545522270872137e-06,
+      "loss": 0.1574,
+      "step": 2766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.539476604134886e-06,
+      "loss": 0.1478,
+      "step": 2767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.533437080883808e-06,
+      "loss": 0.1618,
+      "step": 2768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5274037060922472e-06,
+      "loss": 0.1515,
+      "step": 2769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.521376484728475e-06,
+      "loss": 0.1773,
+      "step": 2770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5153554217557085e-06,
+      "loss": 0.1838,
+      "step": 2771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5093405221320856e-06,
+      "loss": 0.1684,
+      "step": 2772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5033317908106657e-06,
+      "loss": 0.1327,
+      "step": 2773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.497329232739438e-06,
+      "loss": 0.1704,
+      "step": 2774
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4913328528613033e-06,
+      "loss": 0.1499,
+      "step": 2775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4853426561140804e-06,
+      "loss": 0.1469,
+      "step": 2776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4793586474304867e-06,
+      "loss": 0.1518,
+      "step": 2777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.473380831738146e-06,
+      "loss": 0.1619,
+      "step": 2778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4674092139595928e-06,
+      "loss": 0.2022,
+      "step": 2779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.461443799012243e-06,
+      "loss": 0.1531,
+      "step": 2780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4554845918084156e-06,
+      "loss": 0.1632,
+      "step": 2781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.449531597255318e-06,
+      "loss": 0.1606,
+      "step": 2782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.443584820255033e-06,
+      "loss": 0.1481,
+      "step": 2783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.437644265704526e-06,
+      "loss": 0.1411,
+      "step": 2784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.431709938495648e-06,
+      "loss": 0.1412,
+      "step": 2785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.425781843515107e-06,
+      "loss": 0.1813,
+      "step": 2786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4198599856444916e-06,
+      "loss": 0.1897,
+      "step": 2787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4139443697602504e-06,
+      "loss": 0.1353,
+      "step": 2788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4080350007336893e-06,
+      "loss": 0.1559,
+      "step": 2789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4021318834309682e-06,
+      "loss": 0.1521,
+      "step": 2790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.396235022713107e-06,
+      "loss": 0.1386,
+      "step": 2791
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3903444234359697e-06,
+      "loss": 0.1158,
+      "step": 2792
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3844600904502603e-06,
+      "loss": 0.1781,
+      "step": 2793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3785820286015293e-06,
+      "loss": 0.1572,
+      "step": 2794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3727102427301583e-06,
+      "loss": 0.1694,
+      "step": 2795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3668447376713586e-06,
+      "loss": 0.1557,
+      "step": 2796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3609855182551777e-06,
+      "loss": 0.1533,
+      "step": 2797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.355132589306486e-06,
+      "loss": 0.1618,
+      "step": 2798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.349285955644961e-06,
+      "loss": 0.1451,
+      "step": 2799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3434456220851154e-06,
+      "loss": 0.1673,
+      "step": 2800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.33761159343626e-06,
+      "loss": 0.1649,
+      "step": 2801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3317838745025154e-06,
+      "loss": 0.1702,
+      "step": 2802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.325962470082812e-06,
+      "loss": 0.1412,
+      "step": 2803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3201473849708776e-06,
+      "loss": 0.1825,
+      "step": 2804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.314338623955238e-06,
+      "loss": 0.1574,
+      "step": 2805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3085361918192084e-06,
+      "loss": 0.1507,
+      "step": 2806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3027400933408893e-06,
+      "loss": 0.1237,
+      "step": 2807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2969503332931754e-06,
+      "loss": 0.1565,
+      "step": 2808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2911669164437322e-06,
+      "loss": 0.1619,
+      "step": 2809
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2853898475550064e-06,
+      "loss": 0.1727,
+      "step": 2810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.279619131384222e-06,
+      "loss": 0.1385,
+      "step": 2811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2738547726833614e-06,
+      "loss": 0.1615,
+      "step": 2812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.268096776199177e-06,
+      "loss": 0.1711,
+      "step": 2813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.262345146673186e-06,
+      "loss": 0.1361,
+      "step": 2814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.256599888841654e-06,
+      "loss": 0.1551,
+      "step": 2815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2508610074356075e-06,
+      "loss": 0.1664,
+      "step": 2816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2451285071808204e-06,
+      "loss": 0.171,
+      "step": 2817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2394023927978105e-06,
+      "loss": 0.1332,
+      "step": 2818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.233682669001833e-06,
+      "loss": 0.171,
+      "step": 2819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2279693405028892e-06,
+      "loss": 0.1558,
+      "step": 2820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2222624120057114e-06,
+      "loss": 0.1502,
+      "step": 2821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.216561888209754e-06,
+      "loss": 0.126,
+      "step": 2822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2108677738092123e-06,
+      "loss": 0.1481,
+      "step": 2823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2051800734929885e-06,
+      "loss": 0.1468,
+      "step": 2824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1994987919447096e-06,
+      "loss": 0.1556,
+      "step": 2825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.193823933842718e-06,
+      "loss": 0.1589,
+      "step": 2826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1881555038600698e-06,
+      "loss": 0.1704,
+      "step": 2827
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1824935066645158e-06,
+      "loss": 0.1573,
+      "step": 2828
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1768379469185254e-06,
+      "loss": 0.142,
+      "step": 2829
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1711888292792537e-06,
+      "loss": 0.1242,
+      "step": 2830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.165546158398555e-06,
+      "loss": 0.1347,
+      "step": 2831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.159909938922977e-06,
+      "loss": 0.1522,
+      "step": 2832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1542801754937593e-06,
+      "loss": 0.1777,
+      "step": 2833
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1486568727468117e-06,
+      "loss": 0.1632,
+      "step": 2834
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.143040035312739e-06,
+      "loss": 0.1579,
+      "step": 2835
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.137429667816808e-06,
+      "loss": 0.1659,
+      "step": 2836
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1318257748789718e-06,
+      "loss": 0.128,
+      "step": 2837
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.126228361113839e-06,
+      "loss": 0.1534,
+      "step": 2838
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.120637431130691e-06,
+      "loss": 0.1537,
+      "step": 2839
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1150529895334706e-06,
+      "loss": 0.1604,
+      "step": 2840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.109475040920771e-06,
+      "loss": 0.1633,
+      "step": 2841
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.103903589885842e-06,
+      "loss": 0.1641,
+      "step": 2842
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.098338641016585e-06,
+      "loss": 0.1713,
+      "step": 2843
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.092780198895543e-06,
+      "loss": 0.1431,
+      "step": 2844
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0872282680999056e-06,
+      "loss": 0.1524,
+      "step": 2845
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.081682853201499e-06,
+      "loss": 0.1678,
+      "step": 2846
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.076143958766782e-06,
+      "loss": 0.1661,
+      "step": 2847
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.070611589356841e-06,
+      "loss": 0.1375,
+      "step": 2848
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.065085749527397e-06,
+      "loss": 0.1741,
+      "step": 2849
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.059566443828793e-06,
+      "loss": 0.1618,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.054053676805983e-06,
+      "loss": 0.1631,
+      "step": 2851
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0485474529985474e-06,
+      "loss": 0.1313,
+      "step": 2852
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0430477769406697e-06,
+      "loss": 0.156,
+      "step": 2853
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.037554653161146e-06,
+      "loss": 0.1504,
+      "step": 2854
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.032068086183375e-06,
+      "loss": 0.172,
+      "step": 2855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.026588080525362e-06,
+      "loss": 0.1388,
+      "step": 2856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0211146406996994e-06,
+      "loss": 0.1667,
+      "step": 2857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0156477712135815e-06,
+      "loss": 0.1558,
+      "step": 2858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.010187476568788e-06,
+      "loss": 0.1332,
+      "step": 2859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0047337612616803e-06,
+      "loss": 0.1339,
+      "step": 2860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9992866297832104e-06,
+      "loss": 0.1608,
+      "step": 2861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.993846086618908e-06,
+      "loss": 0.1576,
+      "step": 2862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9884121362488683e-06,
+      "loss": 0.1468,
+      "step": 2863
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.98298478314777e-06,
+      "loss": 0.1822,
+      "step": 2864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9775640317848467e-06,
+      "loss": 0.1541,
+      "step": 2865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9721498866239084e-06,
+      "loss": 0.156,
+      "step": 2866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9667423521233132e-06,
+      "loss": 0.1493,
+      "step": 2867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9613414327359824e-06,
+      "loss": 0.1501,
+      "step": 2868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9559471329093937e-06,
+      "loss": 0.1514,
+      "step": 2869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9505594570855637e-06,
+      "loss": 0.145,
+      "step": 2870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9451784097010574e-06,
+      "loss": 0.1676,
+      "step": 2871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9398039951869895e-06,
+      "loss": 0.1641,
+      "step": 2872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.934436217969001e-06,
+      "loss": 0.1597,
+      "step": 2873
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9290750824672766e-06,
+      "loss": 0.1457,
+      "step": 2874
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9237205930965295e-06,
+      "loss": 0.1409,
+      "step": 2875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.918372754265997e-06,
+      "loss": 0.1486,
+      "step": 2876
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.913031570379438e-06,
+      "loss": 0.1698,
+      "step": 2877
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.907697045835142e-06,
+      "loss": 0.1436,
+      "step": 2878
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9023691850259007e-06,
+      "loss": 0.1791,
+      "step": 2879
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8970479923390294e-06,
+      "loss": 0.1696,
+      "step": 2880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8917334721563507e-06,
+      "loss": 0.14,
+      "step": 2881
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8864256288541883e-06,
+      "loss": 0.1337,
+      "step": 2882
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8811244668033658e-06,
+      "loss": 0.1229,
+      "step": 2883
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8758299903692135e-06,
+      "loss": 0.1527,
+      "step": 2884
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8705422039115527e-06,
+      "loss": 0.1531,
+      "step": 2885
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.865261111784691e-06,
+      "loss": 0.1545,
+      "step": 2886
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8599867183374332e-06,
+      "loss": 0.1705,
+      "step": 2887
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8547190279130567e-06,
+      "loss": 0.1697,
+      "step": 2888
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.849458044849325e-06,
+      "loss": 0.1388,
+      "step": 2889
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8442037734784801e-06,
+      "loss": 0.1566,
+      "step": 2890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8389562181272359e-06,
+      "loss": 0.1554,
+      "step": 2891
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8337153831167708e-06,
+      "loss": 0.1492,
+      "step": 2892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8284812727627388e-06,
+      "loss": 0.15,
+      "step": 2893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8232538913752474e-06,
+      "loss": 0.1636,
+      "step": 2894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8180332432588666e-06,
+      "loss": 0.1915,
+      "step": 2895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.812819332712621e-06,
+      "loss": 0.152,
+      "step": 2896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8076121640299904e-06,
+      "loss": 0.1492,
+      "step": 2897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8024117414989007e-06,
+      "loss": 0.1344,
+      "step": 2898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7972180694017216e-06,
+      "loss": 0.1377,
+      "step": 2899
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.792031152015261e-06,
+      "loss": 0.1465,
+      "step": 2900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7868509936107747e-06,
+      "loss": 0.1554,
+      "step": 2901
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7816775984539392e-06,
+      "loss": 0.1755,
+      "step": 2902
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7765109708048721e-06,
+      "loss": 0.1513,
+      "step": 2903
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7713511149181185e-06,
+      "loss": 0.1487,
+      "step": 2904
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7661980350426378e-06,
+      "loss": 0.1729,
+      "step": 2905
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.761051735421816e-06,
+      "loss": 0.1368,
+      "step": 2906
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7559122202934587e-06,
+      "loss": 0.1709,
+      "step": 2907
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.750779493889776e-06,
+      "loss": 0.1283,
+      "step": 2908
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7456535604373949e-06,
+      "loss": 0.1885,
+      "step": 2909
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7405344241573495e-06,
+      "loss": 0.1427,
+      "step": 2910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7354220892650719e-06,
+      "loss": 0.1487,
+      "step": 2911
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7303165599703919e-06,
+      "loss": 0.147,
+      "step": 2912
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.72521784047754e-06,
+      "loss": 0.1616,
+      "step": 2913
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7201259349851418e-06,
+      "loss": 0.1651,
+      "step": 2914
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7150408476862023e-06,
+      "loss": 0.1376,
+      "step": 2915
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.709962582768121e-06,
+      "loss": 0.1512,
+      "step": 2916
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7048911444126736e-06,
+      "loss": 0.1515,
+      "step": 2917
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6998265367960154e-06,
+      "loss": 0.1636,
+      "step": 2918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6947687640886778e-06,
+      "loss": 0.1344,
+      "step": 2919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6897178304555673e-06,
+      "loss": 0.1402,
+      "step": 2920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6846737400559543e-06,
+      "loss": 0.1582,
+      "step": 2921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6796364970434699e-06,
+      "loss": 0.158,
+      "step": 2922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.674606105566118e-06,
+      "loss": 0.1271,
+      "step": 2923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6695825697662483e-06,
+      "loss": 0.1508,
+      "step": 2924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6645658937805753e-06,
+      "loss": 0.1425,
+      "step": 2925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.659556081740158e-06,
+      "loss": 0.1609,
+      "step": 2926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6545531377704126e-06,
+      "loss": 0.131,
+      "step": 2927
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.649557065991081e-06,
+      "loss": 0.1301,
+      "step": 2928
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6445678705162638e-06,
+      "loss": 0.1521,
+      "step": 2929
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6395855554543937e-06,
+      "loss": 0.1296,
+      "step": 2930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6346101249082335e-06,
+      "loss": 0.1511,
+      "step": 2931
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6296415829748857e-06,
+      "loss": 0.1699,
+      "step": 2932
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6246799337457719e-06,
+      "loss": 0.1423,
+      "step": 2933
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6197251813066373e-06,
+      "loss": 0.1515,
+      "step": 2934
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6147773297375546e-06,
+      "loss": 0.1462,
+      "step": 2935
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6098363831129126e-06,
+      "loss": 0.1432,
+      "step": 2936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.604902345501408e-06,
+      "loss": 0.1336,
+      "step": 2937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5999752209660568e-06,
+      "loss": 0.1247,
+      "step": 2938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5950550135641751e-06,
+      "loss": 0.149,
+      "step": 2939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.590141727347383e-06,
+      "loss": 0.1762,
+      "step": 2940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.585235366361606e-06,
+      "loss": 0.1552,
+      "step": 2941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5803359346470659e-06,
+      "loss": 0.1333,
+      "step": 2942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5754434362382786e-06,
+      "loss": 0.135,
+      "step": 2943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5705578751640483e-06,
+      "loss": 0.1339,
+      "step": 2944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5656792554474643e-06,
+      "loss": 0.1348,
+      "step": 2945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5608075811059075e-06,
+      "loss": 0.1561,
+      "step": 2946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5559428561510315e-06,
+      "loss": 0.1388,
+      "step": 2947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5510850845887736e-06,
+      "loss": 0.1686,
+      "step": 2948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5462342704193456e-06,
+      "loss": 0.1485,
+      "step": 2949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.541390417637224e-06,
+      "loss": 0.1579,
+      "step": 2950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.536553530231153e-06,
+      "loss": 0.142,
+      "step": 2951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5317236121841504e-06,
+      "loss": 0.1257,
+      "step": 2952
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5269006674734833e-06,
+      "loss": 0.1426,
+      "step": 2953
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.522084700070684e-06,
+      "loss": 0.1365,
+      "step": 2954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5172757139415395e-06,
+      "loss": 0.1473,
+      "step": 2955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5124737130460843e-06,
+      "loss": 0.1486,
+      "step": 2956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5076787013385996e-06,
+      "loss": 0.1433,
+      "step": 2957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5028906827676148e-06,
+      "loss": 0.1271,
+      "step": 2958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4981096612759027e-06,
+      "loss": 0.1452,
+      "step": 2959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4933356408004663e-06,
+      "loss": 0.1144,
+      "step": 2960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.488568625272554e-06,
+      "loss": 0.1356,
+      "step": 2961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.483808618617637e-06,
+      "loss": 0.1548,
+      "step": 2962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4790556247554167e-06,
+      "loss": 0.1662,
+      "step": 2963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.474309647599822e-06,
+      "loss": 0.1394,
+      "step": 2964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4695706910590058e-06,
+      "loss": 0.1639,
+      "step": 2965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4648387590353342e-06,
+      "loss": 0.1316,
+      "step": 2966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4601138554253935e-06,
+      "loss": 0.1468,
+      "step": 2967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4553959841199794e-06,
+      "loss": 0.1286,
+      "step": 2968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4506851490040962e-06,
+      "loss": 0.1632,
+      "step": 2969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4459813539569567e-06,
+      "loss": 0.156,
+      "step": 2970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4412846028519755e-06,
+      "loss": 0.1357,
+      "step": 2971
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4365948995567691e-06,
+      "loss": 0.157,
+      "step": 2972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.431912247933147e-06,
+      "loss": 0.1494,
+      "step": 2973
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4272366518371095e-06,
+      "loss": 0.1323,
+      "step": 2974
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4225681151188542e-06,
+      "loss": 0.1195,
+      "step": 2975
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4179066416227582e-06,
+      "loss": 0.1424,
+      "step": 2976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4132522351873878e-06,
+      "loss": 0.1613,
+      "step": 2977
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4086048996454893e-06,
+      "loss": 0.1381,
+      "step": 2978
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4039646388239825e-06,
+      "loss": 0.119,
+      "step": 2979
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3993314565439619e-06,
+      "loss": 0.1536,
+      "step": 2980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3947053566207004e-06,
+      "loss": 0.1467,
+      "step": 2981
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3900863428636268e-06,
+      "loss": 0.1315,
+      "step": 2982
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3854744190763447e-06,
+      "loss": 0.1181,
+      "step": 2983
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.380869589056617e-06,
+      "loss": 0.1482,
+      "step": 2984
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3762718565963618e-06,
+      "loss": 0.1397,
+      "step": 2985
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3716812254816535e-06,
+      "loss": 0.1328,
+      "step": 2986
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.367097699492722e-06,
+      "loss": 0.1307,
+      "step": 2987
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3625212824039468e-06,
+      "loss": 0.1456,
+      "step": 2988
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3579519779838456e-06,
+      "loss": 0.1474,
+      "step": 2989
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3533897899950887e-06,
+      "loss": 0.1201,
+      "step": 2990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.348834722194482e-06,
+      "loss": 0.1363,
+      "step": 2991
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3442867783329639e-06,
+      "loss": 0.1569,
+      "step": 2992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.146,
+      "step": 2993
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3352122774016407e-06,
+      "loss": 0.1288,
+      "step": 2994
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.330685727804375e-06,
+      "loss": 0.1696,
+      "step": 2995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3261663170912797e-06,
+      "loss": 0.1428,
+      "step": 2996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3216540489839335e-06,
+      "loss": 0.1411,
+      "step": 2997
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3171489271980331e-06,
+      "loss": 0.1323,
+      "step": 2998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.312650955443393e-06,
+      "loss": 0.1432,
+      "step": 2999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.308160137423944e-06,
+      "loss": 0.1573,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3036764768377142e-06,
+      "loss": 0.137,
+      "step": 3001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2991999773768515e-06,
+      "loss": 0.1245,
+      "step": 3002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2947306427275942e-06,
+      "loss": 0.1529,
+      "step": 3003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2902684765702922e-06,
+      "loss": 0.1621,
+      "step": 3004
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2858134825793833e-06,
+      "loss": 0.1217,
+      "step": 3005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2813656644234029e-06,
+      "loss": 0.1293,
+      "step": 3006
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2769250257649813e-06,
+      "loss": 0.1437,
+      "step": 3007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2724915702608288e-06,
+      "loss": 0.1541,
+      "step": 3008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.268065301561744e-06,
+      "loss": 0.1012,
+      "step": 3009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2636462233126123e-06,
+      "loss": 0.1529,
+      "step": 3010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2592343391523887e-06,
+      "loss": 0.1547,
+      "step": 3011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2548296527141113e-06,
+      "loss": 0.1276,
+      "step": 3012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.250432167624891e-06,
+      "loss": 0.1286,
+      "step": 3013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2460418875059055e-06,
+      "loss": 0.1322,
+      "step": 3014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2416588159723963e-06,
+      "loss": 0.1368,
+      "step": 3015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2372829566336774e-06,
+      "loss": 0.1421,
+      "step": 3016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2329143130931199e-06,
+      "loss": 0.125,
+      "step": 3017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.228552888948149e-06,
+      "loss": 0.1678,
+      "step": 3018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2241986877902534e-06,
+      "loss": 0.1423,
+      "step": 3019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.219851713204966e-06,
+      "loss": 0.1212,
+      "step": 3020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2155119687718696e-06,
+      "loss": 0.1228,
+      "step": 3021
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2111794580645985e-06,
+      "loss": 0.1283,
+      "step": 3022
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2068541846508286e-06,
+      "loss": 0.1331,
+      "step": 3023
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2025361520922707e-06,
+      "loss": 0.1247,
+      "step": 3024
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1982253639446817e-06,
+      "loss": 0.1468,
+      "step": 3025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1939218237578465e-06,
+      "loss": 0.1544,
+      "step": 3026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1896255350755814e-06,
+      "loss": 0.1375,
+      "step": 3027
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1853365014357344e-06,
+      "loss": 0.1237,
+      "step": 3028
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.18105472637018e-06,
+      "loss": 0.1435,
+      "step": 3029
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1767802134048112e-06,
+      "loss": 0.1213,
+      "step": 3030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1725129660595469e-06,
+      "loss": 0.1382,
+      "step": 3031
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.168252987848315e-06,
+      "loss": 0.123,
+      "step": 3032
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1640002822790675e-06,
+      "loss": 0.1374,
+      "step": 3033
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.159754852853756e-06,
+      "loss": 0.1463,
+      "step": 3034
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.15551670306835e-06,
+      "loss": 0.1186,
+      "step": 3035
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1512858364128233e-06,
+      "loss": 0.1274,
+      "step": 3036
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1470622563711464e-06,
+      "loss": 0.1304,
+      "step": 3037
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1428459664212932e-06,
+      "loss": 0.1457,
+      "step": 3038
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1386369700352373e-06,
+      "loss": 0.1192,
+      "step": 3039
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1344352706789396e-06,
+      "loss": 0.1647,
+      "step": 3040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.130240871812358e-06,
+      "loss": 0.1656,
+      "step": 3041
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1260537768894385e-06,
+      "loss": 0.1399,
+      "step": 3042
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1218739893581099e-06,
+      "loss": 0.1212,
+      "step": 3043
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.117701512660281e-06,
+      "loss": 0.1237,
+      "step": 3044
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1135363502318485e-06,
+      "loss": 0.139,
+      "step": 3045
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1093785055026774e-06,
+      "loss": 0.1456,
+      "step": 3046
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1052279818966127e-06,
+      "loss": 0.1276,
+      "step": 3047
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1010847828314708e-06,
+      "loss": 0.1261,
+      "step": 3048
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0969489117190335e-06,
+      "loss": 0.1346,
+      "step": 3049
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0928203719650454e-06,
+      "loss": 0.1103,
+      "step": 3050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0886991669692215e-06,
+      "loss": 0.1215,
+      "step": 3051
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0845853001252349e-06,
+      "loss": 0.1253,
+      "step": 3052
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.08047877482071e-06,
+      "loss": 0.1223,
+      "step": 3053
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0763795944372357e-06,
+      "loss": 0.1121,
+      "step": 3054
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0722877623503425e-06,
+      "loss": 0.1666,
+      "step": 3055
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0682032819295153e-06,
+      "loss": 0.134,
+      "step": 3056
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0641261565381833e-06,
+      "loss": 0.1378,
+      "step": 3057
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0600563895337234e-06,
+      "loss": 0.1216,
+      "step": 3058
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0559939842674461e-06,
+      "loss": 0.1341,
+      "step": 3059
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0519389440846072e-06,
+      "loss": 0.1276,
+      "step": 3060
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0478912723243906e-06,
+      "loss": 0.1418,
+      "step": 3061
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0438509723199153e-06,
+      "loss": 0.1343,
+      "step": 3062
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0398180473982312e-06,
+      "loss": 0.1483,
+      "step": 3063
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.035792500880315e-06,
+      "loss": 0.1571,
+      "step": 3064
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0317743360810683e-06,
+      "loss": 0.1294,
+      "step": 3065
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0277635563093102e-06,
+      "loss": 0.1242,
+      "step": 3066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0237601648677797e-06,
+      "loss": 0.1329,
+      "step": 3067
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0197641650531364e-06,
+      "loss": 0.1157,
+      "step": 3068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0157755601559461e-06,
+      "loss": 0.1304,
+      "step": 3069
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.01179435346069e-06,
+      "loss": 0.1431,
+      "step": 3070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.007820548245758e-06,
+      "loss": 0.1269,
+      "step": 3071
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0038541477834418e-06,
+      "loss": 0.1309,
+      "step": 3072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.998951553399339e-07,
+      "loss": 0.1182,
+      "step": 3073
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.959435741753332e-07,
+      "loss": 0.1284,
+      "step": 3074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.91999407543628e-07,
+      "loss": 0.1256,
+      "step": 3075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.880626586927067e-07,
+      "loss": 0.1285,
+      "step": 3076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.841333308643497e-07,
+      "loss": 0.1439,
+      "step": 3077
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.80211427294222e-07,
+      "loss": 0.1375,
+      "step": 3078
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.76296951211876e-07,
+      "loss": 0.1512,
+      "step": 3079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.723899058407504e-07,
+      "loss": 0.12,
+      "step": 3080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.684902943981644e-07,
+      "loss": 0.1297,
+      "step": 3081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.645981200953137e-07,
+      "loss": 0.1305,
+      "step": 3082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.607133861372719e-07,
+      "loss": 0.1306,
+      "step": 3083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.56836095722985e-07,
+      "loss": 0.1348,
+      "step": 3084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.529662520452665e-07,
+      "loss": 0.1659,
+      "step": 3085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.491038582908018e-07,
+      "loss": 0.1633,
+      "step": 3086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.452489176401425e-07,
+      "loss": 0.1262,
+      "step": 3087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.414014332676968e-07,
+      "loss": 0.1375,
+      "step": 3088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.375614083417417e-07,
+      "loss": 0.1348,
+      "step": 3089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.33728846024402e-07,
+      "loss": 0.1245,
+      "step": 3090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.299037494716612e-07,
+      "loss": 0.1215,
+      "step": 3091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.260861218333595e-07,
+      "loss": 0.1285,
+      "step": 3092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.2227596625318e-07,
+      "loss": 0.1435,
+      "step": 3093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.184732858686585e-07,
+      "loss": 0.1344,
+      "step": 3094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.146780838111702e-07,
+      "loss": 0.123,
+      "step": 3095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.108903632059341e-07,
+      "loss": 0.125,
+      "step": 3096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.071101271720106e-07,
+      "loss": 0.1304,
+      "step": 3097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.033373788222921e-07,
+      "loss": 0.1335,
+      "step": 3098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.995721212635101e-07,
+      "loss": 0.1357,
+      "step": 3099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.958143575962263e-07,
+      "loss": 0.1482,
+      "step": 3100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.920640909148292e-07,
+      "loss": 0.1372,
+      "step": 3101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.88321324307535e-07,
+      "loss": 0.14,
+      "step": 3102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.845860608563861e-07,
+      "loss": 0.1406,
+      "step": 3103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.80858303637242e-07,
+      "loss": 0.1279,
+      "step": 3104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.771380557197839e-07,
+      "loss": 0.1247,
+      "step": 3105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.734253201675114e-07,
+      "loss": 0.1189,
+      "step": 3106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.697201000377331e-07,
+      "loss": 0.1548,
+      "step": 3107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.660223983815708e-07,
+      "loss": 0.1464,
+      "step": 3108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.623322182439564e-07,
+      "loss": 0.1509,
+      "step": 3109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.586495626636282e-07,
+      "loss": 0.1216,
+      "step": 3110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.549744346731248e-07,
+      "loss": 0.1456,
+      "step": 3111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.513068372987909e-07,
+      "loss": 0.1323,
+      "step": 3112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.476467735607675e-07,
+      "loss": 0.124,
+      "step": 3113
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.439942464729899e-07,
+      "loss": 0.1162,
+      "step": 3114
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.403492590431905e-07,
+      "loss": 0.144,
+      "step": 3115
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.367118142728936e-07,
+      "loss": 0.1278,
+      "step": 3116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.330819151574077e-07,
+      "loss": 0.1241,
+      "step": 3117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.294595646858339e-07,
+      "loss": 0.1587,
+      "step": 3118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.258447658410529e-07,
+      "loss": 0.1367,
+      "step": 3119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.222375215997258e-07,
+      "loss": 0.1369,
+      "step": 3120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.186378349322976e-07,
+      "loss": 0.1201,
+      "step": 3121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.150457088029873e-07,
+      "loss": 0.1486,
+      "step": 3122
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.114611461697863e-07,
+      "loss": 0.1569,
+      "step": 3123
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.078841499844625e-07,
+      "loss": 0.1338,
+      "step": 3124
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.043147231925464e-07,
+      "loss": 0.1233,
+      "step": 3125
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.007528687333421e-07,
+      "loss": 0.1262,
+      "step": 3126
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.971985895399126e-07,
+      "loss": 0.1379,
+      "step": 3127
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.936518885390865e-07,
+      "loss": 0.1226,
+      "step": 3128
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.901127686514542e-07,
+      "loss": 0.1303,
+      "step": 3129
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.865812327913524e-07,
+      "loss": 0.1547,
+      "step": 3130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.830572838668859e-07,
+      "loss": 0.1616,
+      "step": 3131
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795409247799069e-07,
+      "loss": 0.1425,
+      "step": 3132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.760321584260134e-07,
+      "loss": 0.1447,
+      "step": 3133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.725309876945552e-07,
+      "loss": 0.1512,
+      "step": 3134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690374154686331e-07,
+      "loss": 0.1385,
+      "step": 3135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655514446250745e-07,
+      "loss": 0.1409,
+      "step": 3136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.620730780344632e-07,
+      "loss": 0.147,
+      "step": 3137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.586023185611136e-07,
+      "loss": 0.1551,
+      "step": 3138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.551391690630805e-07,
+      "loss": 0.1665,
+      "step": 3139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.516836323921439e-07,
+      "loss": 0.1298,
+      "step": 3140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.482357113938265e-07,
+      "loss": 0.1481,
+      "step": 3141
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.447954089073695e-07,
+      "loss": 0.1487,
+      "step": 3142
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.413627277657443e-07,
+      "loss": 0.1157,
+      "step": 3143
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.379376707956464e-07,
+      "loss": 0.1241,
+      "step": 3144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.345202408174968e-07,
+      "loss": 0.1429,
+      "step": 3145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.311104406454284e-07,
+      "loss": 0.1672,
+      "step": 3146
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.277082730872965e-07,
+      "loss": 0.1448,
+      "step": 3147
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.24313740944671e-07,
+      "loss": 0.1311,
+      "step": 3148
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.209268470128306e-07,
+      "loss": 0.1538,
+      "step": 3149
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.175475940807674e-07,
+      "loss": 0.1529,
+      "step": 3150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.141759849311847e-07,
+      "loss": 0.1122,
+      "step": 3151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.108120223404836e-07,
+      "loss": 0.1397,
+      "step": 3152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.074557090787715e-07,
+      "loss": 0.1299,
+      "step": 3153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.041070479098599e-07,
+      "loss": 0.1602,
+      "step": 3154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.007660415912576e-07,
+      "loss": 0.1256,
+      "step": 3155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.97432692874167e-07,
+      "loss": 0.1326,
+      "step": 3156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.941070045034892e-07,
+      "loss": 0.1434,
+      "step": 3157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.907889792178135e-07,
+      "loss": 0.1345,
+      "step": 3158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.874786197494188e-07,
+      "loss": 0.1349,
+      "step": 3159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.841759288242744e-07,
+      "loss": 0.1389,
+      "step": 3160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.80880909162036e-07,
+      "loss": 0.1399,
+      "step": 3161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.775935634760367e-07,
+      "loss": 0.1551,
+      "step": 3162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.743138944732952e-07,
+      "loss": 0.128,
+      "step": 3163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.710419048545059e-07,
+      "loss": 0.1548,
+      "step": 3164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.677775973140388e-07,
+      "loss": 0.1438,
+      "step": 3165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.645209745399405e-07,
+      "loss": 0.1197,
+      "step": 3166
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.612720392139327e-07,
+      "loss": 0.1471,
+      "step": 3167
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.580307940113972e-07,
+      "loss": 0.1449,
+      "step": 3168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.547972416013926e-07,
+      "loss": 0.1498,
+      "step": 3169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.515713846466354e-07,
+      "loss": 0.1205,
+      "step": 3170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.48353225803513e-07,
+      "loss": 0.1437,
+      "step": 3171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.451427677220634e-07,
+      "loss": 0.1495,
+      "step": 3172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.41940013045994e-07,
+      "loss": 0.1343,
+      "step": 3173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.387449644126631e-07,
+      "loss": 0.1193,
+      "step": 3174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.355576244530837e-07,
+      "loss": 0.1433,
+      "step": 3175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.323779957919196e-07,
+      "loss": 0.135,
+      "step": 3176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.292060810474898e-07,
+      "loss": 0.14,
+      "step": 3177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.260418828317539e-07,
+      "loss": 0.1338,
+      "step": 3178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.228854037503218e-07,
+      "loss": 0.1628,
+      "step": 3179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.197366464024502e-07,
+      "loss": 0.1469,
+      "step": 3180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.165956133810303e-07,
+      "loss": 0.1168,
+      "step": 3181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.134623072725932e-07,
+      "loss": 0.127,
+      "step": 3182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.103367306573116e-07,
+      "loss": 0.1288,
+      "step": 3183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.072188861089933e-07,
+      "loss": 0.1381,
+      "step": 3184
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.041087761950737e-07,
+      "loss": 0.1262,
+      "step": 3185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.01006403476625e-07,
+      "loss": 0.134,
+      "step": 3186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.979117705083415e-07,
+      "loss": 0.1566,
+      "step": 3187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.948248798385492e-07,
+      "loss": 0.1282,
+      "step": 3188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.917457340091981e-07,
+      "loss": 0.1197,
+      "step": 3189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.886743355558589e-07,
+      "loss": 0.1371,
+      "step": 3190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.856106870077227e-07,
+      "loss": 0.1296,
+      "step": 3191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.825547908876017e-07,
+      "loss": 0.1425,
+      "step": 3192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.795066497119206e-07,
+      "loss": 0.141,
+      "step": 3193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.764662659907172e-07,
+      "loss": 0.1483,
+      "step": 3194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.734336422276454e-07,
+      "loss": 0.1549,
+      "step": 3195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.704087809199698e-07,
+      "loss": 0.1307,
+      "step": 3196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.67391684558557e-07,
+      "loss": 0.1439,
+      "step": 3197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.643823556278849e-07,
+      "loss": 0.1725,
+      "step": 3198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.613807966060303e-07,
+      "loss": 0.1315,
+      "step": 3199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.583870099646782e-07,
+      "loss": 0.1481,
+      "step": 3200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.554009981691067e-07,
+      "loss": 0.1527,
+      "step": 3201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.524227636781953e-07,
+      "loss": 0.166,
+      "step": 3202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.494523089444203e-07,
+      "loss": 0.148,
+      "step": 3203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.464896364138472e-07,
+      "loss": 0.1183,
+      "step": 3204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.435347485261366e-07,
+      "loss": 0.1621,
+      "step": 3205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.405876477145378e-07,
+      "loss": 0.1299,
+      "step": 3206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.376483364058871e-07,
+      "loss": 0.1277,
+      "step": 3207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.347168170206063e-07,
+      "loss": 0.1455,
+      "step": 3208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.317930919727055e-07,
+      "loss": 0.1497,
+      "step": 3209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.288771636697676e-07,
+      "loss": 0.1331,
+      "step": 3210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.25969034512962e-07,
+      "loss": 0.1422,
+      "step": 3211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.230687068970342e-07,
+      "loss": 0.1296,
+      "step": 3212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.201761832103025e-07,
+      "loss": 0.1272,
+      "step": 3213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.172914658346639e-07,
+      "loss": 0.1303,
+      "step": 3214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.144145571455849e-07,
+      "loss": 0.1531,
+      "step": 3215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.115454595121006e-07,
+      "loss": 0.1411,
+      "step": 3216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.08684175296813e-07,
+      "loss": 0.1658,
+      "step": 3217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.058307068558932e-07,
+      "loss": 0.1427,
+      "step": 3218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.029850565390771e-07,
+      "loss": 0.11,
+      "step": 3219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.001472266896579e-07,
+      "loss": 0.1305,
+      "step": 3220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.973172196444942e-07,
+      "loss": 0.1188,
+      "step": 3221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.944950377339985e-07,
+      "loss": 0.1287,
+      "step": 3222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.916806832821385e-07,
+      "loss": 0.1717,
+      "step": 3223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.888741586064427e-07,
+      "loss": 0.1514,
+      "step": 3224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.860754660179889e-07,
+      "loss": 0.1511,
+      "step": 3225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.832846078214004e-07,
+      "loss": 0.1187,
+      "step": 3226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.805015863148588e-07,
+      "loss": 0.1445,
+      "step": 3227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.777264037900841e-07,
+      "loss": 0.1374,
+      "step": 3228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7495906253234416e-07,
+      "loss": 0.1414,
+      "step": 3229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7219956482045137e-07,
+      "loss": 0.1363,
+      "step": 3230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6944791292675707e-07,
+      "loss": 0.1522,
+      "step": 3231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.667041091171531e-07,
+      "loss": 0.1438,
+      "step": 3232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.639681556510689e-07,
+      "loss": 0.1247,
+      "step": 3233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.612400547814655e-07,
+      "loss": 0.1272,
+      "step": 3234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.585198087548448e-07,
+      "loss": 0.1337,
+      "step": 3235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.558074198112339e-07,
+      "loss": 0.1386,
+      "step": 3236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.531028901841927e-07,
+      "loss": 0.1294,
+      "step": 3237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.50406222100811e-07,
+      "loss": 0.1513,
+      "step": 3238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4771741778170007e-07,
+      "loss": 0.1546,
+      "step": 3239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4503647944100003e-07,
+      "loss": 0.1303,
+      "step": 3240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4236340928637155e-07,
+      "loss": 0.1271,
+      "step": 3241
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.396982095189961e-07,
+      "loss": 0.1199,
+      "step": 3242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.370408823335759e-07,
+      "loss": 0.123,
+      "step": 3243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.343914299183283e-07,
+      "loss": 0.1208,
+      "step": 3244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3174985445498807e-07,
+      "loss": 0.1229,
+      "step": 3245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.291161581187997e-07,
+      "loss": 0.1496,
+      "step": 3246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.264903430785239e-07,
+      "loss": 0.1507,
+      "step": 3247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2387241149642987e-07,
+      "loss": 0.1328,
+      "step": 3248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2126236552829435e-07,
+      "loss": 0.1345,
+      "step": 3249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.186602073234025e-07,
+      "loss": 0.1396,
+      "step": 3250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.160659390245414e-07,
+      "loss": 0.1264,
+      "step": 3251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.134795627679999e-07,
+      "loss": 0.1151,
+      "step": 3252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.109010806835745e-07,
+      "loss": 0.1429,
+      "step": 3253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.083304948945566e-07,
+      "loss": 0.1504,
+      "step": 3254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.057678075177329e-07,
+      "loss": 0.1545,
+      "step": 3255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.032130206633922e-07,
+      "loss": 0.1182,
+      "step": 3256
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.006661364353126e-07,
+      "loss": 0.1428,
+      "step": 3257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.981271569307654e-07,
+      "loss": 0.13,
+      "step": 3258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9559608424051485e-07,
+      "loss": 0.1271,
+      "step": 3259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.930729204488126e-07,
+      "loss": 0.114,
+      "step": 3260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.905576676333989e-07,
+      "loss": 0.1595,
+      "step": 3261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.880503278654979e-07,
+      "loss": 0.1588,
+      "step": 3262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8555090320981594e-07,
+      "loss": 0.1519,
+      "step": 3263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8305939572454766e-07,
+      "loss": 0.121,
+      "step": 3264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.805758074613597e-07,
+      "loss": 0.1359,
+      "step": 3265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7810014046540613e-07,
+      "loss": 0.1341,
+      "step": 3266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.756323967753139e-07,
+      "loss": 0.119,
+      "step": 3267
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.731725784231843e-07,
+      "loss": 0.1365,
+      "step": 3268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.707206874345937e-07,
+      "loss": 0.1451,
+      "step": 3269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.682767258285913e-07,
+      "loss": 0.1474,
+      "step": 3270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.658406956176941e-07,
+      "loss": 0.1285,
+      "step": 3271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.634125988078918e-07,
+      "loss": 0.1231,
+      "step": 3272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6099243739864066e-07,
+      "loss": 0.1327,
+      "step": 3273
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.585802133828575e-07,
+      "loss": 0.1208,
+      "step": 3274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5617592874692795e-07,
+      "loss": 0.137,
+      "step": 3275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.537795854706971e-07,
+      "loss": 0.1585,
+      "step": 3276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.513911855274743e-07,
+      "loss": 0.1555,
+      "step": 3277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4901073088402296e-07,
+      "loss": 0.1509,
+      "step": 3278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.466382235005683e-07,
+      "loss": 0.1424,
+      "step": 3279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4427366533078765e-07,
+      "loss": 0.1307,
+      "step": 3280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4191705832181344e-07,
+      "loss": 0.13,
+      "step": 3281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3956840441423114e-07,
+      "loss": 0.1068,
+      "step": 3282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3722770554207917e-07,
+      "loss": 0.1344,
+      "step": 3283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3489496363284024e-07,
+      "loss": 0.1463,
+      "step": 3284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3257018060744993e-07,
+      "loss": 0.1508,
+      "step": 3285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.302533583802858e-07,
+      "loss": 0.1197,
+      "step": 3286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.279444988591729e-07,
+      "loss": 0.1414,
+      "step": 3287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2564360394537696e-07,
+      "loss": 0.1223,
+      "step": 3288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.233506755336091e-07,
+      "loss": 0.121,
+      "step": 3289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.210657155120145e-07,
+      "loss": 0.1126,
+      "step": 3290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1878872576218247e-07,
+      "loss": 0.1225,
+      "step": 3291
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.165197081591331e-07,
+      "loss": 0.145,
+      "step": 3292
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1425866457132746e-07,
+      "loss": 0.1574,
+      "step": 3293
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.120055968606561e-07,
+      "loss": 0.1491,
+      "step": 3294
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0976050688244387e-07,
+      "loss": 0.1401,
+      "step": 3295
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0752339648544737e-07,
+      "loss": 0.137,
+      "step": 3296
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0529426751184864e-07,
+      "loss": 0.118,
+      "step": 3297
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0307312179725824e-07,
+      "loss": 0.1397,
+      "step": 3298
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0085996117071634e-07,
+      "loss": 0.1224,
+      "step": 3299
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.98654787454683e-07,
+      "loss": 0.1513,
+      "step": 3300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.964576024650445e-07,
+      "loss": 0.1274,
+      "step": 3301
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.942684080111091e-07,
+      "loss": 0.1323,
+      "step": 3302
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.920872058956015e-07,
+      "loss": 0.1429,
+      "step": 3303
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.89913997914667e-07,
+      "loss": 0.1257,
+      "step": 3304
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.877487858578687e-07,
+      "loss": 0.1192,
+      "step": 3305
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.855915715081858e-07,
+      "loss": 0.1301,
+      "step": 3306
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.834423566420086e-07,
+      "loss": 0.1424,
+      "step": 3307
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.813011430291446e-07,
+      "loss": 0.1335,
+      "step": 3308
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7916793243280895e-07,
+      "loss": 0.1424,
+      "step": 3309
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.770427266096265e-07,
+      "loss": 0.1424,
+      "step": 3310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7492552730963185e-07,
+      "loss": 0.1311,
+      "step": 3311
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7281633627626925e-07,
+      "loss": 0.115,
+      "step": 3312
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.707151552463816e-07,
+      "loss": 0.1345,
+      "step": 3313
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.686219859502226e-07,
+      "loss": 0.1278,
+      "step": 3314
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.665368301114457e-07,
+      "loss": 0.1261,
+      "step": 3315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6445968944710185e-07,
+      "loss": 0.1467,
+      "step": 3316
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.623905656676484e-07,
+      "loss": 0.1295,
+      "step": 3317
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6032946047693794e-07,
+      "loss": 0.1411,
+      "step": 3318
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.582763755722195e-07,
+      "loss": 0.1267,
+      "step": 3319
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.562313126441407e-07,
+      "loss": 0.115,
+      "step": 3320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.541942733767366e-07,
+      "loss": 0.1436,
+      "step": 3321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.521652594474433e-07,
+      "loss": 0.1439,
+      "step": 3322
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5014427252708195e-07,
+      "loss": 0.126,
+      "step": 3323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.481313142798669e-07,
+      "loss": 0.1413,
+      "step": 3324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4612638636340335e-07,
+      "loss": 0.1406,
+      "step": 3325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.441294904286773e-07,
+      "loss": 0.1267,
+      "step": 3326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4214062812006446e-07,
+      "loss": 0.1038,
+      "step": 3327
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.401598010753281e-07,
+      "loss": 0.1288,
+      "step": 3328
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3818701092560793e-07,
+      "loss": 0.1293,
+      "step": 3329
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3622225929543109e-07,
+      "loss": 0.1324,
+      "step": 3330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.342655478027045e-07,
+      "loss": 0.1173,
+      "step": 3331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.323168780587126e-07,
+      "loss": 0.1485,
+      "step": 3332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3037625166811628e-07,
+      "loss": 0.1356,
+      "step": 3333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2844367022895719e-07,
+      "loss": 0.1347,
+      "step": 3334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.26519135332649e-07,
+      "loss": 0.1142,
+      "step": 3335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2460264856398072e-07,
+      "loss": 0.1195,
+      "step": 3336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2269421150111548e-07,
+      "loss": 0.1412,
+      "step": 3337
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2079382571558504e-07,
+      "loss": 0.1193,
+      "step": 3338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1890149277228988e-07,
+      "loss": 0.144,
+      "step": 3339
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.170172142295035e-07,
+      "loss": 0.132,
+      "step": 3340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1514099163886581e-07,
+      "loss": 0.1417,
+      "step": 3341
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1327282654537983e-07,
+      "loss": 0.1198,
+      "step": 3342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1141272048741834e-07,
+      "loss": 0.1277,
+      "step": 3343
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0956067499671274e-07,
+      "loss": 0.1512,
+      "step": 3344
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0771669159835971e-07,
+      "loss": 0.1174,
+      "step": 3345
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.05880771810818e-07,
+      "loss": 0.1371,
+      "step": 3346
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.040529171459038e-07,
+      "loss": 0.1376,
+      "step": 3347
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.02233129108792e-07,
+      "loss": 0.1446,
+      "step": 3348
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0042140919801832e-07,
+      "loss": 0.1087,
+      "step": 3349
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9861775890546941e-07,
+      "loss": 0.1166,
+      "step": 3350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9682217971639273e-07,
+      "loss": 0.1256,
+      "step": 3351
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.950346731093833e-07,
+      "loss": 0.1401,
+      "step": 3352
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9325524055639367e-07,
+      "loss": 0.1134,
+      "step": 3353
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9148388352272618e-07,
+      "loss": 0.1293,
+      "step": 3354
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.897206034670318e-07,
+      "loss": 0.1309,
+      "step": 3355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8796540184131018e-07,
+      "loss": 0.1221,
+      "step": 3356
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.862182800909129e-07,
+      "loss": 0.1071,
+      "step": 3357
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8447923965453252e-07,
+      "loss": 0.1257,
+      "step": 3358
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8274828196421014e-07,
+      "loss": 0.1256,
+      "step": 3359
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8102540844533333e-07,
+      "loss": 0.1197,
+      "step": 3360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7931062051662396e-07,
+      "loss": 0.1259,
+      "step": 3361
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7760391959015467e-07,
+      "loss": 0.1369,
+      "step": 3362
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7590530707133567e-07,
+      "loss": 0.1433,
+      "step": 3363
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.742147843589137e-07,
+      "loss": 0.1261,
+      "step": 3364
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.725323528449807e-07,
+      "loss": 0.109,
+      "step": 3365
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7085801391495737e-07,
+      "loss": 0.1233,
+      "step": 3366
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6919176894760747e-07,
+      "loss": 0.134,
+      "step": 3367
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6753361931502454e-07,
+      "loss": 0.1209,
+      "step": 3368
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.658835663826397e-07,
+      "loss": 0.1375,
+      "step": 3369
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6424161150921603e-07,
+      "loss": 0.1615,
+      "step": 3370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.626077560468464e-07,
+      "loss": 0.1341,
+      "step": 3371
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6098200134095242e-07,
+      "loss": 0.1059,
+      "step": 3372
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5936434873029205e-07,
+      "loss": 0.1166,
+      "step": 3373
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5775479954694306e-07,
+      "loss": 0.1332,
+      "step": 3374
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5615335511631636e-07,
+      "loss": 0.1283,
+      "step": 3375
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5456001675714704e-07,
+      "loss": 0.1396,
+      "step": 3376
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.529747857814934e-07,
+      "loss": 0.1331,
+      "step": 3377
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5139766349474004e-07,
+      "loss": 0.1446,
+      "step": 3378
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4982865119559266e-07,
+      "loss": 0.1284,
+      "step": 3379
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4826775017607875e-07,
+      "loss": 0.1174,
+      "step": 3380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4671496172154797e-07,
+      "loss": 0.1319,
+      "step": 3381
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4517028711066862e-07,
+      "loss": 0.1261,
+      "step": 3382
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.436337276154276e-07,
+      "loss": 0.1137,
+      "step": 3383
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4210528450112837e-07,
+      "loss": 0.1401,
+      "step": 3384
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.405849590263919e-07,
+      "loss": 0.1381,
+      "step": 3385
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3907275244315455e-07,
+      "loss": 0.1471,
+      "step": 3386
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3756866599666686e-07,
+      "loss": 0.1193,
+      "step": 3387
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3607270092549252e-07,
+      "loss": 0.1262,
+      "step": 3388
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3458485846150725e-07,
+      "loss": 0.1194,
+      "step": 3389
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3310513982989882e-07,
+      "loss": 0.1138,
+      "step": 3390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.316335462491647e-07,
+      "loss": 0.1213,
+      "step": 3391
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3017007893111333e-07,
+      "loss": 0.1391,
+      "step": 3392
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.287147390808574e-07,
+      "loss": 0.1315,
+      "step": 3393
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2726752789682274e-07,
+      "loss": 0.1495,
+      "step": 3394
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.25828446570736e-07,
+      "loss": 0.1133,
+      "step": 3395
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2439749628763153e-07,
+      "loss": 0.1213,
+      "step": 3396
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2297467822584786e-07,
+      "loss": 0.1202,
+      "step": 3397
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2155999355702886e-07,
+      "loss": 0.1084,
+      "step": 3398
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2015344344611933e-07,
+      "loss": 0.1383,
+      "step": 3399
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1875502905136283e-07,
+      "loss": 0.1327,
+      "step": 3400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1736475152430704e-07,
+      "loss": 0.1485,
+      "step": 3401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1598261200979732e-07,
+      "loss": 0.132,
+      "step": 3402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1460861164597992e-07,
+      "loss": 0.1185,
+      "step": 3403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1324275156429532e-07,
+      "loss": 0.1162,
+      "step": 3404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1188503288948383e-07,
+      "loss": 0.1262,
+      "step": 3405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.105354567395811e-07,
+      "loss": 0.1202,
+      "step": 3406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0919402422591486e-07,
+      "loss": 0.1363,
+      "step": 3407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0786073645311035e-07,
+      "loss": 0.1392,
+      "step": 3408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.065355945190849e-07,
+      "loss": 0.1314,
+      "step": 3409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0521859951504565e-07,
+      "loss": 0.13,
+      "step": 3410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0390975252549618e-07,
+      "loss": 0.1279,
+      "step": 3411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0260905462822435e-07,
+      "loss": 0.1167,
+      "step": 3412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0131650689431116e-07,
+      "loss": 0.123,
+      "step": 3413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0003211038812632e-07,
+      "loss": 0.1401,
+      "step": 3414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.875586616732602e-08,
+      "loss": 0.139,
+      "step": 3415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.748777528285402e-08,
+      "loss": 0.131,
+      "step": 3416
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.622783877894059e-08,
+      "loss": 0.1208,
+      "step": 3417
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.497605769309803e-08,
+      "loss": 0.116,
+      "step": 3418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.373243305612733e-08,
+      "loss": 0.125,
+      "step": 3419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.249696589210933e-08,
+      "loss": 0.1128,
+      "step": 3420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.12696572184113e-08,
+      "loss": 0.124,
+      "step": 3421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.005050804567816e-08,
+      "loss": 0.1274,
+      "step": 3422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.883951937783908e-08,
+      "loss": 0.1222,
+      "step": 3423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.763669221210081e-08,
+      "loss": 0.1359,
+      "step": 3424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.644202753894882e-08,
+      "loss": 0.1205,
+      "step": 3425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.525552634215173e-08,
+      "loss": 0.1216,
+      "step": 3426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.407718959875133e-08,
+      "loss": 0.1078,
+      "step": 3427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.290701827906811e-08,
+      "loss": 0.1113,
+      "step": 3428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.174501334669682e-08,
+      "loss": 0.1367,
+      "step": 3429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.059117575851095e-08,
+      "loss": 0.1349,
+      "step": 3430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.944550646465488e-08,
+      "loss": 0.1216,
+      "step": 3431
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.83080064085484e-08,
+      "loss": 0.1369,
+      "step": 3432
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.717867652688338e-08,
+      "loss": 0.1209,
+      "step": 3433
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.605751774962589e-08,
+      "loss": 0.1145,
+      "step": 3434
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.494453100001186e-08,
+      "loss": 0.0986,
+      "step": 3435
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.383971719454597e-08,
+      "loss": 0.1204,
+      "step": 3436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.2743077243006e-08,
+      "loss": 0.1335,
+      "step": 3437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.165461204843738e-08,
+      "loss": 0.1351,
+      "step": 3438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.057432250715313e-08,
+      "loss": 0.1248,
+      "step": 3439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.95022095087361e-08,
+      "loss": 0.1349,
+      "step": 3440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.843827393603453e-08,
+      "loss": 0.1304,
+      "step": 3441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.738251666516205e-08,
+      "loss": 0.1227,
+      "step": 3442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.633493856549989e-08,
+      "loss": 0.1031,
+      "step": 3443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.529554049969244e-08,
+      "loss": 0.1332,
+      "step": 3444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.42643233236473e-08,
+      "loss": 0.1335,
+      "step": 3445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.32412878865385e-08,
+      "loss": 0.1171,
+      "step": 3446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.222643503079995e-08,
+      "loss": 0.1325,
+      "step": 3447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.121976559212762e-08,
+      "loss": 0.1444,
+      "step": 3448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.02212803994795e-08,
+      "loss": 0.1188,
+      "step": 3449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923098027507568e-08,
+      "loss": 0.1075,
+      "step": 3450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.8248866034392725e-08,
+      "loss": 0.1035,
+      "step": 3451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.727493848616927e-08,
+      "loss": 0.1214,
+      "step": 3452
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.630919843240046e-08,
+      "loss": 0.1392,
+      "step": 3453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.5351646668341253e-08,
+      "loss": 0.1037,
+      "step": 3454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.4402283982502024e-08,
+      "loss": 0.1391,
+      "step": 3455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.346111115665076e-08,
+      "loss": 0.1148,
+      "step": 3456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.252812896581083e-08,
+      "loss": 0.1144,
+      "step": 3457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.160333817826213e-08,
+      "loss": 0.1062,
+      "step": 3458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.06867395555366e-08,
+      "loss": 0.1382,
+      "step": 3459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.977833385242381e-08,
+      "loss": 0.1404,
+      "step": 3460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8878121816963163e-08,
+      "loss": 0.1106,
+      "step": 3461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.798610419045058e-08,
+      "loss": 0.1287,
+      "step": 3462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.710228170742959e-08,
+      "loss": 0.1281,
+      "step": 3463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.622665509570024e-08,
+      "loss": 0.1354,
+      "step": 3464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.535922507630908e-08,
+      "loss": 0.1243,
+      "step": 3465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4499992363556956e-08,
+      "loss": 0.1162,
+      "step": 3466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.364895766499011e-08,
+      "loss": 0.127,
+      "step": 3467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2806121681409076e-08,
+      "loss": 0.1358,
+      "step": 3468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1971485106859776e-08,
+      "loss": 0.1113,
+      "step": 3469
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.114504862863689e-08,
+      "loss": 0.1412,
+      "step": 3470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032681292728158e-08,
+      "loss": 0.124,
+      "step": 3471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9516778676584876e-08,
+      "loss": 0.1258,
+      "step": 3472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8714946543579876e-08,
+      "loss": 0.1088,
+      "step": 3473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.792131718855063e-08,
+      "loss": 0.1233,
+      "step": 3474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.713589126502215e-08,
+      "loss": 0.1516,
+      "step": 3475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6358669419765956e-08,
+      "loss": 0.1116,
+      "step": 3476
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5589652292797874e-08,
+      "loss": 0.129,
+      "step": 3477
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4828840517378003e-08,
+      "loss": 0.153,
+      "step": 3478
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.407623472000743e-08,
+      "loss": 0.1454,
+      "step": 3479
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.333183552043262e-08,
+      "loss": 0.1229,
+      "step": 3480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.259564353163991e-08,
+      "loss": 0.1049,
+      "step": 3481
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1867659359859914e-08,
+      "loss": 0.1332,
+      "step": 3482
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.114788360455978e-08,
+      "loss": 0.1325,
+      "step": 3483
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0436316858453165e-08,
+      "loss": 0.1064,
+      "step": 3484
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.973295970748913e-08,
+      "loss": 0.1341,
+      "step": 3485
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9037812730858816e-08,
+      "loss": 0.1408,
+      "step": 3486
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8350876500993218e-08,
+      "loss": 0.1298,
+      "step": 3487
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7672151583558736e-08,
+      "loss": 0.1079,
+      "step": 3488
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7001638537462737e-08,
+      "loss": 0.1261,
+      "step": 3489
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6339337914851325e-08,
+      "loss": 0.1392,
+      "step": 3490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.568525026110491e-08,
+      "loss": 0.1162,
+      "step": 3491
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.503937611484264e-08,
+      "loss": 0.1142,
+      "step": 3492
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4401716007920183e-08,
+      "loss": 0.1285,
+      "step": 3493
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3772270465428626e-08,
+      "loss": 0.135,
+      "step": 3494
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.315104000569446e-08,
+      "loss": 0.1032,
+      "step": 3495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2538025140280694e-08,
+      "loss": 0.1104,
+      "step": 3496
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.193322637398354e-08,
+      "loss": 0.112,
+      "step": 3497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1336644204834613e-08,
+      "loss": 0.1379,
+      "step": 3498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0748279124097604e-08,
+      "loss": 0.1088,
+      "step": 3499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0168131616272735e-08,
+      "loss": 0.1241,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9596202159090083e-08,
+      "loss": 0.1303,
+      "step": 3501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9032491223515136e-08,
+      "loss": 0.1274,
+      "step": 3502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8476999273744356e-08,
+      "loss": 0.1061,
+      "step": 3503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.792972676720628e-08,
+      "loss": 0.1166,
+      "step": 3504
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7390674154561525e-08,
+      "loss": 0.1228,
+      "step": 3505
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.685984187970058e-08,
+      "loss": 0.1429,
+      "step": 3506
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6337230379747105e-08,
+      "loss": 0.1268,
+      "step": 3507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5822840085052415e-08,
+      "loss": 0.1354,
+      "step": 3508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5316671419199903e-08,
+      "loss": 0.1461,
+      "step": 3509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4818724799002814e-08,
+      "loss": 0.1228,
+      "step": 3510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4329000634502043e-08,
+      "loss": 0.106,
+      "step": 3511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3847499328968338e-08,
+      "loss": 0.1226,
+      "step": 3512
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.337422127890342e-08,
+      "loss": 0.1381,
+      "step": 3513
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2909166874034429e-08,
+      "loss": 0.1196,
+      "step": 3514
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2452336497318363e-08,
+      "loss": 0.1319,
+      "step": 3515
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.200373052493875e-08,
+      "loss": 0.1269,
+      "step": 3516
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.156334932630676e-08,
+      "loss": 0.1217,
+      "step": 3517
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1131193264063422e-08,
+      "loss": 0.1057,
+      "step": 3518
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0707262694071852e-08,
+      "loss": 0.1163,
+      "step": 3519
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0291557965427245e-08,
+      "loss": 0.1208,
+      "step": 3520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.88407942044578e-09,
+      "loss": 0.1285,
+      "step": 3521
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.484827394673491e-09,
+      "loss": 0.1264,
+      "step": 3522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.09380221687961e-09,
+      "loss": 0.1395,
+      "step": 3523
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.711004209061013e-09,
+      "loss": 0.1457,
+      "step": 3524
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.336433686438882e-09,
+      "loss": 0.1214,
+      "step": 3525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.970090957458709e-09,
+      "loss": 0.1114,
+      "step": 3526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.611976323791403e-09,
+      "loss": 0.1086,
+      "step": 3527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.262090080331075e-09,
+      "loss": 0.1329,
+      "step": 3528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.920432515199471e-09,
+      "loss": 0.1217,
+      "step": 3529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.58700390973821e-09,
+      "loss": 0.1296,
+      "step": 3530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.2618045385143265e-09,
+      "loss": 0.1364,
+      "step": 3531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.944834669319166e-09,
+      "loss": 0.1401,
+      "step": 3532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.636094563166161e-09,
+      "loss": 0.0965,
+      "step": 3533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.335584474293054e-09,
+      "loss": 0.1232,
+      "step": 3534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.043304650157454e-09,
+      "loss": 0.1308,
+      "step": 3535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.759255331444612e-09,
+      "loss": 0.1197,
+      "step": 3536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.483436752057424e-09,
+      "loss": 0.1168,
+      "step": 3537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.215849139123096e-09,
+      "loss": 0.1254,
+      "step": 3538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.956492712992033e-09,
+      "loss": 0.1321,
+      "step": 3539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.705367687235617e-09,
+      "loss": 0.1227,
+      "step": 3540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4624742686451e-09,
+      "loss": 0.1074,
+      "step": 3541
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.22781265723604e-09,
+      "loss": 0.1159,
+      "step": 3542
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0013830462438663e-09,
+      "loss": 0.1204,
+      "step": 3543
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7831856221260946e-09,
+      "loss": 0.1235,
+      "step": 3544
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.573220564560108e-09,
+      "loss": 0.1304,
+      "step": 3545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3714880464464907e-09,
+      "loss": 0.1123,
+      "step": 3546
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.177988233902362e-09,
+      "loss": 0.1236,
+      "step": 3547
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9927212862702607e-09,
+      "loss": 0.1245,
+      "step": 3548
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8156873561114841e-09,
+      "loss": 0.1137,
+      "step": 3549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.646886589206087e-09,
+      "loss": 0.105,
+      "step": 3550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4863191245573227e-09,
+      "loss": 0.1161,
+      "step": 3551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.333985094384982e-09,
+      "loss": 0.1204,
+      "step": 3552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.189884624132054e-09,
+      "loss": 0.1335,
+      "step": 3553
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0540178324602857e-09,
+      "loss": 0.1325,
+      "step": 3554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.263848312512924e-10,
+      "loss": 0.1393,
+      "step": 3555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.069857256065572e-10,
+      "loss": 0.1167,
+      "step": 3556
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.958206138474311e-10,
+      "loss": 0.1214,
+      "step": 3557
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.92889587515133e-10,
+      "loss": 0.1308,
+      "step": 3558
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.981927313674195e-10,
+      "loss": 0.1251,
+      "step": 3559
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.117301233863558e-10,
+      "loss": 0.148,
+      "step": 3560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.335018347694341e-10,
+      "loss": 0.1458,
+      "step": 3561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.635079299362353e-10,
+      "loss": 0.1257,
+      "step": 3562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0174846652398773e-10,
+      "loss": 0.1231,
+      "step": 3563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4822349538867742e-10,
+      "loss": 0.1126,
+      "step": 3564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0293306060726871e-10,
+      "loss": 0.1037,
+      "step": 3565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.587719947437343e-11,
+      "loss": 0.1074,
+      "step": 3566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.705594250447142e-11,
+      "loss": 0.1265,
+      "step": 3567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.646931343080027e-11,
+      "loss": 0.1345,
+      "step": 3568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.117329205355347e-12,
+      "loss": 0.1381,
+      "step": 3569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.1127,
+      "step": 3570
+    },
+    {
+      "epoch": 2.0,
+      "step": 3570,
+      "total_flos": 0.0,
+      "train_loss": 0.04371294862582904,
+      "train_runtime": 11613.3895,
+      "train_samples_per_second": 317.219,
+      "train_steps_per_second": 0.307
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3570,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}