diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36216 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6031,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1049723756906078e-07,
+      "loss": 1.8507,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2099447513812156e-07,
+      "loss": 1.7921,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3149171270718233e-07,
+      "loss": 1.7128,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.419889502762431e-07,
+      "loss": 1.9926,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.524861878453039e-07,
+      "loss": 1.8133,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.629834254143647e-07,
+      "loss": 1.8334,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.734806629834254e-07,
+      "loss": 1.9118,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.839779005524863e-07,
+      "loss": 1.6763,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.94475138121547e-07,
+      "loss": 1.8015,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1049723756906078e-06,
+      "loss": 1.7977,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2154696132596686e-06,
+      "loss": 1.6741,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3259668508287293e-06,
+      "loss": 1.6352,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.43646408839779e-06,
+      "loss": 1.5915,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5469613259668508e-06,
+      "loss": 1.4584,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6574585635359118e-06,
+      "loss": 1.4533,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7679558011049725e-06,
+      "loss": 1.4798,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8784530386740332e-06,
+      "loss": 1.4361,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.988950276243094e-06,
+      "loss": 1.5762,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0994475138121547e-06,
+      "loss": 1.3193,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2099447513812157e-06,
+      "loss": 1.4039,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.320441988950276e-06,
+      "loss": 1.488,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.430939226519337e-06,
+      "loss": 1.3907,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.541436464088398e-06,
+      "loss": 1.5017,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6519337016574586e-06,
+      "loss": 1.4494,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7624309392265196e-06,
+      "loss": 1.3517,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.87292817679558e-06,
+      "loss": 1.3888,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.983425414364641e-06,
+      "loss": 1.4031,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0939226519337016e-06,
+      "loss": 1.4164,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.204419889502763e-06,
+      "loss": 1.3908,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3149171270718235e-06,
+      "loss": 1.3286,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4254143646408845e-06,
+      "loss": 1.3008,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.535911602209945e-06,
+      "loss": 1.3668,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.646408839779006e-06,
+      "loss": 1.2774,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7569060773480665e-06,
+      "loss": 1.3326,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.867403314917128e-06,
+      "loss": 1.3799,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.977900552486188e-06,
+      "loss": 1.3032,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.088397790055249e-06,
+      "loss": 1.2042,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1988950276243095e-06,
+      "loss": 1.4156,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.309392265193371e-06,
+      "loss": 1.2945,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.419889502762431e-06,
+      "loss": 1.3001,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.530386740331492e-06,
+      "loss": 1.2289,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.640883977900552e-06,
+      "loss": 1.3266,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.751381215469614e-06,
+      "loss": 1.3077,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.861878453038674e-06,
+      "loss": 1.2923,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972375690607736e-06,
+      "loss": 1.2413,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.082872928176796e-06,
+      "loss": 1.3244,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.193370165745857e-06,
+      "loss": 1.222,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.303867403314917e-06,
+      "loss": 1.2861,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.414364640883978e-06,
+      "loss": 1.2853,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.524861878453039e-06,
+      "loss": 1.284,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6353591160221e-06,
+      "loss": 1.2379,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.74585635359116e-06,
+      "loss": 1.3123,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.856353591160221e-06,
+      "loss": 1.2796,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.966850828729282e-06,
+      "loss": 1.2929,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.077348066298343e-06,
+      "loss": 1.2972,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.187845303867403e-06,
+      "loss": 1.2015,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.298342541436464e-06,
+      "loss": 1.2755,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.408839779005526e-06,
+      "loss": 1.236,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5193370165745865e-06,
+      "loss": 1.2906,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.629834254143647e-06,
+      "loss": 1.1608,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.740331491712708e-06,
+      "loss": 1.2464,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.850828729281769e-06,
+      "loss": 1.2023,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9613259668508295e-06,
+      "loss": 1.2306,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.07182320441989e-06,
+      "loss": 1.1715,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.182320441988951e-06,
+      "loss": 1.2227,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.292817679558012e-06,
+      "loss": 1.2749,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4033149171270724e-06,
+      "loss": 1.2586,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.513812154696133e-06,
+      "loss": 1.2551,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.624309392265194e-06,
+      "loss": 1.2327,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.734806629834256e-06,
+      "loss": 1.1822,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.845303867403316e-06,
+      "loss": 1.1531,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.955801104972377e-06,
+      "loss": 1.19,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.066298342541437e-06,
+      "loss": 1.2485,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.176795580110498e-06,
+      "loss": 1.1669,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.287292817679558e-06,
+      "loss": 1.192,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.397790055248619e-06,
+      "loss": 1.2205,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.508287292817681e-06,
+      "loss": 1.2821,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.618784530386742e-06,
+      "loss": 1.1297,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.729281767955802e-06,
+      "loss": 1.1686,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.839779005524863e-06,
+      "loss": 1.2982,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.950276243093923e-06,
+      "loss": 1.1771,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.060773480662984e-06,
+      "loss": 1.1804,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.171270718232044e-06,
+      "loss": 1.2441,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.281767955801105e-06,
+      "loss": 1.2227,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.392265193370167e-06,
+      "loss": 1.1495,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.502762430939228e-06,
+      "loss": 1.1835,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.613259668508288e-06,
+      "loss": 1.2845,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.723756906077349e-06,
+      "loss": 1.1868,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.834254143646411e-06,
+      "loss": 1.1642,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944751381215471e-06,
+      "loss": 1.2576,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0055248618784532e-05,
+      "loss": 1.1469,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0165745856353592e-05,
+      "loss": 1.1766,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0276243093922653e-05,
+      "loss": 1.1846,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0386740331491714e-05,
+      "loss": 1.2216,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0497237569060774e-05,
+      "loss": 1.2042,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0607734806629835e-05,
+      "loss": 1.1221,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0718232044198895e-05,
+      "loss": 1.2637,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0828729281767956e-05,
+      "loss": 1.1641,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0939226519337018e-05,
+      "loss": 1.2305,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1049723756906078e-05,
+      "loss": 1.1271,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1160220994475139e-05,
+      "loss": 1.2791,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.12707182320442e-05,
+      "loss": 1.1905,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.138121546961326e-05,
+      "loss": 1.1262,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.149171270718232e-05,
+      "loss": 1.2449,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1602209944751381e-05,
+      "loss": 1.1358,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1712707182320442e-05,
+      "loss": 1.1728,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1823204419889504e-05,
+      "loss": 1.1501,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1933701657458564e-05,
+      "loss": 1.233,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2044198895027625e-05,
+      "loss": 1.1512,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2154696132596685e-05,
+      "loss": 1.1938,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2265193370165746e-05,
+      "loss": 1.198,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2375690607734806e-05,
+      "loss": 1.1832,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2486187845303867e-05,
+      "loss": 1.1339,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2596685082872928e-05,
+      "loss": 1.1994,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2707182320441991e-05,
+      "loss": 1.1605,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2817679558011052e-05,
+      "loss": 1.1586,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2928176795580112e-05,
+      "loss": 1.1961,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3038674033149173e-05,
+      "loss": 1.2099,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3149171270718234e-05,
+      "loss": 1.0903,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3259668508287294e-05,
+      "loss": 1.168,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3370165745856355e-05,
+      "loss": 1.1756,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3480662983425417e-05,
+      "loss": 1.2549,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3591160220994477e-05,
+      "loss": 1.143,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3701657458563538e-05,
+      "loss": 1.1463,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3812154696132598e-05,
+      "loss": 1.1715,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3922651933701659e-05,
+      "loss": 1.1593,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.403314917127072e-05,
+      "loss": 1.1317,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.414364640883978e-05,
+      "loss": 1.2092,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.425414364640884e-05,
+      "loss": 1.1596,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4364640883977903e-05,
+      "loss": 1.1682,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4475138121546963e-05,
+      "loss": 1.1696,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4585635359116024e-05,
+      "loss": 1.2506,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4696132596685084e-05,
+      "loss": 1.1693,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4806629834254145e-05,
+      "loss": 1.0766,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4917127071823205e-05,
+      "loss": 1.1971,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5027624309392266e-05,
+      "loss": 1.2341,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5138121546961326e-05,
+      "loss": 1.0954,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5248618784530389e-05,
+      "loss": 1.1725,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.535911602209945e-05,
+      "loss": 1.2646,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546961325966851e-05,
+      "loss": 1.1144,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.558011049723757e-05,
+      "loss": 1.1894,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5690607734806633e-05,
+      "loss": 1.2182,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.580110497237569e-05,
+      "loss": 1.0637,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5911602209944754e-05,
+      "loss": 1.2242,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6022099447513812e-05,
+      "loss": 1.1412,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6132596685082875e-05,
+      "loss": 1.1704,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6243093922651937e-05,
+      "loss": 1.159,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6353591160220996e-05,
+      "loss": 1.1546,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6464088397790058e-05,
+      "loss": 1.1429,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6574585635359117e-05,
+      "loss": 1.2371,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.668508287292818e-05,
+      "loss": 1.0877,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6795580110497238e-05,
+      "loss": 1.1141,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.69060773480663e-05,
+      "loss": 1.211,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7016574585635362e-05,
+      "loss": 1.1934,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.712707182320442e-05,
+      "loss": 1.1159,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7237569060773483e-05,
+      "loss": 1.2276,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7348066298342542e-05,
+      "loss": 1.0779,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7458563535911604e-05,
+      "loss": 1.1643,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7569060773480663e-05,
+      "loss": 1.1663,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7679558011049725e-05,
+      "loss": 1.2093,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7790055248618784e-05,
+      "loss": 1.1395,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7900552486187846e-05,
+      "loss": 1.1258,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.801104972375691e-05,
+      "loss": 1.236,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8121546961325968e-05,
+      "loss": 1.11,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.823204419889503e-05,
+      "loss": 1.2089,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.834254143646409e-05,
+      "loss": 1.0893,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.845303867403315e-05,
+      "loss": 1.2835,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.856353591160221e-05,
+      "loss": 1.0827,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8674033149171272e-05,
+      "loss": 1.1476,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8784530386740334e-05,
+      "loss": 1.1788,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8895027624309393e-05,
+      "loss": 1.1337,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9005524861878455e-05,
+      "loss": 1.1321,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9116022099447514e-05,
+      "loss": 1.1415,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9226519337016576e-05,
+      "loss": 1.093,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9337016574585635e-05,
+      "loss": 1.1895,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9447513812154697e-05,
+      "loss": 1.1215,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9558011049723756e-05,
+      "loss": 1.2069,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9668508287292822e-05,
+      "loss": 1.1269,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977900552486188e-05,
+      "loss": 1.1051,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9889502762430943e-05,
+      "loss": 1.1376,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.1878,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998558024085e-05,
+      "loss": 1.1945,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999423209675e-05,
+      "loss": 1.138,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999987022219246e-05,
+      "loss": 1.1545,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976928393646e-05,
+      "loss": 1.1828,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963950622868e-05,
+      "loss": 1.1391,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999948088910656e-05,
+      "loss": 1.186,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999929343261576e-05,
+      "loss": 1.1659,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990771368104e-05,
+      "loss": 1.067,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999883200175286e-05,
+      "loss": 1.125,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999855802751384e-05,
+      "loss": 1.1937,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999825521417232e-05,
+      "loss": 1.173,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999792356181565e-05,
+      "loss": 1.1156,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999756307053947e-05,
+      "loss": 1.1743,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999717374044777e-05,
+      "loss": 1.1425,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999675557165282e-05,
+      "loss": 1.1551,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999630856427518e-05,
+      "loss": 1.1538,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999958327184438e-05,
+      "loss": 1.1501,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999953280342959e-05,
+      "loss": 1.1405,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999479451197703e-05,
+      "loss": 1.1052,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999423215164105e-05,
+      "loss": 1.1668,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999364095345015e-05,
+      "loss": 1.1819,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999302091757486e-05,
+      "loss": 1.1953,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999237204419392e-05,
+      "loss": 1.0868,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999169433349454e-05,
+      "loss": 1.2267,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999909877856721e-05,
+      "loss": 1.1132,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 1.1548,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998948817948157e-05,
+      "loss": 1.2231,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998869512154597e-05,
+      "loss": 1.1156,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998787322735223e-05,
+      "loss": 1.1235,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998702249713747e-05,
+      "loss": 1.1112,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998614293114708e-05,
+      "loss": 1.2415,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999852345296346e-05,
+      "loss": 1.1391,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998429729286207e-05,
+      "loss": 1.137,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998333122109982e-05,
+      "loss": 1.2003,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999823363146264e-05,
+      "loss": 1.1539,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998131257372878e-05,
+      "loss": 1.0841,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998025999870216e-05,
+      "loss": 1.2182,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997917858985013e-05,
+      "loss": 1.1459,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 1.1,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997692927192562e-05,
+      "loss": 1.1413,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997576136350184e-05,
+      "loss": 1.185,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997456462255003e-05,
+      "loss": 1.1336,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997333904941533e-05,
+      "loss": 1.1222,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997208464445113e-05,
+      "loss": 1.1215,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997080140801932e-05,
+      "loss": 1.1731,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996948934048984e-05,
+      "loss": 1.1158,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996814844224118e-05,
+      "loss": 1.1324,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996677871366003e-05,
+      "loss": 1.1855,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996538015514135e-05,
+      "loss": 1.1111,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996395276708856e-05,
+      "loss": 1.1337,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996249654991328e-05,
+      "loss": 1.1708,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996101150403543e-05,
+      "loss": 1.1007,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999594976298834e-05,
+      "loss": 1.129,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995795492789368e-05,
+      "loss": 1.1321,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995638339851122e-05,
+      "loss": 1.1913,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995478304218925e-05,
+      "loss": 1.1272,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999531538593893e-05,
+      "loss": 1.1494,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995149585058117e-05,
+      "loss": 1.1159,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999498090162431e-05,
+      "loss": 1.154,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994809335686152e-05,
+      "loss": 1.1605,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994634887293124e-05,
+      "loss": 1.1505,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999445755649553e-05,
+      "loss": 1.2246,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999427734334452e-05,
+      "loss": 1.111,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994094247892062e-05,
+      "loss": 1.1191,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 1.1742,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999371941029485e-05,
+      "loss": 1.0691,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993527668258195e-05,
+      "loss": 1.0911,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99933330441363e-05,
+      "loss": 1.0885,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993135537985285e-05,
+      "loss": 1.1944,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992935149862116e-05,
+      "loss": 1.1555,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999273187982458e-05,
+      "loss": 1.0967,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992525727931303e-05,
+      "loss": 1.1829,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992316694241735e-05,
+      "loss": 1.1353,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992104778816162e-05,
+      "loss": 1.175,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991889981715696e-05,
+      "loss": 1.1368,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991672303002293e-05,
+      "loss": 1.2389,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991451742738718e-05,
+      "loss": 1.1267,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 1.0878,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991001977816337e-05,
+      "loss": 1.1588,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999077277328724e-05,
+      "loss": 1.1413,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990540687467394e-05,
+      "loss": 1.0586,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990305720423735e-05,
+      "loss": 1.1447,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999006787222403e-05,
+      "loss": 1.2019,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989827142936864e-05,
+      "loss": 1.1002,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998958353263167e-05,
+      "loss": 1.126,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989337041378698e-05,
+      "loss": 1.1654,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998908766924904e-05,
+      "loss": 1.0834,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988835416314613e-05,
+      "loss": 1.1512,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998858028264816e-05,
+      "loss": 1.148,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998832226832327e-05,
+      "loss": 1.2001,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988061373414342e-05,
+      "loss": 1.0552,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987797597996625e-05,
+      "loss": 1.1198,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998753094214619e-05,
+      "loss": 1.1466,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987261405939934e-05,
+      "loss": 1.1415,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986988989455592e-05,
+      "loss": 1.1233,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986713692771732e-05,
+      "loss": 1.1658,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986435515967743e-05,
+      "loss": 1.1158,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986154459123854e-05,
+      "loss": 1.1046,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985870522321118e-05,
+      "loss": 1.1769,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985583705641418e-05,
+      "loss": 1.1095,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998529400916748e-05,
+      "loss": 1.1862,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998500143298284e-05,
+      "loss": 1.1407,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998470597717188e-05,
+      "loss": 1.1674,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984407641819812e-05,
+      "loss": 1.1548,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984106427012667e-05,
+      "loss": 1.0901,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998380233283732e-05,
+      "loss": 1.1649,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983495359381472e-05,
+      "loss": 1.1629,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983185506733643e-05,
+      "loss": 1.1411,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99828727749832e-05,
+      "loss": 1.1287,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982557164220335e-05,
+      "loss": 1.1766,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982238674536064e-05,
+      "loss": 1.1582,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998191730602224e-05,
+      "loss": 1.1039,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981593058771543e-05,
+      "loss": 1.0825,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981265932877486e-05,
+      "loss": 1.1737,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980935928434412e-05,
+      "loss": 1.1193,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998060304553749e-05,
+      "loss": 1.1434,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 1.075,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979928644766936e-05,
+      "loss": 1.1578,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99795871270878e-05,
+      "loss": 1.123,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979242731343803e-05,
+      "loss": 1.1027,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978895457634272e-05,
+      "loss": 1.1649,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997854530605935e-05,
+      "loss": 1.1001,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997819227672003e-05,
+      "loss": 1.1131,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977836369718117e-05,
+      "loss": 1.163,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977477585156252e-05,
+      "loss": 1.1596,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977115923137912e-05,
+      "loss": 1.1027,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976751383767393e-05,
+      "loss": 1.1376,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997638396714983e-05,
+      "loss": 1.1204,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976013673391185e-05,
+      "loss": 1.1177,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 1.0355,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975264454878634e-05,
+      "loss": 1.1476,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974885530340796e-05,
+      "loss": 1.1359,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974503729094023e-05,
+      "loss": 1.0535,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974119051248415e-05,
+      "loss": 1.2348,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973731496914914e-05,
+      "loss": 1.1544,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973341066205287e-05,
+      "loss": 1.1089,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972947759232136e-05,
+      "loss": 1.1046,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972551576108885e-05,
+      "loss": 1.0968,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972152516949795e-05,
+      "loss": 1.1246,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971750581869955e-05,
+      "loss": 1.13,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971345770985268e-05,
+      "loss": 1.1118,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970938084412498e-05,
+      "loss": 1.1964,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970527522269204e-05,
+      "loss": 1.0671,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970114084673796e-05,
+      "loss": 1.1724,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969697771745513e-05,
+      "loss": 1.1566,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996927858360441e-05,
+      "loss": 1.0825,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968856520371382e-05,
+      "loss": 1.1439,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968431582168152e-05,
+      "loss": 1.1132,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968003769117264e-05,
+      "loss": 1.1808,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 1.1184,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967139518966872e-05,
+      "loss": 1.0423,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966703082116613e-05,
+      "loss": 1.1924,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966263770917192e-05,
+      "loss": 1.133,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99658215854953e-05,
+      "loss": 1.1003,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996537652597847e-05,
+      "loss": 1.1455,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 1.1412,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996447778517421e-05,
+      "loss": 1.0877,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996402410414598e-05,
+      "loss": 1.1226,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996356754954119e-05,
+      "loss": 1.1776,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963108121491508e-05,
+      "loss": 1.0838,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962645820129433e-05,
+      "loss": 1.0826,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996218064558829e-05,
+      "loss": 1.1128,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961712598002234e-05,
+      "loss": 1.1399,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961241677506242e-05,
+      "loss": 1.1628,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960767884236132e-05,
+      "loss": 1.1057,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960291218328537e-05,
+      "loss": 1.1301,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959811679920938e-05,
+      "loss": 1.0488,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959329269151614e-05,
+      "loss": 1.1438,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958843986159705e-05,
+      "loss": 1.1883,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958355831085155e-05,
+      "loss": 1.1152,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957864804068752e-05,
+      "loss": 1.1399,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99573709052521e-05,
+      "loss": 1.0545,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995687413477764e-05,
+      "loss": 1.1688,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995637449278864e-05,
+      "loss": 1.1123,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955871979429188e-05,
+      "loss": 1.0737,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955366594844214e-05,
+      "loss": 1.1494,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954858339179462e-05,
+      "loss": 1.1705,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954347212581514e-05,
+      "loss": 1.0804,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995383321519778e-05,
+      "loss": 1.1183,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995331634717649e-05,
+      "loss": 1.1922,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952796608666704e-05,
+      "loss": 1.1069,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 1.1219,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951748520782038e-05,
+      "loss": 1.1265,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951220171709426e-05,
+      "loss": 1.1177,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950688952752846e-05,
+      "loss": 1.142,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950154864065497e-05,
+      "loss": 1.1178,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949617905801414e-05,
+      "loss": 1.1767,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994907807811545e-05,
+      "loss": 1.0801,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994853538116329e-05,
+      "loss": 1.1025,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947989815101444e-05,
+      "loss": 1.1666,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994744138008725e-05,
+      "loss": 1.0812,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946890076278874e-05,
+      "loss": 1.137,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946335903835314e-05,
+      "loss": 1.1593,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945778862916385e-05,
+      "loss": 1.1311,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 1.1103,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944656176295842e-05,
+      "loss": 1.1065,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944090530918005e-05,
+      "loss": 1.1437,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994352201771236e-05,
+      "loss": 1.1474,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994295063684286e-05,
+      "loss": 1.077,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942376388474282e-05,
+      "loss": 1.0866,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941799272772243e-05,
+      "loss": 1.1402,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994121928990318e-05,
+      "loss": 1.0802,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940636440034357e-05,
+      "loss": 1.0994,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940050723333867e-05,
+      "loss": 1.1018,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939462139970623e-05,
+      "loss": 1.1652,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938870690114372e-05,
+      "loss": 1.074,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938276373935688e-05,
+      "loss": 1.1192,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 1.1452,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937079143297427e-05,
+      "loss": 1.082,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936476229183133e-05,
+      "loss": 1.0732,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993587044943695e-05,
+      "loss": 1.1802,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935261804233587e-05,
+      "loss": 1.1263,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934650293748574e-05,
+      "loss": 1.0948,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934035918158265e-05,
+      "loss": 1.1713,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993341867763985e-05,
+      "loss": 1.0667,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993279857237133e-05,
+      "loss": 1.0625,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932175602531548e-05,
+      "loss": 1.1529,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993154976830016e-05,
+      "loss": 1.1228,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930921069857653e-05,
+      "loss": 1.1315,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930289507385344e-05,
+      "loss": 1.1317,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992965508106537e-05,
+      "loss": 1.0885,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99290177910807e-05,
+      "loss": 1.1897,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992837763761512e-05,
+      "loss": 1.1183,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927734620853255e-05,
+      "loss": 0.9895,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 1.2274,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992643999818325e-05,
+      "loss": 1.0914,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925788392648474e-05,
+      "loss": 1.0905,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925133924564136e-05,
+      "loss": 1.1573,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924476594118975e-05,
+      "loss": 1.1528,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992381640150257e-05,
+      "loss": 1.1217,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923153346905312e-05,
+      "loss": 1.1068,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922487430518426e-05,
+      "loss": 1.158,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921818652533958e-05,
+      "loss": 1.0569,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 1.1219,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992047251254459e-05,
+      "loss": 1.1394,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991979515092791e-05,
+      "loss": 1.1292,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919114928490085e-05,
+      "loss": 1.0169,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918431845427295e-05,
+      "loss": 1.139,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917745901936532e-05,
+      "loss": 1.1747,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917057098215624e-05,
+      "loss": 1.0663,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991636543446321e-05,
+      "loss": 1.1105,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991567091087877e-05,
+      "loss": 1.151,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99149735276626e-05,
+      "loss": 1.0932,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914273285015822e-05,
+      "loss": 1.1,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913570183140378e-05,
+      "loss": 1.0762,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912864222239045e-05,
+      "loss": 1.1856,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991215540251542e-05,
+      "loss": 1.1253,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911443724173916e-05,
+      "loss": 1.1024,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991072918741978e-05,
+      "loss": 1.1004,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910011792459086e-05,
+      "loss": 1.1538,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990929153949872e-05,
+      "loss": 1.0391,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908568428746408e-05,
+      "loss": 1.0924,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907842460410682e-05,
+      "loss": 1.1852,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907113634700917e-05,
+      "loss": 1.0874,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906381951827295e-05,
+      "loss": 1.06,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905647412000834e-05,
+      "loss": 1.1373,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990491001543337e-05,
+      "loss": 1.1347,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904169762337567e-05,
+      "loss": 1.0395,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990342665292691e-05,
+      "loss": 1.1078,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 1.1568,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901931866019087e-05,
+      "loss": 1.0717,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901180188953014e-05,
+      "loss": 1.0778,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900425656434263e-05,
+      "loss": 1.2004,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899668268680438e-05,
+      "loss": 1.1344,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989890802590997e-05,
+      "loss": 1.1134,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898144928342105e-05,
+      "loss": 1.1282,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989737897619692e-05,
+      "loss": 1.1728,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989661016969531e-05,
+      "loss": 1.099,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895838509058992e-05,
+      "loss": 1.1758,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895063994510512e-05,
+      "loss": 1.0715,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989428662627324e-05,
+      "loss": 1.0858,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989350640457136e-05,
+      "loss": 1.1499,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 1.0882,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989193740167465e-05,
+      "loss": 1.1285,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989114862093232e-05,
+      "loss": 1.1404,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890356987630362e-05,
+      "loss": 1.0306,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988956250199709e-05,
+      "loss": 1.18,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888765164261626e-05,
+      "loss": 1.0981,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988796497465392e-05,
+      "loss": 1.0724,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887161933404743e-05,
+      "loss": 1.1262,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886356040745683e-05,
+      "loss": 1.0755,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988554729690916e-05,
+      "loss": 1.0775,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884735702128414e-05,
+      "loss": 1.1041,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98839212566375e-05,
+      "loss": 1.1848,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883103960671305e-05,
+      "loss": 1.0529,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 1.1672,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98814608182567e-05,
+      "loss": 1.0958,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880634972282168e-05,
+      "loss": 1.1031,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98798062767801e-05,
+      "loss": 1.0896,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878974731989487e-05,
+      "loss": 1.1177,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878140338150143e-05,
+      "loss": 1.1521,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987730309550271e-05,
+      "loss": 1.0659,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876463004288637e-05,
+      "loss": 1.1334,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98756200647502e-05,
+      "loss": 1.1244,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874774277130512e-05,
+      "loss": 1.1628,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987392564167348e-05,
+      "loss": 1.0715,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987307415862385e-05,
+      "loss": 1.1003,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872219828227187e-05,
+      "loss": 1.1646,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987136265072988e-05,
+      "loss": 1.1181,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 1.0759,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869639755422963e-05,
+      "loss": 1.0958,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868774038110232e-05,
+      "loss": 1.1144,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98679054746906e-05,
+      "loss": 1.0485,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986703406541456e-05,
+      "loss": 1.1448,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866159810533417e-05,
+      "loss": 1.1643,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986528271029931e-05,
+      "loss": 1.1085,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864402764965187e-05,
+      "loss": 1.0959,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863519974784817e-05,
+      "loss": 1.1935,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862634340012796e-05,
+      "loss": 1.082,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861745860904538e-05,
+      "loss": 1.0934,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986085453771627e-05,
+      "loss": 1.0698,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 1.1615,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859063360128753e-05,
+      "loss": 1.1387,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858163506246068e-05,
+      "loss": 1.1009,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985726080931651e-05,
+      "loss": 1.1669,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856355269600415e-05,
+      "loss": 1.0491,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855446887358934e-05,
+      "loss": 1.1452,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985453566285404e-05,
+      "loss": 1.0633,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853621596348525e-05,
+      "loss": 1.1131,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852704688106003e-05,
+      "loss": 1.0675,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851784938390907e-05,
+      "loss": 1.0849,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850862347468486e-05,
+      "loss": 1.2209,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849936915604814e-05,
+      "loss": 1.1144,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849008643066774e-05,
+      "loss": 1.1365,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 1.079,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847143577039263e-05,
+      "loss": 1.1513,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846206784087672e-05,
+      "loss": 1.1503,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845267151537465e-05,
+      "loss": 1.0724,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984432467965963e-05,
+      "loss": 1.1318,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843379368725978e-05,
+      "loss": 1.0598,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984243121900913e-05,
+      "loss": 1.1373,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984148023078252e-05,
+      "loss": 1.1064,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840526404320415e-05,
+      "loss": 1.1189,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839569739897893e-05,
+      "loss": 1.1241,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838610237790853e-05,
+      "loss": 1.1211,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837647898276008e-05,
+      "loss": 1.1127,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836682721630892e-05,
+      "loss": 1.0669,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 1.1369,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983474385806408e-05,
+      "loss": 1.0884,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983377017170154e-05,
+      "loss": 1.1591,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983279364932705e-05,
+      "loss": 1.0951,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831814291222233e-05,
+      "loss": 1.1028,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983083209766953e-05,
+      "loss": 1.1334,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829847068952202e-05,
+      "loss": 1.0561,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828859205354326e-05,
+      "loss": 1.091,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827868507160794e-05,
+      "loss": 1.1721,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826874974657323e-05,
+      "loss": 0.9972,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825878608130438e-05,
+      "loss": 1.1024,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824879407867496e-05,
+      "loss": 1.0901,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823877374156647e-05,
+      "loss": 1.1731,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 1.141,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982186480754801e-05,
+      "loss": 1.1076,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820854275230626e-05,
+      "loss": 1.148,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819840910626174e-05,
+      "loss": 1.1166,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818824714026904e-05,
+      "loss": 1.0561,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817805685725875e-05,
+      "loss": 1.1275,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816783826016982e-05,
+      "loss": 1.1896,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981575913519491e-05,
+      "loss": 1.0725,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814731613555186e-05,
+      "loss": 1.0615,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813701261394136e-05,
+      "loss": 1.142,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812668079008912e-05,
+      "loss": 1.1546,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811632066697478e-05,
+      "loss": 1.0387,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810593224758614e-05,
+      "loss": 1.1391,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 1.1272,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808507053197804e-05,
+      "loss": 1.0837,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807459724177497e-05,
+      "loss": 1.1081,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806409566733043e-05,
+      "loss": 1.1874,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805356581167308e-05,
+      "loss": 1.0642,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804300767783958e-05,
+      "loss": 1.1196,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803242126887496e-05,
+      "loss": 1.0228,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980218065878322e-05,
+      "loss": 1.1441,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801116363777252e-05,
+      "loss": 1.1077,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980004924217654e-05,
+      "loss": 1.0818,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979897929428882e-05,
+      "loss": 1.1249,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979790652042268e-05,
+      "loss": 1.102,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796830920887483e-05,
+      "loss": 1.0776,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979575249599344e-05,
+      "loss": 1.1085,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979467124605156e-05,
+      "loss": 1.1515,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979358717137367e-05,
+      "loss": 1.053,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979250027227241e-05,
+      "loss": 1.0824,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979141054906124e-05,
+      "loss": 1.1698,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790318002054426e-05,
+      "loss": 1.0974,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789222631567058e-05,
+      "loss": 1.105,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788124437915034e-05,
+      "loss": 1.1342,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787023421415065e-05,
+      "loss": 1.0755,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785919582384684e-05,
+      "loss": 1.076,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784812921142232e-05,
+      "loss": 1.0696,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978370343800686e-05,
+      "loss": 1.1868,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782591133298542e-05,
+      "loss": 1.1067,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.1425,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978035806044701e-05,
+      "loss": 1.1231,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779237292947805e-05,
+      "loss": 1.0606,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778113705163665e-05,
+      "loss": 1.1111,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977698729741863e-05,
+      "loss": 1.0931,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977585807003755e-05,
+      "loss": 1.1891,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977472602334609e-05,
+      "loss": 1.0169,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977359115767073e-05,
+      "loss": 1.1065,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772453473338747e-05,
+      "loss": 1.1229,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771312970678258e-05,
+      "loss": 1.138,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977016965001817e-05,
+      "loss": 1.0732,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769023511688215e-05,
+      "loss": 1.1433,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767874556018935e-05,
+      "loss": 1.0802,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766722783341682e-05,
+      "loss": 1.1312,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976556819398862e-05,
+      "loss": 1.0716,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764410788292724e-05,
+      "loss": 1.1874,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976325056658779e-05,
+      "loss": 1.1017,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762087529208414e-05,
+      "loss": 1.0789,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976092167649002e-05,
+      "loss": 1.1144,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975975300876883e-05,
+      "loss": 1.178,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758581526381878e-05,
+      "loss": 1.0203,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757407229667015e-05,
+      "loss": 1.0572,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975623011896291e-05,
+      "loss": 1.1333,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755050194609025e-05,
+      "loss": 1.0492,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753867456945653e-05,
+      "loss": 1.1022,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975268190631389e-05,
+      "loss": 1.0854,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751493543055634e-05,
+      "loss": 1.1381,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975030236751361e-05,
+      "loss": 1.1025,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974910838003135e-05,
+      "loss": 1.0588,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974791158095319e-05,
+      "loss": 1.1085,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746711970624282e-05,
+      "loss": 1.0556,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745509549390585e-05,
+      "loss": 1.1237,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744304317598874e-05,
+      "loss": 1.051,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743096275596735e-05,
+      "loss": 1.1274,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974188542373256e-05,
+      "loss": 1.0725,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740671762355548e-05,
+      "loss": 1.11,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973945529181572e-05,
+      "loss": 1.1222,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738236012463892e-05,
+      "loss": 1.1258,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737013924651706e-05,
+      "loss": 1.1211,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735789028731603e-05,
+      "loss": 1.1426,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973456132505684e-05,
+      "loss": 1.058,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733330813981476e-05,
+      "loss": 1.0846,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732097495860388e-05,
+      "loss": 1.1305,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973086137104926e-05,
+      "loss": 1.1846,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729622439904578e-05,
+      "loss": 1.0714,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728380702783644e-05,
+      "loss": 1.042,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972713616004458e-05,
+      "loss": 1.1148,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97258888120463e-05,
+      "loss": 1.1476,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972463865914853e-05,
+      "loss": 1.0661,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723385701711808e-05,
+      "loss": 1.1193,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972212994009749e-05,
+      "loss": 1.0988,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720871374667714e-05,
+      "loss": 1.0807,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 1.1111,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718345833814504e-05,
+      "loss": 1.0851,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971707885911941e-05,
+      "loss": 1.0497,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715809082065576e-05,
+      "loss": 1.0902,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714536503019204e-05,
+      "loss": 1.0933,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713261122347295e-05,
+      "loss": 1.1368,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711982940417663e-05,
+      "loss": 1.1207,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710701957598926e-05,
+      "loss": 1.0335,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 1.1906,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970813159077268e-05,
+      "loss": 1.0557,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970684220750645e-05,
+      "loss": 1.1035,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970555002483368e-05,
+      "loss": 1.1345,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970425504312703e-05,
+      "loss": 1.1124,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 1.1315,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701656684106764e-05,
+      "loss": 1.0068,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700353307542505e-05,
+      "loss": 1.1157,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699047133443072e-05,
+      "loss": 1.073,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697738162185163e-05,
+      "loss": 1.1143,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696426394146278e-05,
+      "loss": 1.1493,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695111829704724e-05,
+      "loss": 1.1757,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969379446923962e-05,
+      "loss": 1.0217,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969247431313088e-05,
+      "loss": 1.1086,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691151361759232e-05,
+      "loss": 1.1022,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968982561550621e-05,
+      "loss": 1.062,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968849707475415e-05,
+      "loss": 1.1253,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96871657398862e-05,
+      "loss": 1.1058,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 1.0805,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968449468933924e-05,
+      "loss": 1.1263,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683154974430544e-05,
+      "loss": 1.1304,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681812466946595e-05,
+      "loss": 1.1753,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680467167274566e-05,
+      "loss": 1.0688,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967911907580243e-05,
+      "loss": 1.1096,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677768192918973e-05,
+      "loss": 1.0714,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676414519013782e-05,
+      "loss": 1.0986,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675058054477255e-05,
+      "loss": 1.0965,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673698799700582e-05,
+      "loss": 1.0533,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967233675507577e-05,
+      "loss": 1.1724,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967097192099563e-05,
+      "loss": 1.0584,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669604297853766e-05,
+      "loss": 1.072,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668233886044597e-05,
+      "loss": 1.1061,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666860685963344e-05,
+      "loss": 1.1378,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665484698006028e-05,
+      "loss": 1.0743,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664105922569483e-05,
+      "loss": 1.0986,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662724360051334e-05,
+      "loss": 1.1488,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661340010850025e-05,
+      "loss": 1.0926,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965995287536479e-05,
+      "loss": 1.0558,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658562953995676e-05,
+      "loss": 1.2068,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657170247143526e-05,
+      "loss": 1.099,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965577475520999e-05,
+      "loss": 1.0175,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654376478597526e-05,
+      "loss": 1.1119,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652975417709386e-05,
+      "loss": 1.1813,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651571572949627e-05,
+      "loss": 1.0929,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 1.09,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648755533435517e-05,
+      "loss": 1.0682,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647343339493294e-05,
+      "loss": 1.1658,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964592836330372e-05,
+      "loss": 1.091,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964451060527487e-05,
+      "loss": 1.089,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964309006581561e-05,
+      "loss": 1.1662,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641666745335626e-05,
+      "loss": 1.0517,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964024064424539e-05,
+      "loss": 1.0565,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963881176295619e-05,
+      "loss": 1.1367,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637380101880098e-05,
+      "loss": 1.0889,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635945661430006e-05,
+      "loss": 1.1358,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96345084420196e-05,
+      "loss": 1.0898,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963306844406336e-05,
+      "loss": 1.1446,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 1.0951,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630180114175355e-05,
+      "loss": 1.1191,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628731783076565e-05,
+      "loss": 1.1532,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962728067509791e-05,
+      "loss": 1.0468,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625826790657875e-05,
+      "loss": 1.0751,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962437013017576e-05,
+      "loss": 1.088,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622910694071654e-05,
+      "loss": 1.1911,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621448482766455e-05,
+      "loss": 1.0546,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619983496681857e-05,
+      "loss": 1.113,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618515736240353e-05,
+      "loss": 1.1804,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961704520186524e-05,
+      "loss": 1.0605,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615571893980612e-05,
+      "loss": 1.1275,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614095813011366e-05,
+      "loss": 1.1385,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 1.096,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611135333522586e-05,
+      "loss": 1.0886,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609650935856847e-05,
+      "loss": 1.1036,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960816376681406e-05,
+      "loss": 1.0784,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960667382682312e-05,
+      "loss": 1.1013,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605181116313725e-05,
+      "loss": 1.0989,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603685635716356e-05,
+      "loss": 1.1006,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960218738546231e-05,
+      "loss": 1.1517,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600686365983674e-05,
+      "loss": 1.0349,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599182577713327e-05,
+      "loss": 1.0827,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597676021084962e-05,
+      "loss": 1.1569,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596166696533062e-05,
+      "loss": 1.1023,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594654604492908e-05,
+      "loss": 1.0272,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593139745400575e-05,
+      "loss": 1.1269,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591622119692953e-05,
+      "loss": 1.0906,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590101727807703e-05,
+      "loss": 1.0638,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588578570183312e-05,
+      "loss": 1.14,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587052647259044e-05,
+      "loss": 1.167,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958552395947497e-05,
+      "loss": 1.0126,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583992507271958e-05,
+      "loss": 1.0441,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582458291091664e-05,
+      "loss": 1.1059,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580921311376557e-05,
+      "loss": 1.1075,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957938156856989e-05,
+      "loss": 1.1055,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957783906311572e-05,
+      "loss": 1.1073,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576293795458895e-05,
+      "loss": 1.1455,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574745766045065e-05,
+      "loss": 1.1226,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 1.0935,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571641423732965e-05,
+      "loss": 1.1234,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570085111729968e-05,
+      "loss": 1.1394,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956852603976052e-05,
+      "loss": 1.0922,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566964208274254e-05,
+      "loss": 1.059,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565399617721587e-05,
+      "loss": 1.114,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563832268553746e-05,
+      "loss": 1.1057,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956226216122274e-05,
+      "loss": 1.0708,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560689296181388e-05,
+      "loss": 1.0853,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955911367388329e-05,
+      "loss": 1.105,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557535294782854e-05,
+      "loss": 1.1142,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955595415933527e-05,
+      "loss": 1.1067,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554370267996537e-05,
+      "loss": 1.1237,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552783621223437e-05,
+      "loss": 1.0995,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551194219473552e-05,
+      "loss": 1.0065,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954960206320526e-05,
+      "loss": 1.1216,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954800715287773e-05,
+      "loss": 1.1224,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546409488950925e-05,
+      "loss": 1.0767,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544809071885603e-05,
+      "loss": 1.048,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954320590214332e-05,
+      "loss": 1.1632,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954159998018642e-05,
+      "loss": 1.0463,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539991306478046e-05,
+      "loss": 1.0882,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538379881482128e-05,
+      "loss": 1.1038,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536765705663396e-05,
+      "loss": 1.1393,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535148779487365e-05,
+      "loss": 1.0938,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533529103420356e-05,
+      "loss": 1.0432,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531906677929472e-05,
+      "loss": 1.1767,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530281503482614e-05,
+      "loss": 1.0844,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528653580548474e-05,
+      "loss": 1.0852,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527022909596537e-05,
+      "loss": 1.207,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952538949109708e-05,
+      "loss": 1.1254,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523753325521174e-05,
+      "loss": 1.089,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522114413340683e-05,
+      "loss": 1.0961,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520472755028256e-05,
+      "loss": 1.0918,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518828351057345e-05,
+      "loss": 1.1199,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517181201902183e-05,
+      "loss": 1.0414,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951553130803781e-05,
+      "loss": 1.1179,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513878669940035e-05,
+      "loss": 1.0691,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512223288085476e-05,
+      "loss": 1.0687,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 1.0954,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508904295016415e-05,
+      "loss": 1.1164,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507240684759095e-05,
+      "loss": 1.0135,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505574332659353e-05,
+      "loss": 1.1091,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950390523919776e-05,
+      "loss": 1.1713,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502233404855672e-05,
+      "loss": 1.09,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500558830115236e-05,
+      "loss": 1.0876,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498881515459395e-05,
+      "loss": 1.138,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949720146137188e-05,
+      "loss": 1.0435,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495518668337204e-05,
+      "loss": 1.1039,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493833136840683e-05,
+      "loss": 1.0818,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949214486736841e-05,
+      "loss": 1.1647,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949045386040728e-05,
+      "loss": 1.1016,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 1.055,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487063635969942e-05,
+      "loss": 1.0494,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 1.1534,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948366246743956e-05,
+      "loss": 1.0139,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948195778036509e-05,
+      "loss": 1.0916,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480250358739667e-05,
+      "loss": 1.1069,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94785402030557e-05,
+      "loss": 1.1006,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476827313806395e-05,
+      "loss": 0.9595,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475111691485737e-05,
+      "loss": 1.151,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473393336588508e-05,
+      "loss": 1.067,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471672249610268e-05,
+      "loss": 1.0941,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946994843104737e-05,
+      "loss": 1.1174,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946822188139696e-05,
+      "loss": 1.0872,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466492601156964e-05,
+      "loss": 1.0706,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94647605908261e-05,
+      "loss": 1.0603,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463025850903866e-05,
+      "loss": 1.1373,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946128838189056e-05,
+      "loss": 1.0454,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459548184287254e-05,
+      "loss": 1.0854,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457805258595814e-05,
+      "loss": 1.1081,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945605960531889e-05,
+      "loss": 1.1319,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454311224959927e-05,
+      "loss": 1.0728,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945256011802314e-05,
+      "loss": 1.1131,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945080628501355e-05,
+      "loss": 1.0991,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944904972643694e-05,
+      "loss": 1.0626,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447290442799906e-05,
+      "loss": 1.047,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445528434609812e-05,
+      "loss": 1.0683,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 1.1714,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441996246603848e-05,
+      "loss": 1.056,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440226067806646e-05,
+      "loss": 1.1262,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438453166493713e-05,
+      "loss": 1.0628,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436677543176347e-05,
+      "loss": 1.1498,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434899198366632e-05,
+      "loss": 1.0394,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433118132577432e-05,
+      "loss": 1.1278,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431334346322398e-05,
+      "loss": 1.0964,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429547840115967e-05,
+      "loss": 1.0648,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427758614473356e-05,
+      "loss": 1.0567,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425966669910572e-05,
+      "loss": 1.1521,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94241720069444e-05,
+      "loss": 1.0811,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422374626092414e-05,
+      "loss": 1.0894,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942057452787297e-05,
+      "loss": 1.1305,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418771712805203e-05,
+      "loss": 1.0555,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416966181409047e-05,
+      "loss": 1.074,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94151579342052e-05,
+      "loss": 1.121,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413346971715152e-05,
+      "loss": 1.106,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941153329446118e-05,
+      "loss": 1.0985,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409716902966336e-05,
+      "loss": 1.1339,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940789779775446e-05,
+      "loss": 1.1032,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406075979350175e-05,
+      "loss": 1.1245,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404251448278878e-05,
+      "loss": 1.0746,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402424205066766e-05,
+      "loss": 0.9943,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94005942502408e-05,
+      "loss": 1.1599,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398761584328726e-05,
+      "loss": 1.153,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 1.072,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395088121361187e-05,
+      "loss": 1.128,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393247325365126e-05,
+      "loss": 1.0984,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939140382040178e-05,
+      "loss": 1.0819,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389557607002808e-05,
+      "loss": 1.0704,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387708685700648e-05,
+      "loss": 1.1137,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938585705702852e-05,
+      "loss": 1.138,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384002721520423e-05,
+      "loss": 1.0871,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938214567971114e-05,
+      "loss": 1.0596,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380285932136235e-05,
+      "loss": 1.1462,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378423479332045e-05,
+      "loss": 1.0204,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93765583218357e-05,
+      "loss": 1.1398,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374690460185094e-05,
+      "loss": 1.0871,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937281989491892e-05,
+      "loss": 1.0458,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370946626576624e-05,
+      "loss": 1.0628,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369070655698462e-05,
+      "loss": 1.1309,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936719198282545e-05,
+      "loss": 1.074,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936531060849939e-05,
+      "loss": 1.0933,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363426533262856e-05,
+      "loss": 1.1025,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361539757659212e-05,
+      "loss": 1.0861,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935965028223259e-05,
+      "loss": 1.0835,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357758107527913e-05,
+      "loss": 1.1062,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355863234090864e-05,
+      "loss": 1.0811,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353965662467926e-05,
+      "loss": 1.1197,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352065393206346e-05,
+      "loss": 1.133,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 1.0209,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348256763960146e-05,
+      "loss": 1.1293,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934634840507392e-05,
+      "loss": 1.114,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934443735074583e-05,
+      "loss": 0.9978,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342523601527013e-05,
+      "loss": 1.12,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340607157969393e-05,
+      "loss": 1.0551,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338688020625658e-05,
+      "loss": 1.1031,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933676619004928e-05,
+      "loss": 1.1146,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334841666794503e-05,
+      "loss": 1.1603,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933291445141635e-05,
+      "loss": 1.0241,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933098454447062e-05,
+      "loss": 1.1723,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329051946513898e-05,
+      "loss": 1.0899,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327116658103523e-05,
+      "loss": 1.0866,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932517867979763e-05,
+      "loss": 1.0738,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323238012155125e-05,
+      "loss": 1.1237,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932129465573568e-05,
+      "loss": 1.1365,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319348611099754e-05,
+      "loss": 1.0943,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317399878808575e-05,
+      "loss": 1.0859,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931544845942415e-05,
+      "loss": 1.1007,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931349435350926e-05,
+      "loss": 1.1518,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311537561627456e-05,
+      "loss": 1.0839,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930957808434307e-05,
+      "loss": 1.0778,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307615922221205e-05,
+      "loss": 1.0417,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305651075827734e-05,
+      "loss": 1.052,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930368354572932e-05,
+      "loss": 1.102,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301713332493386e-05,
+      "loss": 1.1557,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299740436688124e-05,
+      "loss": 1.1223,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 1.0116,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295786599646303e-05,
+      "loss": 1.0454,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293805659550013e-05,
+      "loss": 1.1678,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291822039164934e-05,
+      "loss": 1.0584,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289835739063133e-05,
+      "loss": 1.1367,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287846759817452e-05,
+      "loss": 1.1452,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285855102001498e-05,
+      "loss": 1.0651,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928386076618966e-05,
+      "loss": 1.0382,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281863752957095e-05,
+      "loss": 1.0673,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927986406287973e-05,
+      "loss": 1.1274,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277861696534268e-05,
+      "loss": 1.0939,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927585665449818e-05,
+      "loss": 1.0272,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273848937349712e-05,
+      "loss": 1.185,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 1.1364,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926982548003246e-05,
+      "loss": 1.0296,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926780974102403e-05,
+      "loss": 1.098,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926579132922391e-05,
+      "loss": 1.1208,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263770245214196e-05,
+      "loss": 1.1055,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261746489577767e-05,
+      "loss": 1.0791,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925972006289826e-05,
+      "loss": 1.1231,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257690965760085e-05,
+      "loss": 1.0786,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255659198748428e-05,
+      "loss": 1.0838,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925362476244924e-05,
+      "loss": 1.0675,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925158765744924e-05,
+      "loss": 1.0859,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249547884335917e-05,
+      "loss": 1.0907,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247505443697537e-05,
+      "loss": 1.0676,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245460336123136e-05,
+      "loss": 1.1694,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92434125622025e-05,
+      "loss": 1.0948,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92413621225262e-05,
+      "loss": 1.0199,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923930901768558e-05,
+      "loss": 1.0681,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923725324827274e-05,
+      "loss": 1.1299,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235194814880554e-05,
+      "loss": 1.0809,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923313371810267e-05,
+      "loss": 1.1019,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923106995853349e-05,
+      "loss": 1.1004,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229003536768197e-05,
+      "loss": 1.0635,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922693445340274e-05,
+      "loss": 1.0429,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224862709033823e-05,
+      "loss": 1.1612,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922278830425894e-05,
+      "loss": 1.0578,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220711239676325e-05,
+      "loss": 1.1214,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218631515885007e-05,
+      "loss": 1.0746,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216549133484758e-05,
+      "loss": 1.0905,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214464093076136e-05,
+      "loss": 1.0921,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212376395260447e-05,
+      "loss": 1.1033,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210286040639783e-05,
+      "loss": 1.1219,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208193029816984e-05,
+      "loss": 1.0224,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206097363395668e-05,
+      "loss": 1.1043,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203999041980215e-05,
+      "loss": 1.0424,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920189806617577e-05,
+      "loss": 1.1623,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 1.0815,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197688153824315e-05,
+      "loss": 1.0585,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195579218491423e-05,
+      "loss": 1.1756,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193467631197776e-05,
+      "loss": 1.0634,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 1.1172,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918923650316487e-05,
+      "loss": 1.1256,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187116963645845e-05,
+      "loss": 1.0762,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918499477460654e-05,
+      "loss": 1.1111,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182869936658978e-05,
+      "loss": 1.092,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180742450415962e-05,
+      "loss": 1.1623,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917861231649104e-05,
+      "loss": 1.0321,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917647953549854e-05,
+      "loss": 1.1172,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174344108053537e-05,
+      "loss": 1.0819,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172206034771884e-05,
+      "loss": 1.1399,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917006531627019e-05,
+      "loss": 1.0719,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167921953165827e-05,
+      "loss": 1.0763,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165775946076927e-05,
+      "loss": 1.1239,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163627295622397e-05,
+      "loss": 1.0298,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916147600242189e-05,
+      "loss": 1.0696,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159322067095833e-05,
+      "loss": 1.1145,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915716549026541e-05,
+      "loss": 1.0627,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915500627255256e-05,
+      "loss": 1.1002,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152844414580003e-05,
+      "loss": 1.0995,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91506799169712e-05,
+      "loss": 1.0185,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148512780350384e-05,
+      "loss": 1.0847,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146343005342546e-05,
+      "loss": 1.0821,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144170592573442e-05,
+      "loss": 1.0812,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914199554266958e-05,
+      "loss": 1.1191,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913981785625824e-05,
+      "loss": 1.1059,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913763753396745e-05,
+      "loss": 1.0474,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.134,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133268984263473e-05,
+      "loss": 1.1093,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131080758110147e-05,
+      "loss": 1.1265,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128889898597117e-05,
+      "loss": 1.1486,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126696406356206e-05,
+      "loss": 1.0263,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124500282020016e-05,
+      "loss": 1.056,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912230152622189e-05,
+      "loss": 1.1562,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912010013959594e-05,
+      "loss": 1.0462,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117896122777044e-05,
+      "loss": 1.0808,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115689476400817e-05,
+      "loss": 1.0843,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113480201103658e-05,
+      "loss": 1.0963,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91112682975227e-05,
+      "loss": 1.162,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910905376629585e-05,
+      "loss": 1.1245,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910683660806177e-05,
+      "loss": 1.0451,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104616823459882e-05,
+      "loss": 1.0885,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102394413130348e-05,
+      "loss": 1.0963,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910016937771411e-05,
+      "loss": 1.0678,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097941717852855e-05,
+      "loss": 1.1353,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095711434189032e-05,
+      "loss": 1.073,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909347852736584e-05,
+      "loss": 1.1064,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909124299802724e-05,
+      "loss": 1.0896,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089004846817947e-05,
+      "loss": 1.0415,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086764074383435e-05,
+      "loss": 1.1055,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908452068136993e-05,
+      "loss": 1.0343,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082274668424423e-05,
+      "loss": 1.2068,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908002603619464e-05,
+      "loss": 1.1028,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907777478532909e-05,
+      "loss": 0.9669,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907552091647701e-05,
+      "loss": 1.1165,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073264430288415e-05,
+      "loss": 1.1337,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071005327414055e-05,
+      "loss": 1.0706,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068743608505454e-05,
+      "loss": 1.1205,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906647927421488e-05,
+      "loss": 1.1224,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064212325195348e-05,
+      "loss": 1.0313,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906194276210064e-05,
+      "loss": 1.0754,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905967058558529e-05,
+      "loss": 1.1241,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905739579630458e-05,
+      "loss": 1.1266,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055118394914545e-05,
+      "loss": 1.0636,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052838382071984e-05,
+      "loss": 1.0612,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050555758434433e-05,
+      "loss": 1.1101,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048270524660197e-05,
+      "loss": 1.1276,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045982681408324e-05,
+      "loss": 1.0422,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043692229338612e-05,
+      "loss": 1.0498,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041399169111626e-05,
+      "loss": 1.1539,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039103501388668e-05,
+      "loss": 1.1191,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036805226831795e-05,
+      "loss": 1.0527,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034504346103825e-05,
+      "loss": 1.1248,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032200859868316e-05,
+      "loss": 1.0954,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029894768789582e-05,
+      "loss": 1.0557,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902758607353269e-05,
+      "loss": 1.0373,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025274774763457e-05,
+      "loss": 1.1311,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902296087314845e-05,
+      "loss": 1.1074,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020644369354987e-05,
+      "loss": 1.0404,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901832526405114e-05,
+      "loss": 1.1144,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901600355790572e-05,
+      "loss": 1.0664,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013679251588304e-05,
+      "loss": 1.0797,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011352345769204e-05,
+      "loss": 1.16,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009022841119496e-05,
+      "loss": 1.105,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006690738310988e-05,
+      "loss": 1.063,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004356038016255e-05,
+      "loss": 1.1119,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900201874090861e-05,
+      "loss": 1.0788,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999678847662124e-05,
+      "loss": 1.0705,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89973363589516e-05,
+      "loss": 1.0675,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994991275452613e-05,
+      "loss": 1.1576,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899264359784146e-05,
+      "loss": 1.0917,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990293326795212e-05,
+      "loss": 1.081,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987940462991673e-05,
+      "loss": 1.1398,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898558500710939e-05,
+      "loss": 1.0583,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983226959827676e-05,
+      "loss": 1.1046,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980866321826574e-05,
+      "loss": 1.0566,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978503093786882e-05,
+      "loss": 1.0539,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976137276390145e-05,
+      "loss": 1.1755,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897376887031865e-05,
+      "loss": 1.0987,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971397876255433e-05,
+      "loss": 0.9947,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969024294884285e-05,
+      "loss": 1.0939,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966648126889727e-05,
+      "loss": 1.0415,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896426937295704e-05,
+      "loss": 1.0797,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961888033772243e-05,
+      "loss": 1.1221,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959504110022104e-05,
+      "loss": 1.1274,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 1.067,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954728511576587e-05,
+      "loss": 1.1398,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895233683825847e-05,
+      "loss": 1.0552,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949942583129527e-05,
+      "loss": 1.0471,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947545746880257e-05,
+      "loss": 1.088,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945146330201885e-05,
+      "loss": 1.1797,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89427443337864e-05,
+      "loss": 1.0215,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894033975832652e-05,
+      "loss": 1.1105,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937932604515717e-05,
+      "loss": 1.0716,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89355228730482e-05,
+      "loss": 1.0887,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893311056461893e-05,
+      "loss": 1.0971,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930695679923594e-05,
+      "loss": 1.1053,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892827821965864e-05,
+      "loss": 1.135,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892585818452126e-05,
+      "loss": 1.0851,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923435575209363e-05,
+      "loss": 1.0776,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921010392421628e-05,
+      "loss": 1.1492,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891858263685747e-05,
+      "loss": 1.0797,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891615230921703e-05,
+      "loss": 1.032,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891371941020121e-05,
+      "loss": 1.0701,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911283940511652e-05,
+      "loss": 1.112,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908845900850725e-05,
+      "loss": 1.1232,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890640529192155e-05,
+      "loss": 1.0766,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903962114427985e-05,
+      "loss": 1.051,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901516369074636e-05,
+      "loss": 1.1171,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889906805656684e-05,
+      "loss": 1.0878,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896617177610677e-05,
+      "loss": 1.0487,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889416373291298e-05,
+      "loss": 1.0726,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891707723181294e-05,
+      "loss": 1.0751,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888924914912393e-05,
+      "loss": 1.1016,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886788011449928e-05,
+      "loss": 1.0924,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884324310869068e-05,
+      "loss": 1.1316,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888185804809187e-05,
+      "loss": 0.9799,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879389223829592e-05,
+      "loss": 1.1069,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876917838794226e-05,
+      "loss": 1.1154,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874443893698514e-05,
+      "loss": 1.0631,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871967389255927e-05,
+      "loss": 1.1068,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869488326180682e-05,
+      "loss": 1.0584,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886700670518772e-05,
+      "loss": 1.0778,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864522526992735e-05,
+      "loss": 1.0998,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862035792312148e-05,
+      "loss": 0.9691,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859546501863125e-05,
+      "loss": 1.1825,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857054656363564e-05,
+      "loss": 1.0608,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 1.0404,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852063303088108e-05,
+      "loss": 1.149,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849563796751695e-05,
+      "loss": 1.1107,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884706173824371e-05,
+      "loss": 1.1011,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884455712828573e-05,
+      "loss": 1.1273,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842049967600075e-05,
+      "loss": 1.0333,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88395402569098e-05,
+      "loss": 1.1314,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883702799693869e-05,
+      "loss": 1.0924,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883451318841127e-05,
+      "loss": 1.1704,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831995832052802e-05,
+      "loss": 1.0218,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 1.1071,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826953478747412e-05,
+      "loss": 1.0784,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824428483254687e-05,
+      "loss": 1.1606,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821900942839292e-05,
+      "loss": 1.0984,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819370858230153e-05,
+      "loss": 1.0633,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881683823015694e-05,
+      "loss": 1.1482,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881430305935005e-05,
+      "loss": 1.0141,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811765346540612e-05,
+      "loss": 1.0784,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809225092460488e-05,
+      "loss": 1.1252,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880668229784228e-05,
+      "loss": 1.0436,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804136963419316e-05,
+      "loss": 1.1012,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801589089925656e-05,
+      "loss": 1.1101,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799038678096097e-05,
+      "loss": 1.0826,
+      "step": 1103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879648572866617e-05,
+      "loss": 1.0491,
+      "step": 1104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793930242372117e-05,
+      "loss": 1.1462,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879137221995095e-05,
+      "loss": 1.047,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788811662140372e-05,
+      "loss": 1.1791,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786248569678847e-05,
+      "loss": 1.0227,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783682943305552e-05,
+      "loss": 1.0669,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781114783760403e-05,
+      "loss": 1.0626,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778544091784047e-05,
+      "loss": 1.1252,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775970868117856e-05,
+      "loss": 1.0688,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773395113503938e-05,
+      "loss": 1.081,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877081682868513e-05,
+      "loss": 1.1021,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768236014404994e-05,
+      "loss": 1.0544,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876565267140782e-05,
+      "loss": 1.0341,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 1.122,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760478402243198e-05,
+      "loss": 1.0426,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757887477567984e-05,
+      "loss": 1.0092,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755294027160203e-05,
+      "loss": 1.0979,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87526980517678e-05,
+      "loss": 1.1245,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750099552139432e-05,
+      "loss": 1.1469,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87474985290245e-05,
+      "loss": 1.0556,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874489498317313e-05,
+      "loss": 1.0926,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742288915336165e-05,
+      "loss": 1.035,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873968032626518e-05,
+      "loss": 1.0656,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737069216712488e-05,
+      "loss": 1.1944,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734455587431117e-05,
+      "loss": 0.9802,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731839439174825e-05,
+      "loss": 1.1325,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729220772698096e-05,
+      "loss": 1.0309,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726599588756144e-05,
+      "loss": 1.1133,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87239758881049e-05,
+      "loss": 1.0795,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721349671501033e-05,
+      "loss": 1.0849,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718720939701925e-05,
+      "loss": 1.1068,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716089693465696e-05,
+      "loss": 1.0891,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713455933551176e-05,
+      "loss": 1.0714,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710819660717936e-05,
+      "loss": 1.035,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708180875726266e-05,
+      "loss": 1.144,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705539579337172e-05,
+      "loss": 1.1055,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702895772312396e-05,
+      "loss": 1.0668,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700249455414394e-05,
+      "loss": 1.1082,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869760062940636e-05,
+      "loss": 1.0646,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 1.0142,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869229545311653e-05,
+      "loss": 1.155,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689639104364727e-05,
+      "loss": 1.1078,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868698024956286e-05,
+      "loss": 1.1006,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868431888947773e-05,
+      "loss": 1.0626,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681655024876864e-05,
+      "loss": 1.1586,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678988656528504e-05,
+      "loss": 1.0173,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676319785201617e-05,
+      "loss": 1.1334,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673648411665895e-05,
+      "loss": 1.035,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670974536691748e-05,
+      "loss": 1.1236,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668298161050308e-05,
+      "loss": 1.0938,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665619285513436e-05,
+      "loss": 1.0908,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662937910853697e-05,
+      "loss": 1.1116,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 1.1141,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657567667259527e-05,
+      "loss": 1.0298,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654878799873856e-05,
+      "loss": 1.1264,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652187436462825e-05,
+      "loss": 1.003,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864949357780261e-05,
+      "loss": 1.0968,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864679722467011e-05,
+      "loss": 1.0608,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644098377842934e-05,
+      "loss": 1.1216,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641397038099426e-05,
+      "loss": 1.0614,
+      "step": 1163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638693206218634e-05,
+      "loss": 1.0692,
+      "step": 1164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635986882980325e-05,
+      "loss": 1.1218,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633278069165e-05,
+      "loss": 1.1592,
+      "step": 1166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630566765553864e-05,
+      "loss": 1.0636,
+      "step": 1167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862785297292884e-05,
+      "loss": 1.06,
+      "step": 1168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625136692072577e-05,
+      "loss": 1.1538,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622417923768433e-05,
+      "loss": 1.0396,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619696668800494e-05,
+      "loss": 1.0577,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861697292795355e-05,
+      "loss": 1.1069,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614246702013122e-05,
+      "loss": 1.0716,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611517991765433e-05,
+      "loss": 1.1039,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608786797997436e-05,
+      "loss": 1.0498,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606053121496793e-05,
+      "loss": 1.0824,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860331696305188e-05,
+      "loss": 1.1093,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860057832345179e-05,
+      "loss": 1.0191,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859783720348634e-05,
+      "loss": 1.0741,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595093603946053e-05,
+      "loss": 1.1158,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859234752562217e-05,
+      "loss": 1.0558,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589598969306646e-05,
+      "loss": 1.1218,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586847935792152e-05,
+      "loss": 1.1431,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584094425872073e-05,
+      "loss": 1.0893,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581338440340508e-05,
+      "loss": 0.9628,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578579979992266e-05,
+      "loss": 1.1413,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575819045622883e-05,
+      "loss": 1.0406,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857305563802859e-05,
+      "loss": 1.1198,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570289758006346e-05,
+      "loss": 1.0552,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567521406353813e-05,
+      "loss": 1.1005,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564750583869374e-05,
+      "loss": 1.1284,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856197729135212e-05,
+      "loss": 1.0752,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559201529601854e-05,
+      "loss": 1.1035,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556423299419096e-05,
+      "loss": 1.0753,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 1.0838,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550859436961716e-05,
+      "loss": 1.087,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548073806291685e-05,
+      "loss": 1.1204,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545285710398343e-05,
+      "loss": 1.038,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854249515008576e-05,
+      "loss": 1.0438,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539702126158724e-05,
+      "loss": 1.1486,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536906639422724e-05,
+      "loss": 1.0803,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534108690683968e-05,
+      "loss": 1.0602,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531308280749374e-05,
+      "loss": 1.1401,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528505410426563e-05,
+      "loss": 1.009,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852570008052387e-05,
+      "loss": 1.0114,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522892291850335e-05,
+      "loss": 1.101,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852008204521572e-05,
+      "loss": 1.105,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851726934143048e-05,
+      "loss": 1.0634,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514454181305784e-05,
+      "loss": 1.0235,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851163656565351e-05,
+      "loss": 1.0767,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850881649528625e-05,
+      "loss": 1.1679,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505993971017303e-05,
+      "loss": 1.0292,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503168993660658e-05,
+      "loss": 1.0796,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850034156403103e-05,
+      "loss": 1.081,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849751168294384e-05,
+      "loss": 1.112,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494679351215212e-05,
+      "loss": 1.0053,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849184456966197e-05,
+      "loss": 1.1356,
+      "step": 1217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489007339101663e-05,
+      "loss": 1.0861,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486167660352523e-05,
+      "loss": 1.0053,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483325534233503e-05,
+      "loss": 1.0876,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 1.1152,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477633943165157e-05,
+      "loss": 1.0609,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474784479857257e-05,
+      "loss": 1.0354,
+      "step": 1223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847193257246233e-05,
+      "loss": 1.0993,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846907822180286e-05,
+      "loss": 1.0825,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846622142870202e-05,
+      "loss": 1.1116,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463362193983704e-05,
+      "loss": 1.0794,
+      "step": 1227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846050051847249e-05,
+      "loss": 1.091,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457636402993678e-05,
+      "loss": 1.076,
+      "step": 1229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454769848373262e-05,
+      "loss": 1.0959,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 1.1009,
+      "step": 1231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449029425015138e-05,
+      "loss": 1.1229,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446155557932935e-05,
+      "loss": 1.1248,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443279255020153e-05,
+      "loss": 1.0567,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84404005171063e-05,
+      "loss": 1.1806,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843751934502159e-05,
+      "loss": 1.0314,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434635739596945e-05,
+      "loss": 1.0643,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843174970166398e-05,
+      "loss": 1.0927,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842886123205501e-05,
+      "loss": 1.0811,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842597033160306e-05,
+      "loss": 1.0526,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423077001141848e-05,
+      "loss": 1.1387,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420181241505805e-05,
+      "loss": 1.0983,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417283053530047e-05,
+      "loss": 1.0494,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414382438050395e-05,
+      "loss": 1.0627,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841147939590338e-05,
+      "loss": 1.0741,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408573927926225e-05,
+      "loss": 1.1106,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405666034956842e-05,
+      "loss": 1.0531,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402755717833866e-05,
+      "loss": 1.0556,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399842977396615e-05,
+      "loss": 1.1892,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396927814485107e-05,
+      "loss": 0.9797,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839401022994006e-05,
+      "loss": 1.0585,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391090224602895e-05,
+      "loss": 1.1146,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388167799315727e-05,
+      "loss": 1.102,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385242954921366e-05,
+      "loss": 1.0136,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382315692263324e-05,
+      "loss": 1.0118,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379386012185813e-05,
+      "loss": 1.1301,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376453915533736e-05,
+      "loss": 1.0664,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83735194031527e-05,
+      "loss": 1.0429,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370582475888992e-05,
+      "loss": 1.1353,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836764313458962e-05,
+      "loss": 1.0861,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364701380102267e-05,
+      "loss": 1.0309,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836175721327533e-05,
+      "loss": 1.1376,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358810634957888e-05,
+      "loss": 1.0918,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835586164599972e-05,
+      "loss": 1.0537,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352910247251298e-05,
+      "loss": 1.0962,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349956439563796e-05,
+      "loss": 1.0782,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834700022378907e-05,
+      "loss": 1.0803,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344041600779688e-05,
+      "loss": 0.9945,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83410805713889e-05,
+      "loss": 1.0986,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338117136470648e-05,
+      "loss": 1.1021,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335151296879576e-05,
+      "loss": 1.0558,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332183053471015e-05,
+      "loss": 1.0764,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 1.1045,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832623935862624e-05,
+      "loss": 1.0612,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832326390890415e-05,
+      "loss": 1.0667,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320286058792845e-05,
+      "loss": 1.1542,
+      "step": 1276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317305809151115e-05,
+      "loss": 1.0436,
+      "step": 1277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314323160838446e-05,
+      "loss": 1.07,
+      "step": 1278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831133811471503e-05,
+      "loss": 1.0572,
+      "step": 1279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830835067164173e-05,
+      "loss": 1.1387,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305360832480118e-05,
+      "loss": 1.0207,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302368598092443e-05,
+      "loss": 1.0725,
+      "step": 1282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299373969341655e-05,
+      "loss": 1.0729,
+      "step": 1283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296376947091388e-05,
+      "loss": 1.0873,
+      "step": 1284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829337753220597e-05,
+      "loss": 1.0471,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 1.027,
+      "step": 1286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828737152799044e-05,
+      "loss": 1.1803,
+      "step": 1287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284364940392426e-05,
+      "loss": 1.041,
+      "step": 1288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828135596362347e-05,
+      "loss": 1.0736,
+      "step": 1289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278344598551337e-05,
+      "loss": 1.091,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82753308460445e-05,
+      "loss": 1.0592,
+      "step": 1291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272314706972105e-05,
+      "loss": 1.077,
+      "step": 1292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826929618220399e-05,
+      "loss": 1.0665,
+      "step": 1293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826627527261069e-05,
+      "loss": 1.1585,
+      "step": 1294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263251979063413e-05,
+      "loss": 0.9503,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826022630243407e-05,
+      "loss": 1.0779,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825719824359524e-05,
+      "loss": 1.1124,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254167803420215e-05,
+      "loss": 1.0519,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251134982782952e-05,
+      "loss": 1.1016,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248099782558103e-05,
+      "loss": 1.0891,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245062203621003e-05,
+      "loss": 1.1419,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242022246847674e-05,
+      "loss": 1.0942,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823897991311483e-05,
+      "loss": 1.0706,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235935203299866e-05,
+      "loss": 1.0764,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232888118280856e-05,
+      "loss": 1.0888,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 1.0582,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226786826146448e-05,
+      "loss": 1.0515,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223732620790634e-05,
+      "loss": 1.1576,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822067604374994e-05,
+      "loss": 1.0512,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217617095905874e-05,
+      "loss": 1.0582,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821455577814062e-05,
+      "loss": 1.0571,
+      "step": 1311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821149209133704e-05,
+      "loss": 1.117,
+      "step": 1312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208426036378697e-05,
+      "loss": 1.0163,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205357614149823e-05,
+      "loss": 0.988,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820228682553533e-05,
+      "loss": 1.1451,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199213671420827e-05,
+      "loss": 1.0819,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196138152692594e-05,
+      "loss": 1.0911,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193060270237596e-05,
+      "loss": 1.1392,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189980024943476e-05,
+      "loss": 1.0586,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186897417698562e-05,
+      "loss": 1.0427,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818381244939187e-05,
+      "loss": 1.126,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180725120913084e-05,
+      "loss": 1.0867,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817763543315258e-05,
+      "loss": 1.076,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174543387001403e-05,
+      "loss": 1.1097,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171448983351284e-05,
+      "loss": 1.1442,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816835222309464e-05,
+      "loss": 1.0228,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816525310712456e-05,
+      "loss": 1.0787,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816215163633481e-05,
+      "loss": 1.0846,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159047811619843e-05,
+      "loss": 1.0771,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155941633874787e-05,
+      "loss": 1.0987,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152833103995443e-05,
+      "loss": 0.9628,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149722222878306e-05,
+      "loss": 1.1708,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146608991420533e-05,
+      "loss": 1.0597,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143493410519967e-05,
+      "loss": 1.0502,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140375481075122e-05,
+      "loss": 1.1239,
+      "step": 1335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81372552039852e-05,
+      "loss": 1.0944,
+      "step": 1336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134132580150064e-05,
+      "loss": 1.0284,
+      "step": 1337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131007610470278e-05,
+      "loss": 0.9665,
+      "step": 1338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127880295847058e-05,
+      "loss": 1.1074,
+      "step": 1339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124750637182313e-05,
+      "loss": 1.139,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121618635378616e-05,
+      "loss": 1.0461,
+      "step": 1341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811848429133922e-05,
+      "loss": 1.1123,
+      "step": 1342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115347605968063e-05,
+      "loss": 1.0681,
+      "step": 1343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811220858016974e-05,
+      "loss": 1.0455,
+      "step": 1344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810906721484954e-05,
+      "loss": 1.0288,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810592351091341e-05,
+      "loss": 1.1146,
+      "step": 1346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102777469267985e-05,
+      "loss": 0.992,
+      "step": 1347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099629090820565e-05,
+      "loss": 1.0702,
+      "step": 1348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096478376479127e-05,
+      "loss": 1.119,
+      "step": 1349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093325327152324e-05,
+      "loss": 1.0594,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.0648,
+      "step": 1351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087012227180583e-05,
+      "loss": 1.0265,
+      "step": 1352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808385217835632e-05,
+      "loss": 1.1209,
+      "step": 1353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808068979818802e-05,
+      "loss": 1.0297,
+      "step": 1354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077525087587712e-05,
+      "loss": 1.0998,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807435804746807e-05,
+      "loss": 1.1028,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071188678742457e-05,
+      "loss": 1.0661,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068016982324906e-05,
+      "loss": 1.0778,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806484295913012e-05,
+      "loss": 1.0633,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061666610073465e-05,
+      "loss": 1.0898,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058487936070992e-05,
+      "loss": 1.0461,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055306938039413e-05,
+      "loss": 1.0221,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805212361689611e-05,
+      "loss": 1.1222,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804893797355914e-05,
+      "loss": 1.0526,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045750008947232e-05,
+      "loss": 1.0221,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804255972397977e-05,
+      "loss": 1.0711,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039367119576825e-05,
+      "loss": 1.1091,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036172196659118e-05,
+      "loss": 1.0304,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032974956148064e-05,
+      "loss": 1.0936,
+      "step": 1369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802977539896572e-05,
+      "loss": 1.1126,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802657352603483e-05,
+      "loss": 1.0456,
+      "step": 1371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023369338278795e-05,
+      "loss": 1.0519,
+      "step": 1372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020162836621686e-05,
+      "loss": 1.0934,
+      "step": 1373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016954021988248e-05,
+      "loss": 1.0708,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013742895303883e-05,
+      "loss": 1.0685,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010529457494664e-05,
+      "loss": 1.0626,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007313709487334e-05,
+      "loss": 1.1508,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004095652209304e-05,
+      "loss": 0.969,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000875286588635e-05,
+      "loss": 1.0696,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799765261355407e-05,
+      "loss": 1.1291,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 1.0398,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991200348961533e-05,
+      "loss": 1.0069,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987970759264365e-05,
+      "loss": 1.0261,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984738865874902e-05,
+      "loss": 1.1488,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798150466972521e-05,
+      "loss": 1.0823,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978268171748015e-05,
+      "loss": 1.0169,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975029372876706e-05,
+      "loss": 1.1428,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797178827404534e-05,
+      "loss": 1.0266,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968544876188634e-05,
+      "loss": 1.0967,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965299180241963e-05,
+      "loss": 1.0651,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962051187141377e-05,
+      "loss": 1.0959,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958800897823582e-05,
+      "loss": 1.0038,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955548313225936e-05,
+      "loss": 1.1072,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795229343428648e-05,
+      "loss": 1.0994,
+      "step": 1394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949036261943894e-05,
+      "loss": 1.0721,
+      "step": 1395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945776797137544e-05,
+      "loss": 1.0821,
+      "step": 1396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942515040807436e-05,
+      "loss": 1.0135,
+      "step": 1397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939250993894246e-05,
+      "loss": 1.141,
+      "step": 1398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793598465733931e-05,
+      "loss": 1.061,
+      "step": 1399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932716032084618e-05,
+      "loss": 1.0852,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929445119072837e-05,
+      "loss": 1.1398,
+      "step": 1401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926171919247275e-05,
+      "loss": 1.0483,
+      "step": 1402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792289643355191e-05,
+      "loss": 1.0865,
+      "step": 1403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919618662931373e-05,
+      "loss": 1.0381,
+      "step": 1404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791633860833096e-05,
+      "loss": 1.0936,
+      "step": 1405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791305627069662e-05,
+      "loss": 1.0708,
+      "step": 1406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790977165097497e-05,
+      "loss": 1.042,
+      "step": 1407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906484750113267e-05,
+      "loss": 1.0867,
+      "step": 1408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903195569059453e-05,
+      "loss": 1.0563,
+      "step": 1409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78999041087621e-05,
+      "loss": 0.9942,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896610370170452e-05,
+      "loss": 1.0112,
+      "step": 1411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789331435423441e-05,
+      "loss": 1.1206,
+      "step": 1412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890016061904522e-05,
+      "loss": 1.1132,
+      "step": 1413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886715494132008e-05,
+      "loss": 0.9687,
+      "step": 1414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883412651868735e-05,
+      "loss": 1.143,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 1.0656,
+      "step": 1416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787680014768065e-05,
+      "loss": 1.0653,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873490487662858e-05,
+      "loss": 1.0408,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787017855696833e-05,
+      "loss": 1.154,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866864356552215e-05,
+      "loss": 1.1053,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786354788737031e-05,
+      "loss": 1.049,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786022915037907e-05,
+      "loss": 1.1139,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856908146535602e-05,
+      "loss": 1.0918,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785358487679767e-05,
+      "loss": 1.0565,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850259342123687e-05,
+      "loss": 1.0143,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846931543472722e-05,
+      "loss": 1.0994,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843601481804493e-05,
+      "loss": 1.1027,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840269158079378e-05,
+      "loss": 1.044,
+      "step": 1428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 1.1427,
+      "step": 1429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833597728303238e-05,
+      "loss": 1.0163,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830258624176224e-05,
+      "loss": 1.0737,
+      "step": 1431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826917261840337e-05,
+      "loss": 1.0947,
+      "step": 1432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782357364225921e-05,
+      "loss": 1.0591,
+      "step": 1433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782022776639713e-05,
+      "loss": 1.0559,
+      "step": 1434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816879635219028e-05,
+      "loss": 1.0045,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781352924969049e-05,
+      "loss": 1.0938,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781017661077775e-05,
+      "loss": 1.0627,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806821719447696e-05,
+      "loss": 1.0829,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780346457666786e-05,
+      "loss": 1.1229,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800105183406424e-05,
+      "loss": 1.0856,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796743540632226e-05,
+      "loss": 0.9835,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793379649314743e-05,
+      "loss": 1.0863,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790013510424107e-05,
+      "loss": 1.1724,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786645124931095e-05,
+      "loss": 1.0164,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783274493807135e-05,
+      "loss": 1.0662,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779901618024303e-05,
+      "loss": 1.1227,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777652649855531e-05,
+      "loss": 0.9836,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773149136373536e-05,
+      "loss": 1.1019,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776976953245299e-05,
+      "loss": 1.1014,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766387687768338e-05,
+      "loss": 1.1633,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776300360329488e-05,
+      "loss": 1.0646,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759617280008576e-05,
+      "loss": 1.0586,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756228718886025e-05,
+      "loss": 1.0763,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752837920904468e-05,
+      "loss": 1.0731,
+      "step": 1454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749444887041797e-05,
+      "loss": 1.033,
+      "step": 1455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746049618276545e-05,
+      "loss": 1.0734,
+      "step": 1456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77426521155879e-05,
+      "loss": 1.1398,
+      "step": 1457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773925237995567e-05,
+      "loss": 1.0219,
+      "step": 1458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735850412360332e-05,
+      "loss": 1.0768,
+      "step": 1459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732446213782997e-05,
+      "loss": 1.1715,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772903978520542e-05,
+      "loss": 1.0794,
+      "step": 1461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725631127609993e-05,
+      "loss": 1.0828,
+      "step": 1462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722220241979762e-05,
+      "loss": 1.0135,
+      "step": 1463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718807129298407e-05,
+      "loss": 1.133,
+      "step": 1464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715391790550255e-05,
+      "loss": 1.1068,
+      "step": 1465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711974226720272e-05,
+      "loss": 1.0348,
+      "step": 1466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708554438794067e-05,
+      "loss": 1.1152,
+      "step": 1467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705132427757895e-05,
+      "loss": 1.0633,
+      "step": 1468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770170819459864e-05,
+      "loss": 1.0442,
+      "step": 1469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698281740303836e-05,
+      "loss": 1.0958,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769485306586166e-05,
+      "loss": 1.1065,
+      "step": 1471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691422172260923e-05,
+      "loss": 1.0626,
+      "step": 1472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768798906049108e-05,
+      "loss": 1.0746,
+      "step": 1473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768455373154222e-05,
+      "loss": 1.0678,
+      "step": 1474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681116186405073e-05,
+      "loss": 1.0347,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767767642607102e-05,
+      "loss": 1.1072,
+      "step": 1476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674234451532065e-05,
+      "loss": 1.0028,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767079026378086e-05,
+      "loss": 1.0981,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766734386381068e-05,
+      "loss": 1.0922,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766389525261547e-05,
+      "loss": 1.0629,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 1.087,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656991400528815e-05,
+      "loss": 1.0985,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765353616162841e-05,
+      "loss": 1.096,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765007871548504e-05,
+      "loss": 1.0837,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646619063095816e-05,
+      "loss": 1.0901,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643157205458483e-05,
+      "loss": 1.0905,
+      "step": 1486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763969314357143e-05,
+      "loss": 0.979,
+      "step": 1487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636226878433668e-05,
+      "loss": 1.1132,
+      "step": 1488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632758411044855e-05,
+      "loss": 1.0692,
+      "step": 1489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629287742405285e-05,
+      "loss": 1.0691,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762581487351587e-05,
+      "loss": 1.0572,
+      "step": 1491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762233980537818e-05,
+      "loss": 1.0778,
+      "step": 1492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618862538994404e-05,
+      "loss": 1.0736,
+      "step": 1493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761538307536737e-05,
+      "loss": 1.0929,
+      "step": 1494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611901415500536e-05,
+      "loss": 1.1127,
+      "step": 1495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608417560397996e-05,
+      "loss": 1.0013,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604931511064477e-05,
+      "loss": 1.0819,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601443268505343e-05,
+      "loss": 1.0557,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597952833726583e-05,
+      "loss": 1.1365,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594460207734823e-05,
+      "loss": 1.0899,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590965391537316e-05,
+      "loss": 1.0253,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587468386141954e-05,
+      "loss": 1.0133,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758396919255726e-05,
+      "loss": 1.0353,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580467811792374e-05,
+      "loss": 1.0795,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576964244857082e-05,
+      "loss": 1.0754,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573458492761802e-05,
+      "loss": 1.1012,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 1.0375,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566440437136053e-05,
+      "loss": 1.0068,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562928135629566e-05,
+      "loss": 1.0907,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559413653011027e-05,
+      "loss": 1.1093,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555896990294003e-05,
+      "loss": 1.058,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755237814849268e-05,
+      "loss": 1.0825,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548857128621878e-05,
+      "loss": 1.1037,
+      "step": 1513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545333931697038e-05,
+      "loss": 1.0558,
+      "step": 1514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541808558734236e-05,
+      "loss": 1.0798,
+      "step": 1515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753828101075017e-05,
+      "loss": 1.1244,
+      "step": 1516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753475128876217e-05,
+      "loss": 1.0713,
+      "step": 1517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753121939378819e-05,
+      "loss": 1.0504,
+      "step": 1518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527685326846816e-05,
+      "loss": 1.0883,
+      "step": 1519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524149088957244e-05,
+      "loss": 0.9807,
+      "step": 1520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520610681139322e-05,
+      "loss": 1.0464,
+      "step": 1521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517070104413497e-05,
+      "loss": 1.1192,
+      "step": 1522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513527359800865e-05,
+      "loss": 1.133,
+      "step": 1523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509982448323132e-05,
+      "loss": 0.9934,
+      "step": 1524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506435371002635e-05,
+      "loss": 1.0386,
+      "step": 1525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750288612886233e-05,
+      "loss": 1.1468,
+      "step": 1526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499334722925805e-05,
+      "loss": 1.0499,
+      "step": 1527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495781154217267e-05,
+      "loss": 1.0307,
+      "step": 1528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492225423761547e-05,
+      "loss": 1.0659,
+      "step": 1529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488667532584106e-05,
+      "loss": 1.0777,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 1.075,
+      "step": 1531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481545272168977e-05,
+      "loss": 1.0912,
+      "step": 1532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747798090498532e-05,
+      "loss": 1.1336,
+      "step": 1533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747441438118799e-05,
+      "loss": 1.0126,
+      "step": 1534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470845701805552e-05,
+      "loss": 1.0356,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74672748678672e-05,
+      "loss": 1.0178,
+      "step": 1536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463701880402738e-05,
+      "loss": 1.078,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746012674044261e-05,
+      "loss": 1.0536,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745654944901786e-05,
+      "loss": 1.1302,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745297000716016e-05,
+      "loss": 1.0615,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744938841590181e-05,
+      "loss": 1.064,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445804676275722e-05,
+      "loss": 1.0446,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442218789315434e-05,
+      "loss": 1.0682,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438630756055088e-05,
+      "loss": 1.1415,
+      "step": 1544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743504057752946e-05,
+      "loss": 0.981,
+      "step": 1545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 1.1301,
+      "step": 1546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427853788824546e-05,
+      "loss": 1.0805,
+      "step": 1547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742425718071789e-05,
+      "loss": 0.98,
+      "step": 1548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420658431491224e-05,
+      "loss": 1.0124,
+      "step": 1549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417057542182407e-05,
+      "loss": 1.1146,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741345451382992e-05,
+      "loss": 1.0496,
+      "step": 1551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740984934747286e-05,
+      "loss": 1.0679,
+      "step": 1552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740624204415094e-05,
+      "loss": 1.0506,
+      "step": 1553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402632604904486e-05,
+      "loss": 1.1234,
+      "step": 1554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399021030774443e-05,
+      "loss": 1.0319,
+      "step": 1555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395407322802374e-05,
+      "loss": 1.0624,
+      "step": 1556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391791482030455e-05,
+      "loss": 1.1573,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388173509501475e-05,
+      "loss": 1.0412,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384553406258842e-05,
+      "loss": 1.0379,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738093117334657e-05,
+      "loss": 1.0365,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377306811809306e-05,
+      "loss": 1.121,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373680322692283e-05,
+      "loss": 1.0383,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370051707041377e-05,
+      "loss": 1.0738,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366420965903056e-05,
+      "loss": 1.1155,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362788100324408e-05,
+      "loss": 1.0975,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359153111353136e-05,
+      "loss": 1.0311,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355516000037555e-05,
+      "loss": 1.0561,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351876767426585e-05,
+      "loss": 1.1178,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348235414569766e-05,
+      "loss": 1.0247,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734459194251725e-05,
+      "loss": 1.0705,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340946352319795e-05,
+      "loss": 1.1052,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337298645028764e-05,
+      "loss": 1.0711,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733364882169615e-05,
+      "loss": 1.006,
+      "step": 1573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329996883374538e-05,
+      "loss": 1.0621,
+      "step": 1574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732634283111713e-05,
+      "loss": 1.1274,
+      "step": 1575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322686665977738e-05,
+      "loss": 1.0565,
+      "step": 1576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731902838901078e-05,
+      "loss": 1.0016,
+      "step": 1577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315368001271287e-05,
+      "loss": 1.1316,
+      "step": 1578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73117055038149e-05,
+      "loss": 1.0371,
+      "step": 1579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308040897697864e-05,
+      "loss": 1.0976,
+      "step": 1580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304374183977032e-05,
+      "loss": 1.0345,
+      "step": 1581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300705363709868e-05,
+      "loss": 1.1066,
+      "step": 1582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297034437954442e-05,
+      "loss": 1.0423,
+      "step": 1583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729336140776943e-05,
+      "loss": 1.0495,
+      "step": 1584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 1.0814,
+      "step": 1585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286009038348395e-05,
+      "loss": 1.0234,
+      "step": 1586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7282329701232758e-05,
+      "loss": 1.0577,
+      "step": 1587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278648263928313e-05,
+      "loss": 1.0984,
+      "step": 1588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727496472749677e-05,
+      "loss": 1.1047,
+      "step": 1589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271279093000434e-05,
+      "loss": 1.0602,
+      "step": 1590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267591361502233e-05,
+      "loss": 1.0271,
+      "step": 1591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726390153406569e-05,
+      "loss": 1.1293,
+      "step": 1592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726020961175493e-05,
+      "loss": 1.073,
+      "step": 1593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256515595634688e-05,
+      "loss": 1.018,
+      "step": 1594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72528194867703e-05,
+      "loss": 1.1266,
+      "step": 1595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249121286227705e-05,
+      "loss": 1.0683,
+      "step": 1596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245420995073453e-05,
+      "loss": 1.0389,
+      "step": 1597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7241718614374678e-05,
+      "loss": 1.0683,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238014145199135e-05,
+      "loss": 1.1256,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234307588615177e-05,
+      "loss": 1.048,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230598945691755e-05,
+      "loss": 1.0431,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226888217498426e-05,
+      "loss": 1.1263,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722317540510534e-05,
+      "loss": 1.0451,
+      "step": 1603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721946050958326e-05,
+      "loss": 1.0372,
+      "step": 1604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215743532003545e-05,
+      "loss": 1.0623,
+      "step": 1605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212024473438145e-05,
+      "loss": 1.0936,
+      "step": 1606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208303334959627e-05,
+      "loss": 1.1054,
+      "step": 1607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720458011764115e-05,
+      "loss": 1.091,
+      "step": 1608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200854822556467e-05,
+      "loss": 1.0759,
+      "step": 1609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197127450779936e-05,
+      "loss": 1.0886,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 1.0022,
+      "step": 1611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189666481451755e-05,
+      "loss": 0.982,
+      "step": 1612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185932886051813e-05,
+      "loss": 1.1277,
+      "step": 1613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182197218263438e-05,
+      "loss": 1.1038,
+      "step": 1614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717845947916398e-05,
+      "loss": 1.0267,
+      "step": 1615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174719669831383e-05,
+      "loss": 1.1429,
+      "step": 1616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717097779134419e-05,
+      "loss": 1.0365,
+      "step": 1617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167233844781545e-05,
+      "loss": 1.0708,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716348783122318e-05,
+      "loss": 1.0709,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159739751749423e-05,
+      "loss": 1.0683,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715598960744121e-05,
+      "loss": 1.076,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152237399380066e-05,
+      "loss": 1.0139,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71484831286481e-05,
+      "loss": 1.1392,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144726796328034e-05,
+      "loss": 1.0747,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140968403503175e-05,
+      "loss": 1.0402,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713720795125742e-05,
+      "loss": 1.1437,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133445440675268e-05,
+      "loss": 1.1256,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129680872841812e-05,
+      "loss": 1.0424,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125914248842735e-05,
+      "loss": 1.072,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712214556976431e-05,
+      "loss": 1.1052,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118374836693407e-05,
+      "loss": 0.9764,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711460205071749e-05,
+      "loss": 1.0458,
+      "step": 1632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110827212924604e-05,
+      "loss": 1.0106,
+      "step": 1633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71070503244034e-05,
+      "loss": 1.111,
+      "step": 1634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710327138624312e-05,
+      "loss": 1.0523,
+      "step": 1635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099490399533583e-05,
+      "loss": 1.0082,
+      "step": 1636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709570736536521e-05,
+      "loss": 1.1125,
+      "step": 1637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709192228482901e-05,
+      "loss": 1.0472,
+      "step": 1638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088135159016584e-05,
+      "loss": 1.1236,
+      "step": 1639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084345989020116e-05,
+      "loss": 1.0923,
+      "step": 1640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080554775932386e-05,
+      "loss": 1.0812,
+      "step": 1641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707676152084677e-05,
+      "loss": 1.0498,
+      "step": 1642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707296622485721e-05,
+      "loss": 1.0818,
+      "step": 1643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706916888905826e-05,
+      "loss": 1.0075,
+      "step": 1644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065369514545054e-05,
+      "loss": 1.0553,
+      "step": 1645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706156810241331e-05,
+      "loss": 1.0852,
+      "step": 1646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057764653759337e-05,
+      "loss": 1.0744,
+      "step": 1647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053959169680033e-05,
+      "loss": 1.0812,
+      "step": 1648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705015165127288e-05,
+      "loss": 0.9978,
+      "step": 1649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046342099635948e-05,
+      "loss": 1.0804,
+      "step": 1650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042530515867897e-05,
+      "loss": 1.1068,
+      "step": 1651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703871690106796e-05,
+      "loss": 1.0191,
+      "step": 1652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703490125633598e-05,
+      "loss": 1.0764,
+      "step": 1653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031083582772353e-05,
+      "loss": 1.1039,
+      "step": 1654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027263881478092e-05,
+      "loss": 1.0358,
+      "step": 1655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023442153554776e-05,
+      "loss": 1.0892,
+      "step": 1656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019618400104572e-05,
+      "loss": 1.0818,
+      "step": 1657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015792622230226e-05,
+      "loss": 1.0578,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011964821035088e-05,
+      "loss": 1.0676,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7008134997623066e-05,
+      "loss": 1.0189,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004303153098668e-05,
+      "loss": 1.1007,
+      "step": 1661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000469288566974e-05,
+      "loss": 1.0186,
+      "step": 1662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 1.0454,
+      "step": 1663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992795503904963e-05,
+      "loss": 1.1004,
+      "step": 1664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988955585987733e-05,
+      "loss": 1.0848,
+      "step": 1665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985113652489374e-05,
+      "loss": 1.0402,
+      "step": 1666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981269704517877e-05,
+      "loss": 1.0465,
+      "step": 1667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977423743181826e-05,
+      "loss": 1.1219,
+      "step": 1668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697357576959038e-05,
+      "loss": 1.0927,
+      "step": 1669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969725784853264e-05,
+      "loss": 1.0263,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965873790080806e-05,
+      "loss": 1.1109,
+      "step": 1671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962019786383894e-05,
+      "loss": 0.9903,
+      "step": 1672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958163774874012e-05,
+      "loss": 1.0579,
+      "step": 1673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954305756663214e-05,
+      "loss": 1.0576,
+      "step": 1674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695044573286413e-05,
+      "loss": 1.1364,
+      "step": 1675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 1.0126,
+      "step": 1676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694271967295454e-05,
+      "loss": 1.0809,
+      "step": 1677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938853639072186e-05,
+      "loss": 1.0609,
+      "step": 1678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693498560405787e-05,
+      "loss": 1.0736,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931115569027104e-05,
+      "loss": 1.0685,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 1.0425,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923369503381218e-05,
+      "loss": 1.1076,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691949347500002e-05,
+      "loss": 1.001,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915615451070234e-05,
+      "loss": 1.1017,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911735432710263e-05,
+      "loss": 1.0536,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690785342103908e-05,
+      "loss": 1.1764,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903969417176244e-05,
+      "loss": 1.0683,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900083422241883e-05,
+      "loss": 0.9623,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68961954373567e-05,
+      "loss": 1.1063,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892305463641967e-05,
+      "loss": 1.0794,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888413502219534e-05,
+      "loss": 0.9873,
+      "step": 1691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688451955421183e-05,
+      "loss": 1.0046,
+      "step": 1692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880623620741843e-05,
+      "loss": 1.1607,
+      "step": 1693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876725702933147e-05,
+      "loss": 1.0231,
+      "step": 1694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687282580190988e-05,
+      "loss": 1.0468,
+      "step": 1695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868923918796753e-05,
+      "loss": 1.1115,
+      "step": 1696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865020054719057e-05,
+      "loss": 0.9595,
+      "step": 1697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686111421080264e-05,
+      "loss": 1.0603,
+      "step": 1698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685720638817393e-05,
+      "loss": 1.1409,
+      "step": 1699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853296587959934e-05,
+      "loss": 1.0483,
+      "step": 1700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849384811288202e-05,
+      "loss": 1.07,
+      "step": 1701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684547105928689e-05,
+      "loss": 1.071,
+      "step": 1702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684155533308469e-05,
+      "loss": 1.0963,
+      "step": 1703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683763763381089e-05,
+      "loss": 0.9972,
+      "step": 1704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833717962595327e-05,
+      "loss": 1.1059,
+      "step": 1705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829796320568416e-05,
+      "loss": 0.9999,
+      "step": 1706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682587270886115e-05,
+      "loss": 1.1172,
+      "step": 1707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821947128605068e-05,
+      "loss": 1.0055,
+      "step": 1708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818019580932294e-05,
+      "loss": 1.1042,
+      "step": 1709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681409006697551e-05,
+      "loss": 1.1151,
+      "step": 1710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810158587867973e-05,
+      "loss": 1.0646,
+      "step": 1711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68062251447435e-05,
+      "loss": 1.0687,
+      "step": 1712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680228973873648e-05,
+      "loss": 1.1014,
+      "step": 1713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798352370981864e-05,
+      "loss": 1.0394,
+      "step": 1714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794413042615168e-05,
+      "loss": 1.0103,
+      "step": 1715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790471754772474e-05,
+      "loss": 1.0386,
+      "step": 1716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786528508590436e-05,
+      "loss": 1.1437,
+      "step": 1717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782583305206263e-05,
+      "loss": 1.0617,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778636145757735e-05,
+      "loss": 1.0252,
+      "step": 1719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677468703138319e-05,
+      "loss": 1.1041,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770735963221542e-05,
+      "loss": 1.0444,
+      "step": 1721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766782942412246e-05,
+      "loss": 0.9673,
+      "step": 1722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762827970095344e-05,
+      "loss": 1.0804,
+      "step": 1723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758871047411433e-05,
+      "loss": 1.0355,
+      "step": 1724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754912175501666e-05,
+      "loss": 1.0224,
+      "step": 1725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750951355507763e-05,
+      "loss": 1.0932,
+      "step": 1726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746988588572005e-05,
+      "loss": 1.1103,
+      "step": 1727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743023875837233e-05,
+      "loss": 1.0356,
+      "step": 1728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673905721844686e-05,
+      "loss": 1.0724,
+      "step": 1729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673508861754484e-05,
+      "loss": 1.0151,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67311180742757e-05,
+      "loss": 1.0514,
+      "step": 1731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727145589784536e-05,
+      "loss": 1.0472,
+      "step": 1732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672317116521698e-05,
+      "loss": 1.0937,
+      "step": 1733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719194801719244e-05,
+      "loss": 1.1247,
+      "step": 1734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715216500438093e-05,
+      "loss": 1.0638,
+      "step": 1735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671123626252085e-05,
+      "loss": 1.0881,
+      "step": 1736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670725408911539e-05,
+      "loss": 1.0198,
+      "step": 1737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670326998137016e-05,
+      "loss": 1.1092,
+      "step": 1738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669928394043415e-05,
+      "loss": 1.0587,
+      "step": 1739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695295967456918e-05,
+      "loss": 1.0418,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 1.0373,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687314229979806e-05,
+      "loss": 1.1175,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683320467781817e-05,
+      "loss": 1.0202,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679324778146394e-05,
+      "loss": 1.0528,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675327162225878e-05,
+      "loss": 1.1303,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667132762117316e-05,
+      "loss": 1.0191,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666732615614169e-05,
+      "loss": 1.0451,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666332276828547e-05,
+      "loss": 1.1009,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659317458759056e-05,
+      "loss": 0.9974,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655310228717565e-05,
+      "loss": 1.085,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665130107931666e-05,
+      "loss": 1.1029,
+      "step": 1751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664729001171256e-05,
+      "loss": 1.0967,
+      "step": 1752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643277027062036e-05,
+      "loss": 1.0372,
+      "step": 1753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639262126522417e-05,
+      "loss": 1.0671,
+      "step": 1754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663524531125158e-05,
+      "loss": 1.0203,
+      "step": 1755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 1.0854,
+      "step": 1756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662720594115052e-05,
+      "loss": 1.0368,
+      "step": 1757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623183388638815e-05,
+      "loss": 1.0661,
+      "step": 1758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619158926032922e-05,
+      "loss": 1.028,
+      "step": 1759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615132554493477e-05,
+      "loss": 1.0985,
+      "step": 1760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611104275181664e-05,
+      "loss": 1.0438,
+      "step": 1761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607074089259222e-05,
+      "loss": 1.1665,
+      "step": 1762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603041997888437e-05,
+      "loss": 1.0461,
+      "step": 1763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599008002232145e-05,
+      "loss": 1.066,
+      "step": 1764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594972103453727e-05,
+      "loss": 1.0366,
+      "step": 1765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659093430271712e-05,
+      "loss": 1.1324,
+      "step": 1766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586894601186804e-05,
+      "loss": 1.0222,
+      "step": 1767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582853000027818e-05,
+      "loss": 1.0191,
+      "step": 1768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657880950040573e-05,
+      "loss": 1.0953,
+      "step": 1769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574764103486668e-05,
+      "loss": 1.0438,
+      "step": 1770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657071681043731e-05,
+      "loss": 1.0623,
+      "step": 1771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656666762242487e-05,
+      "loss": 1.1497,
+      "step": 1772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656261654061712e-05,
+      "loss": 1.0799,
+      "step": 1773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6558563566182365e-05,
+      "loss": 0.9814,
+      "step": 1774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655450870028947e-05,
+      "loss": 1.0497,
+      "step": 1775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550451944107835e-05,
+      "loss": 1.1488,
+      "step": 1776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546393298807405e-05,
+      "loss": 1.0687,
+      "step": 1777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542332765558688e-05,
+      "loss": 1.0579,
+      "step": 1778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.653827034553271e-05,
+      "loss": 1.0854,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534206039901057e-05,
+      "loss": 1.0561,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653013984983585e-05,
+      "loss": 1.1042,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526071776509768e-05,
+      "loss": 1.0465,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652200182109602e-05,
+      "loss": 1.1375,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651792998476836e-05,
+      "loss": 1.0456,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513856268701088e-05,
+      "loss": 1.0281,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650978067406904e-05,
+      "loss": 1.0213,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65057032020476e-05,
+      "loss": 1.0576,
+      "step": 1787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501623853812695e-05,
+      "loss": 1.0023,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497542630540783e-05,
+      "loss": 1.1031,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493459533408877e-05,
+      "loss": 1.098,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648937456359451e-05,
+      "loss": 1.0101,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485287722275783e-05,
+      "loss": 1.1021,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481199010631312e-05,
+      "loss": 1.032,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477108429840263e-05,
+      "loss": 1.1417,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647301598108234e-05,
+      "loss": 1.0206,
+      "step": 1795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468921665537786e-05,
+      "loss": 1.0805,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464825484387382e-05,
+      "loss": 1.1167,
+      "step": 1797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460727438812446e-05,
+      "loss": 0.9645,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645662752999484e-05,
+      "loss": 1.0999,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452525759116945e-05,
+      "loss": 1.1229,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448422127361707e-05,
+      "loss": 1.021,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444316635912585e-05,
+      "loss": 0.9487,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440209285953587e-05,
+      "loss": 1.1204,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643610007866925e-05,
+      "loss": 1.0458,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431989015244646e-05,
+      "loss": 1.0693,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 1.0931,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423761324717636e-05,
+      "loss": 1.0556,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419644699988053e-05,
+      "loss": 1.0873,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641552622386386e-05,
+      "loss": 1.0801,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64114058975328e-05,
+      "loss": 1.0254,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640728372218317e-05,
+      "loss": 1.1078,
+      "step": 1811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640315969900377e-05,
+      "loss": 1.049,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639903382918395e-05,
+      "loss": 1.027,
+      "step": 1813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394906113913603e-05,
+      "loss": 1.086,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639077655438313e-05,
+      "loss": 1.0936,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638664515178348e-05,
+      "loss": 1.0508,
+      "step": 1816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382511907306132e-05,
+      "loss": 1.1098,
+      "step": 1817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378376822143094e-05,
+      "loss": 0.9493,
+      "step": 1818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 1.115,
+      "step": 1819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637010113453062e-05,
+      "loss": 1.0418,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636596053446786e-05,
+      "loss": 1.0703,
+      "step": 1821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636181809849274e-05,
+      "loss": 1.0423,
+      "step": 1822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357673827799923e-05,
+      "loss": 1.08,
+      "step": 1823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353527723584596e-05,
+      "loss": 1.0337,
+      "step": 1824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349379787042478e-05,
+      "loss": 1.1323,
+      "step": 1825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634523001936981e-05,
+      "loss": 1.034,
+      "step": 1826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634107842176337e-05,
+      "loss": 1.1152,
+      "step": 1827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336924995420453e-05,
+      "loss": 1.1314,
+      "step": 1828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6332769741538892e-05,
+      "loss": 1.0154,
+      "step": 1829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328612661317038e-05,
+      "loss": 1.0339,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 1.1327,
+      "step": 1831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632029302664851e-05,
+      "loss": 1.0248,
+      "step": 1832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631613047460118e-05,
+      "loss": 1.0646,
+      "step": 1833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311966101012246e-05,
+      "loss": 1.0442,
+      "step": 1834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307799907082685e-05,
+      "loss": 1.1006,
+      "step": 1835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303631894014015e-05,
+      "loss": 1.0484,
+      "step": 1836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299462063008272e-05,
+      "loss": 1.0359,
+      "step": 1837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295290415268008e-05,
+      "loss": 1.0951,
+      "step": 1838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291116951996312e-05,
+      "loss": 1.0837,
+      "step": 1839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286941674396788e-05,
+      "loss": 0.9893,
+      "step": 1840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282764583673566e-05,
+      "loss": 1.0347,
+      "step": 1841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278585681031304e-05,
+      "loss": 1.1001,
+      "step": 1842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274404967675168e-05,
+      "loss": 1.0774,
+      "step": 1843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270222444810864e-05,
+      "loss": 1.0638,
+      "step": 1844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266038113644605e-05,
+      "loss": 1.0986,
+      "step": 1845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626185197538314e-05,
+      "loss": 1.0066,
+      "step": 1846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257664031233723e-05,
+      "loss": 1.0329,
+      "step": 1847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253474282404137e-05,
+      "loss": 1.0952,
+      "step": 1848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249282730102693e-05,
+      "loss": 1.0495,
+      "step": 1849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245089375538208e-05,
+      "loss": 0.9369,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624089421992003e-05,
+      "loss": 1.0886,
+      "step": 1851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236697264458013e-05,
+      "loss": 1.1275,
+      "step": 1852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623249851036255e-05,
+      "loss": 1.0678,
+      "step": 1853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228297958844536e-05,
+      "loss": 1.063,
+      "step": 1854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224095611115385e-05,
+      "loss": 1.0815,
+      "step": 1855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621989146838704e-05,
+      "loss": 1.1097,
+      "step": 1856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215685531871957e-05,
+      "loss": 1.0902,
+      "step": 1857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211477802783105e-05,
+      "loss": 1.0304,
+      "step": 1858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207268282333973e-05,
+      "loss": 1.1334,
+      "step": 1859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203056971738566e-05,
+      "loss": 1.0303,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198843872211404e-05,
+      "loss": 1.1003,
+      "step": 1861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194628984967525e-05,
+      "loss": 1.0574,
+      "step": 1862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190412311222486e-05,
+      "loss": 1.0812,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186193852192356e-05,
+      "loss": 0.9778,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6181973609093716e-05,
+      "loss": 1.023,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617775158314366e-05,
+      "loss": 1.0949,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61735277755598e-05,
+      "loss": 1.0756,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616930218756027e-05,
+      "loss": 1.0409,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165074820363703e-05,
+      "loss": 1.0372,
+      "step": 1869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160845675189254e-05,
+      "loss": 1.1244,
+      "step": 1870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 1.0495,
+      "step": 1871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152382055785873e-05,
+      "loss": 1.0736,
+      "step": 1872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148147583997813e-05,
+      "loss": 1.0669,
+      "step": 1873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61439113391136e-05,
+      "loss": 0.9974,
+      "step": 1874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139673322354953e-05,
+      "loss": 1.1014,
+      "step": 1875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613543353494409e-05,
+      "loss": 1.1212,
+      "step": 1876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131191978103748e-05,
+      "loss": 1.0203,
+      "step": 1877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126948653057172e-05,
+      "loss": 1.0052,
+      "step": 1878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122703561028114e-05,
+      "loss": 1.0185,
+      "step": 1879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118456703240842e-05,
+      "loss": 1.0773,
+      "step": 1880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114208080920125e-05,
+      "loss": 1.0644,
+      "step": 1881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109957695291246e-05,
+      "loss": 1.0379,
+      "step": 1882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105705547579998e-05,
+      "loss": 1.0488,
+      "step": 1883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610145163901268e-05,
+      "loss": 1.0814,
+      "step": 1884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097195970816094e-05,
+      "loss": 0.9961,
+      "step": 1885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609293854421756e-05,
+      "loss": 1.0531,
+      "step": 1886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088679360444896e-05,
+      "loss": 1.1612,
+      "step": 1887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084418420726432e-05,
+      "loss": 1.0399,
+      "step": 1888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080155726291005e-05,
+      "loss": 1.0021,
+      "step": 1889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607589127836795e-05,
+      "loss": 1.0984,
+      "step": 1890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071625078187113e-05,
+      "loss": 1.0819,
+      "step": 1891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067357126978852e-05,
+      "loss": 1.031,
+      "step": 1892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063087425974014e-05,
+      "loss": 1.1018,
+      "step": 1893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058815976403973e-05,
+      "loss": 1.0551,
+      "step": 1894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605454277950058e-05,
+      "loss": 0.9909,
+      "step": 1895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050267836496217e-05,
+      "loss": 1.1187,
+      "step": 1896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 1.0007,
+      "step": 1897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041712717116554e-05,
+      "loss": 1.0518,
+      "step": 1898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037432543208518e-05,
+      "loss": 1.0802,
+      "step": 1899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603315062813401e-05,
+      "loss": 1.0152,
+      "step": 1900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028866973127922e-05,
+      "loss": 1.1049,
+      "step": 1901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602458157942564e-05,
+      "loss": 1.0468,
+      "step": 1902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020294448263046e-05,
+      "loss": 1.0726,
+      "step": 1903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601600558087653e-05,
+      "loss": 1.0562,
+      "step": 1904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601171497850298e-05,
+      "loss": 1.0219,
+      "step": 1905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600742264237979e-05,
+      "loss": 1.0666,
+      "step": 1906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003128573744847e-05,
+      "loss": 1.0866,
+      "step": 1907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599883277383653e-05,
+      "loss": 1.0576,
+      "step": 1908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994535243893742e-05,
+      "loss": 1.1018,
+      "step": 1909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599023598515586e-05,
+      "loss": 1.0832,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985934998862775e-05,
+      "loss": 1.052,
+      "step": 1911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981632286254863e-05,
+      "loss": 1.0871,
+      "step": 1912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977327848573017e-05,
+      "loss": 0.9976,
+      "step": 1913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973021687058607e-05,
+      "loss": 1.068,
+      "step": 1914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596871380295351e-05,
+      "loss": 1.0934,
+      "step": 1915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5964404197500104e-05,
+      "loss": 1.0393,
+      "step": 1916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960092871941253e-05,
+      "loss": 0.9766,
+      "step": 1917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955779827520327e-05,
+      "loss": 1.1152,
+      "step": 1918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951465065481185e-05,
+      "loss": 1.0624,
+      "step": 1919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947148587068185e-05,
+      "loss": 1.0934,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942830393526176e-05,
+      "loss": 1.1328,
+      "step": 1921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938510486100506e-05,
+      "loss": 1.0254,
+      "step": 1922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934188866037017e-05,
+      "loss": 1.0652,
+      "step": 1923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929865534582038e-05,
+      "loss": 1.0771,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925540492982406e-05,
+      "loss": 1.0611,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592121374248543e-05,
+      "loss": 1.0268,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916885284338937e-05,
+      "loss": 1.0594,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912555119791222e-05,
+      "loss": 1.0727,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908223250091093e-05,
+      "loss": 1.1033,
+      "step": 1929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903889676487832e-05,
+      "loss": 1.0631,
+      "step": 1930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899554400231233e-05,
+      "loss": 1.0568,
+      "step": 1931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895217422571554e-05,
+      "loss": 1.0901,
+      "step": 1932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890878744759564e-05,
+      "loss": 1.0274,
+      "step": 1933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5886538368046524e-05,
+      "loss": 1.0787,
+      "step": 1934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882196293684166e-05,
+      "loss": 1.0625,
+      "step": 1935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 1.0663,
+      "step": 1936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873507057020942e-05,
+      "loss": 1.0646,
+      "step": 1937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869159897226007e-05,
+      "loss": 1.0406,
+      "step": 1938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864811044793627e-05,
+      "loss": 1.0046,
+      "step": 1939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586046050097799e-05,
+      "loss": 1.0415,
+      "step": 1940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5856108267033768e-05,
+      "loss": 1.0633,
+      "step": 1941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851754344216135e-05,
+      "loss": 1.0882,
+      "step": 1942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584739873378073e-05,
+      "loss": 1.1235,
+      "step": 1943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843041436983697e-05,
+      "loss": 1.0342,
+      "step": 1944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838682455081657e-05,
+      "loss": 1.0387,
+      "step": 1945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834321789331717e-05,
+      "loss": 1.1147,
+      "step": 1946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582995944099148e-05,
+      "loss": 1.0583,
+      "step": 1947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825595411319016e-05,
+      "loss": 1.0718,
+      "step": 1948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821229701572897e-05,
+      "loss": 1.1101,
+      "step": 1949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581686231301217e-05,
+      "loss": 0.9963,
+      "step": 1950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812493246896368e-05,
+      "loss": 1.0332,
+      "step": 1951
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808122504485514e-05,
+      "loss": 1.116,
+      "step": 1952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58037500870401e-05,
+      "loss": 0.9935,
+      "step": 1953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799375995821116e-05,
+      "loss": 1.0301,
+      "step": 1954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795000232090033e-05,
+      "loss": 1.1004,
+      "step": 1955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579062279710879e-05,
+      "loss": 1.0938,
+      "step": 1956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786243692139826e-05,
+      "loss": 1.053,
+      "step": 1957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781862918446053e-05,
+      "loss": 1.0115,
+      "step": 1958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577748047729086e-05,
+      "loss": 1.0736,
+      "step": 1959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5773096369938125e-05,
+      "loss": 1.1347,
+      "step": 1960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5768710597652205e-05,
+      "loss": 1.0613,
+      "step": 1961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764323161697933e-05,
+      "loss": 1.0438,
+      "step": 1962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759934063340627e-05,
+      "loss": 1.1032,
+      "step": 1963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755543303846083e-05,
+      "loss": 1.0379,
+      "step": 1964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751150884480568e-05,
+      "loss": 1.0391,
+      "step": 1965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574675680651084e-05,
+      "loss": 1.0634,
+      "step": 1966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742361071204127e-05,
+      "loss": 1.1185,
+      "step": 1967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573796367982814e-05,
+      "loss": 1.1163,
+      "step": 1968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573356463365107e-05,
+      "loss": 1.0381,
+      "step": 1969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729163933941574e-05,
+      "loss": 1.1018,
+      "step": 1970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572476158196879e-05,
+      "loss": 0.9984,
+      "step": 1971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720357579002346e-05,
+      "loss": 1.0227,
+      "step": 1972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715951926312327e-05,
+      "loss": 1.0994,
+      "step": 1973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711544625169303e-05,
+      "loss": 1.0235,
+      "step": 1974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 1.0307,
+      "step": 1975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702725082608894e-05,
+      "loss": 1.0841,
+      "step": 1976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698312843735022e-05,
+      "loss": 1.1036,
+      "step": 1977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693898961495176e-05,
+      "loss": 1.0262,
+      "step": 1978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689483437162293e-05,
+      "loss": 1.0596,
+      "step": 1979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568506627200979e-05,
+      "loss": 1.0924,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 1.0053,
+      "step": 1981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567622702434196e-05,
+      "loss": 1.0695,
+      "step": 1982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671804944375826e-05,
+      "loss": 1.0105,
+      "step": 1983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667381228688464e-05,
+      "loss": 1.076,
+      "step": 1984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662955878555654e-05,
+      "loss": 1.0698,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565852889525364e-05,
+      "loss": 1.0931,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654100280059155e-05,
+      "loss": 1.0663,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564967003424938e-05,
+      "loss": 1.0276,
+      "step": 1988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564523815910198e-05,
+      "loss": 1.0502,
+      "step": 1989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640804655895086e-05,
+      "loss": 1.0292,
+      "step": 1990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636369525907297e-05,
+      "loss": 1.1165,
+      "step": 1991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5631932770417685e-05,
+      "loss": 1.0175,
+      "step": 1992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562749439070579e-05,
+      "loss": 1.0052,
+      "step": 1993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623054388051618e-05,
+      "loss": 1.0593,
+      "step": 1994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618612763735642e-05,
+      "loss": 1.0493,
+      "step": 1995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561416951903881e-05,
+      "loss": 1.0105,
+      "step": 1996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560972465524253e-05,
+      "loss": 1.1477,
+      "step": 1997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560527817362868e-05,
+      "loss": 1.0517,
+      "step": 1998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600830075479604e-05,
+      "loss": 1.016,
+      "step": 1999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596380362078108e-05,
+      "loss": 1.0498,
+      "step": 2000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 1.0936,
+      "step": 2001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587476094651436e-05,
+      "loss": 1.0943,
+      "step": 2002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583021543194206e-05,
+      "loss": 1.0333,
+      "step": 2003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578565381620453e-05,
+      "loss": 1.0579,
+      "step": 2004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557410761121532e-05,
+      "loss": 1.0973,
+      "step": 2005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569648233264395e-05,
+      "loss": 1.0795,
+      "step": 2006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565187249053747e-05,
+      "loss": 0.991,
+      "step": 2007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560724659869905e-05,
+      "loss": 1.025,
+      "step": 2008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555626046699985e-05,
+      "loss": 1.0404,
+      "step": 2009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555179467173104e-05,
+      "loss": 1.0334,
+      "step": 2010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554732727535139e-05,
+      "loss": 1.0846,
+      "step": 2011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554285827914927e-05,
+      "loss": 1.0265,
+      "step": 2012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538387684413522e-05,
+      "loss": 1.0866,
+      "step": 2013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553391549243344e-05,
+      "loss": 1.0565,
+      "step": 2014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552944170449879e-05,
+      "loss": 1.1474,
+      "step": 2015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552496632189978e-05,
+      "loss": 1.0302,
+      "step": 2016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5520489345927095e-05,
+      "loss": 0.9841,
+      "step": 2017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516010777871876e-05,
+      "loss": 1.0852,
+      "step": 2018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511530619025716e-05,
+      "loss": 1.0922,
+      "step": 2019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507048870680668e-05,
+      "loss": 1.0531,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502565534129253e-05,
+      "loss": 0.9152,
+      "step": 2021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498080610664444e-05,
+      "loss": 1.1001,
+      "step": 2022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493594101579668e-05,
+      "loss": 1.0777,
+      "step": 2023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548910600816881e-05,
+      "loss": 1.0154,
+      "step": 2024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5484616331726225e-05,
+      "loss": 1.115,
+      "step": 2025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480125073546705e-05,
+      "loss": 1.0249,
+      "step": 2026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475632234925505e-05,
+      "loss": 1.0209,
+      "step": 2027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5471137817158342e-05,
+      "loss": 1.061,
+      "step": 2028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546664182154139e-05,
+      "loss": 1.1092,
+      "step": 2029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462144249371266e-05,
+      "loss": 1.0683,
+      "step": 2030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457645101945046e-05,
+      "loss": 1.0391,
+      "step": 2031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453144380560268e-05,
+      "loss": 1.0446,
+      "step": 2032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448642086514917e-05,
+      "loss": 1.1483,
+      "step": 2033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544413822110743e-05,
+      "loss": 1.0709,
+      "step": 2034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439632785636707e-05,
+      "loss": 0.9999,
+      "step": 2035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435125781402086e-05,
+      "loss": 1.0991,
+      "step": 2036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543061720970337e-05,
+      "loss": 0.9764,
+      "step": 2037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426107071840805e-05,
+      "loss": 1.0347,
+      "step": 2038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54215953691151e-05,
+      "loss": 1.0512,
+      "step": 2039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54170821028274e-05,
+      "loss": 1.0808,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412567274279316e-05,
+      "loss": 1.052,
+      "step": 2041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54080508847729e-05,
+      "loss": 1.0092,
+      "step": 2042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403532935610655e-05,
+      "loss": 1.0516,
+      "step": 2043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539901342809554e-05,
+      "loss": 1.0845,
+      "step": 2044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394492363530957e-05,
+      "loss": 1.0276,
+      "step": 2045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538996974322076e-05,
+      "loss": 1.0408,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538544556846925e-05,
+      "loss": 1.0834,
+      "step": 2047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380919840581178e-05,
+      "loss": 1.0874,
+      "step": 2048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537639256086174e-05,
+      "loss": 0.9673,
+      "step": 2049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371863730616586e-05,
+      "loss": 1.0962,
+      "step": 2050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367333351151804e-05,
+      "loss": 1.0451,
+      "step": 2051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362801423773935e-05,
+      "loss": 1.0855,
+      "step": 2052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 1.1012,
+      "step": 2053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353732930507332e-05,
+      "loss": 1.0451,
+      "step": 2054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349196367233903e-05,
+      "loss": 0.9736,
+      "step": 2055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344658261278013e-05,
+      "loss": 1.1243,
+      "step": 2056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340118613948418e-05,
+      "loss": 1.0519,
+      "step": 2057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533557742655434e-05,
+      "loss": 1.0405,
+      "step": 2058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533103470040543e-05,
+      "loss": 1.0747,
+      "step": 2059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532649043681179e-05,
+      "loss": 0.9572,
+      "step": 2060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321944637083963e-05,
+      "loss": 1.1406,
+      "step": 2061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317397302532933e-05,
+      "loss": 1.0977,
+      "step": 2062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531284843447014e-05,
+      "loss": 1.0443,
+      "step": 2063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308298034207444e-05,
+      "loss": 1.1343,
+      "step": 2064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303746103057163e-05,
+      "loss": 0.9891,
+      "step": 2065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529919264233205e-05,
+      "loss": 1.0525,
+      "step": 2066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294637653345307e-05,
+      "loss": 1.0711,
+      "step": 2067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290081137410563e-05,
+      "loss": 1.0763,
+      "step": 2068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285523095841904e-05,
+      "loss": 1.012,
+      "step": 2069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528096352995384e-05,
+      "loss": 1.1154,
+      "step": 2070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527640244106133e-05,
+      "loss": 1.0106,
+      "step": 2071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527183983047977e-05,
+      "loss": 1.0222,
+      "step": 2072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267275699524992e-05,
+      "loss": 1.0798,
+      "step": 2073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526271004951328e-05,
+      "loss": 1.067,
+      "step": 2074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258142881761332e-05,
+      "loss": 1.048,
+      "step": 2075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253574197586308e-05,
+      "loss": 1.0619,
+      "step": 2076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249003998305787e-05,
+      "loss": 1.0786,
+      "step": 2077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244432285237796e-05,
+      "loss": 1.1156,
+      "step": 2078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239859059700794e-05,
+      "loss": 1.0081,
+      "step": 2079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235284323013674e-05,
+      "loss": 1.0161,
+      "step": 2080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230708076495777e-05,
+      "loss": 1.0881,
+      "step": 2081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226130321466866e-05,
+      "loss": 1.0789,
+      "step": 2082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221551059247137e-05,
+      "loss": 1.0271,
+      "step": 2083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216970291157233e-05,
+      "loss": 1.0637,
+      "step": 2084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212388018518226e-05,
+      "loss": 0.9689,
+      "step": 2085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207804242651625e-05,
+      "loss": 1.0707,
+      "step": 2086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520321896487936e-05,
+      "loss": 0.9558,
+      "step": 2087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198632186523807e-05,
+      "loss": 1.0974,
+      "step": 2088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194043908907774e-05,
+      "loss": 1.0317,
+      "step": 2089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189454133354494e-05,
+      "loss": 0.9905,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184862861187637e-05,
+      "loss": 1.0733,
+      "step": 2091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180270093731305e-05,
+      "loss": 1.0954,
+      "step": 2092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175675832310028e-05,
+      "loss": 1.0167,
+      "step": 2093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517108007824877e-05,
+      "loss": 1.1012,
+      "step": 2094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166482832872923e-05,
+      "loss": 1.1215,
+      "step": 2095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161884097508316e-05,
+      "loss": 1.0208,
+      "step": 2096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157283873481197e-05,
+      "loss": 1.0512,
+      "step": 2097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515268216211825e-05,
+      "loss": 1.1063,
+      "step": 2098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148078964746583e-05,
+      "loss": 0.9806,
+      "step": 2099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514347428269374e-05,
+      "loss": 0.9917,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138868117287689e-05,
+      "loss": 1.0639,
+      "step": 2101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134260469856825e-05,
+      "loss": 1.0853,
+      "step": 2102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512965134172997e-05,
+      "loss": 1.0575,
+      "step": 2103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125040734236374e-05,
+      "loss": 1.0458,
+      "step": 2104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120428648705716e-05,
+      "loss": 1.072,
+      "step": 2105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115815086468103e-05,
+      "loss": 1.0927,
+      "step": 2106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111200048854055e-05,
+      "loss": 1.072,
+      "step": 2107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106583537194533e-05,
+      "loss": 1.0038,
+      "step": 2108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101965552820916e-05,
+      "loss": 1.0549,
+      "step": 2109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097346097065008e-05,
+      "loss": 1.0859,
+      "step": 2110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092725171259037e-05,
+      "loss": 1.0513,
+      "step": 2111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088102776735656e-05,
+      "loss": 1.0523,
+      "step": 2112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508347891482794e-05,
+      "loss": 1.0249,
+      "step": 2113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078853586869393e-05,
+      "loss": 1.0603,
+      "step": 2114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074226794193934e-05,
+      "loss": 1.1042,
+      "step": 2115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069598538135905e-05,
+      "loss": 0.9958,
+      "step": 2116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064968820030079e-05,
+      "loss": 1.0518,
+      "step": 2117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060337641211637e-05,
+      "loss": 0.9492,
+      "step": 2118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055705003016199e-05,
+      "loss": 1.1329,
+      "step": 2119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051070906779789e-05,
+      "loss": 1.0705,
+      "step": 2120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046435353838855e-05,
+      "loss": 1.0191,
+      "step": 2121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041798345530272e-05,
+      "loss": 1.1098,
+      "step": 2122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503715988319133e-05,
+      "loss": 1.032,
+      "step": 2123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503251996815974e-05,
+      "loss": 1.0254,
+      "step": 2124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027878601773633e-05,
+      "loss": 1.036,
+      "step": 2125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023235785371552e-05,
+      "loss": 1.0971,
+      "step": 2126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018591520292467e-05,
+      "loss": 1.0115,
+      "step": 2127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013945807875758e-05,
+      "loss": 1.0765,
+      "step": 2128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500929864946123e-05,
+      "loss": 1.0891,
+      "step": 2129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50046500463891e-05,
+      "loss": 1.0676,
+      "step": 2130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.0507,
+      "step": 2131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995348511634986e-05,
+      "loss": 0.9446,
+      "step": 2132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499069558263552e-05,
+      "loss": 1.1153,
+      "step": 2133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986041214343487e-05,
+      "loss": 1.0807,
+      "step": 2134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981385408101181e-05,
+      "loss": 1.0626,
+      "step": 2135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976728165251317e-05,
+      "loss": 1.0055,
+      "step": 2136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972069487137024e-05,
+      "loss": 1.121,
+      "step": 2137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967409375101839e-05,
+      "loss": 1.0415,
+      "step": 2138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962747830489713e-05,
+      "loss": 1.075,
+      "step": 2139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958084854645018e-05,
+      "loss": 1.0679,
+      "step": 2140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953420448912535e-05,
+      "loss": 1.0958,
+      "step": 2141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948754614637448e-05,
+      "loss": 0.9748,
+      "step": 2142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494408735316537e-05,
+      "loss": 1.0496,
+      "step": 2143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493941866584231e-05,
+      "loss": 1.0356,
+      "step": 2144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934748554014702e-05,
+      "loss": 1.0245,
+      "step": 2145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930077019029376e-05,
+      "loss": 1.0907,
+      "step": 2146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925404062233586e-05,
+      "loss": 1.1137,
+      "step": 2147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920729684974986e-05,
+      "loss": 1.0609,
+      "step": 2148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916053888601649e-05,
+      "loss": 1.0509,
+      "step": 2149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911376674462048e-05,
+      "loss": 1.0315,
+      "step": 2150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906698043905067e-05,
+      "loss": 1.0232,
+      "step": 2151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490201799828001e-05,
+      "loss": 1.0092,
+      "step": 2152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489733653893657e-05,
+      "loss": 1.0914,
+      "step": 2153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892653667224861e-05,
+      "loss": 1.1005,
+      "step": 2154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887969384495403e-05,
+      "loss": 1.0681,
+      "step": 2155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883283692099114e-05,
+      "loss": 1.0519,
+      "step": 2156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878596591387329e-05,
+      "loss": 1.112,
+      "step": 2157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873908083711785e-05,
+      "loss": 0.9991,
+      "step": 2158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869218170424626e-05,
+      "loss": 1.0594,
+      "step": 2159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864526852878402e-05,
+      "loss": 1.0163,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485983413242606e-05,
+      "loss": 1.0981,
+      "step": 2161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855140010420966e-05,
+      "loss": 1.0827,
+      "step": 2162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850444488216877e-05,
+      "loss": 0.9739,
+      "step": 2163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845747567167964e-05,
+      "loss": 1.0518,
+      "step": 2164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4841049248628785e-05,
+      "loss": 1.099,
+      "step": 2165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836349533954325e-05,
+      "loss": 0.9773,
+      "step": 2166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831648424499953e-05,
+      "loss": 1.0789,
+      "step": 2167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826945921621447e-05,
+      "loss": 1.0592,
+      "step": 2168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4822242026674989e-05,
+      "loss": 1.0729,
+      "step": 2169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817536741017153e-05,
+      "loss": 1.0251,
+      "step": 2170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481283006600493e-05,
+      "loss": 1.0766,
+      "step": 2171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808122002995693e-05,
+      "loss": 1.03,
+      "step": 2172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803412553347232e-05,
+      "loss": 1.0329,
+      "step": 2173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798701718417727e-05,
+      "loss": 1.0693,
+      "step": 2174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793989499565757e-05,
+      "loss": 1.014,
+      "step": 2175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789275898150309e-05,
+      "loss": 1.0383,
+      "step": 2176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478456091553076e-05,
+      "loss": 1.0413,
+      "step": 2177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779844553066887e-05,
+      "loss": 1.0912,
+      "step": 2178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775126812118865e-05,
+      "loss": 0.9405,
+      "step": 2179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770407694047273e-05,
+      "loss": 1.03,
+      "step": 2180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765687200213079e-05,
+      "loss": 1.0912,
+      "step": 2181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760965331977653e-05,
+      "loss": 1.0663,
+      "step": 2182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756242090702756e-05,
+      "loss": 1.0615,
+      "step": 2183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751517477750544e-05,
+      "loss": 1.008,
+      "step": 2184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746791494483584e-05,
+      "loss": 1.0987,
+      "step": 2185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742064142264819e-05,
+      "loss": 1.0396,
+      "step": 2186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737335422457592e-05,
+      "loss": 1.0695,
+      "step": 2187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732605336425651e-05,
+      "loss": 1.0349,
+      "step": 2188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727873885533123e-05,
+      "loss": 1.0806,
+      "step": 2189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723141071144542e-05,
+      "loss": 0.9985,
+      "step": 2190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471840689462482e-05,
+      "loss": 1.0042,
+      "step": 2191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713671357339284e-05,
+      "loss": 1.1527,
+      "step": 2192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470893446065363e-05,
+      "loss": 1.07,
+      "step": 2193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704196205933957e-05,
+      "loss": 1.07,
+      "step": 2194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699456594546754e-05,
+      "loss": 0.9401,
+      "step": 2195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469471562785891e-05,
+      "loss": 1.1326,
+      "step": 2196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689973307237687e-05,
+      "loss": 0.9777,
+      "step": 2197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685229634050755e-05,
+      "loss": 1.0651,
+      "step": 2198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468048460966616e-05,
+      "loss": 1.0614,
+      "step": 2199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675738235452352e-05,
+      "loss": 1.0999,
+      "step": 2200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670990512778156e-05,
+      "loss": 0.9807,
+      "step": 2201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666241443012794e-05,
+      "loss": 1.0854,
+      "step": 2202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661491027525874e-05,
+      "loss": 0.9942,
+      "step": 2203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656739267687397e-05,
+      "loss": 1.0417,
+      "step": 2204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651986164867744e-05,
+      "loss": 1.0768,
+      "step": 2205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 1.0935,
+      "step": 2206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642475935768387e-05,
+      "loss": 1.0861,
+      "step": 2207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637718812231385e-05,
+      "loss": 1.0489,
+      "step": 2208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 1.0002,
+      "step": 2209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628200554042403e-05,
+      "loss": 1.0768,
+      "step": 2210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623439422135435e-05,
+      "loss": 1.0877,
+      "step": 2211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461867695685081e-05,
+      "loss": 1.025,
+      "step": 2212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613913159561996e-05,
+      "loss": 1.0879,
+      "step": 2213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609148031642852e-05,
+      "loss": 1.0454,
+      "step": 2214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604381574467616e-05,
+      "loss": 1.0191,
+      "step": 2215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459961378941091e-05,
+      "loss": 1.1029,
+      "step": 2216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594844677847743e-05,
+      "loss": 1.0311,
+      "step": 2217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45900742411535e-05,
+      "loss": 1.0374,
+      "step": 2218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585302480703959e-05,
+      "loss": 1.0322,
+      "step": 2219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580529397875265e-05,
+      "loss": 1.082,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575754994043956e-05,
+      "loss": 1.0524,
+      "step": 2221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4570979270586944e-05,
+      "loss": 1.05,
+      "step": 2222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566202228881525e-05,
+      "loss": 1.0108,
+      "step": 2223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561423870305383e-05,
+      "loss": 1.0434,
+      "step": 2224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556644196236564e-05,
+      "loss": 1.0793,
+      "step": 2225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551863208053507e-05,
+      "loss": 1.1231,
+      "step": 2226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547080907135024e-05,
+      "loss": 0.971,
+      "step": 2227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542297294860312e-05,
+      "loss": 1.0081,
+      "step": 2228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537512372608937e-05,
+      "loss": 1.0733,
+      "step": 2229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532726141760849e-05,
+      "loss": 1.0935,
+      "step": 2230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527938603696376e-05,
+      "loss": 1.0158,
+      "step": 2231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523149759796215e-05,
+      "loss": 1.0181,
+      "step": 2232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518359611441452e-05,
+      "loss": 1.0928,
+      "step": 2233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513568160013538e-05,
+      "loss": 1.0014,
+      "step": 2234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508775406894308e-05,
+      "loss": 1.0345,
+      "step": 2235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450398135346597e-05,
+      "loss": 1.1116,
+      "step": 2236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44991860011111e-05,
+      "loss": 1.0549,
+      "step": 2237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494389351212662e-05,
+      "loss": 1.0632,
+      "step": 2238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448959140515398e-05,
+      "loss": 1.0408,
+      "step": 2239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484792164318761e-05,
+      "loss": 1.0369,
+      "step": 2240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479991630091083e-05,
+      "loss": 1.0286,
+      "step": 2241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475189803855399e-05,
+      "loss": 1.0611,
+      "step": 2242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470386686996529e-05,
+      "loss": 0.9996,
+      "step": 2243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465582280899671e-05,
+      "loss": 1.1002,
+      "step": 2244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460776586950393e-05,
+      "loss": 1.0356,
+      "step": 2245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455969606534632e-05,
+      "loss": 0.9971,
+      "step": 2246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44511613410387e-05,
+      "loss": 1.0909,
+      "step": 2247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446351791849276e-05,
+      "loss": 1.0456,
+      "step": 2248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441540960353413e-05,
+      "loss": 1.0107,
+      "step": 2249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443672884793853e-05,
+      "loss": 1.0222,
+      "step": 2250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431915455992416e-05,
+      "loss": 1.0867,
+      "step": 2251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427100785903231e-05,
+      "loss": 1.0768,
+      "step": 2252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422284839059505e-05,
+      "loss": 1.0972,
+      "step": 2253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417467616850128e-05,
+      "loss": 1.0797,
+      "step": 2254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441264912066437e-05,
+      "loss": 0.9755,
+      "step": 2255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407829351891858e-05,
+      "loss": 1.0346,
+      "step": 2256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403008311922593e-05,
+      "loss": 1.0205,
+      "step": 2257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439818600214694e-05,
+      "loss": 1.0762,
+      "step": 2258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393362423955628e-05,
+      "loss": 1.0285,
+      "step": 2259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438853757873975e-05,
+      "loss": 1.0311,
+      "step": 2260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 1.0827,
+      "step": 2261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378884092800527e-05,
+      "loss": 0.9582,
+      "step": 2262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43740554548612e-05,
+      "loss": 1.0502,
+      "step": 2263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369225555465346e-05,
+      "loss": 1.1015,
+      "step": 2264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364394396005885e-05,
+      "loss": 1.075,
+      "step": 2265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359561977876102e-05,
+      "loss": 1.0315,
+      "step": 2266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354728302469645e-05,
+      "loss": 1.0177,
+      "step": 2267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434989337118052e-05,
+      "loss": 1.1123,
+      "step": 2268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43450571854031e-05,
+      "loss": 1.0958,
+      "step": 2269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340219746532111e-05,
+      "loss": 0.9708,
+      "step": 2270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335381055962657e-05,
+      "loss": 1.0115,
+      "step": 2271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330541115090188e-05,
+      "loss": 1.0865,
+      "step": 2272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432569992531052e-05,
+      "loss": 1.0394,
+      "step": 2273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320857488019826e-05,
+      "loss": 0.9955,
+      "step": 2274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316013804614644e-05,
+      "loss": 1.1368,
+      "step": 2275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311168876491869e-05,
+      "loss": 0.9913,
+      "step": 2276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306322705048759e-05,
+      "loss": 0.9782,
+      "step": 2277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430147529168292e-05,
+      "loss": 1.033,
+      "step": 2278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296626637792325e-05,
+      "loss": 1.0916,
+      "step": 2279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291776744775305e-05,
+      "loss": 0.9689,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 1.1018,
+      "step": 2281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282073246957082e-05,
+      "loss": 1.0483,
+      "step": 2282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277219644954323e-05,
+      "loss": 1.0631,
+      "step": 2283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272364809422017e-05,
+      "loss": 1.0167,
+      "step": 2284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267508741760281e-05,
+      "loss": 1.0197,
+      "step": 2285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262651443369577e-05,
+      "loss": 1.0799,
+      "step": 2286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 1.065,
+      "step": 2287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425293316000491e-05,
+      "loss": 1.0177,
+      "step": 2288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248072177833654e-05,
+      "loss": 1.0859,
+      "step": 2289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243209970538846e-05,
+      "loss": 0.9681,
+      "step": 2290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238346539522716e-05,
+      "loss": 1.0417,
+      "step": 2291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233481886187862e-05,
+      "loss": 1.0979,
+      "step": 2292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422861601193722e-05,
+      "loss": 1.1149,
+      "step": 2293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422374891817409e-05,
+      "loss": 0.9995,
+      "step": 2294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218880606302115e-05,
+      "loss": 1.0698,
+      "step": 2295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214011077725293e-05,
+      "loss": 1.0862,
+      "step": 2296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209140333847972e-05,
+      "loss": 1.085,
+      "step": 2297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204268376074858e-05,
+      "loss": 1.0365,
+      "step": 2298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199395205810988e-05,
+      "loss": 1.0003,
+      "step": 2299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194520824461773e-05,
+      "loss": 1.1166,
+      "step": 2300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189645233432953e-05,
+      "loss": 1.048,
+      "step": 2301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418476843413063e-05,
+      "loss": 0.9878,
+      "step": 2302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179890427961242e-05,
+      "loss": 1.0311,
+      "step": 2303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417501121633159e-05,
+      "loss": 1.0857,
+      "step": 2304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170130800648814e-05,
+      "loss": 1.0176,
+      "step": 2305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165249182320401e-05,
+      "loss": 1.1006,
+      "step": 2306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160366362754183e-05,
+      "loss": 1.0412,
+      "step": 2307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155482343358347e-05,
+      "loss": 1.0384,
+      "step": 2308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150597125541416e-05,
+      "loss": 1.028,
+      "step": 2309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145710710712271e-05,
+      "loss": 1.0529,
+      "step": 2310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414082310028012e-05,
+      "loss": 1.0936,
+      "step": 2311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135934295654534e-05,
+      "loss": 1.043,
+      "step": 2312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413104429824542e-05,
+      "loss": 1.0644,
+      "step": 2313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126153109463025e-05,
+      "loss": 1.0363,
+      "step": 2314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412126073071795e-05,
+      "loss": 1.1218,
+      "step": 2315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411636716342113e-05,
+      "loss": 0.9558,
+      "step": 2316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111472408983843e-05,
+      "loss": 1.0458,
+      "step": 2317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106576468817721e-05,
+      "loss": 1.0376,
+      "step": 2318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101679344334722e-05,
+      "loss": 1.0474,
+      "step": 2319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096781036947159e-05,
+      "loss": 1.0795,
+      "step": 2320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4091881548067676e-05,
+      "loss": 0.9995,
+      "step": 2321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086980879109264e-05,
+      "loss": 1.0391,
+      "step": 2322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082079031485253e-05,
+      "loss": 1.046,
+      "step": 2323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077176006609307e-05,
+      "loss": 1.084,
+      "step": 2324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072271805895443e-05,
+      "loss": 1.0699,
+      "step": 2325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 1.0467,
+      "step": 2326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062459882611676e-05,
+      "loss": 1.0761,
+      "step": 2327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057552162871486e-05,
+      "loss": 0.9912,
+      "step": 2328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052643272952795e-05,
+      "loss": 1.043,
+      "step": 2329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047733214271304e-05,
+      "loss": 1.084,
+      "step": 2330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404282198824305e-05,
+      "loss": 1.1022,
+      "step": 2331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403790959628441e-05,
+      "loss": 0.9475,
+      "step": 2332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032996039812088e-05,
+      "loss": 1.0282,
+      "step": 2333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4028081320243137e-05,
+      "loss": 1.0945,
+      "step": 2334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023165438994933e-05,
+      "loss": 0.9774,
+      "step": 2335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018248397485193e-05,
+      "loss": 1.0824,
+      "step": 2336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401333019713197e-05,
+      "loss": 1.0068,
+      "step": 2337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008410839353652e-05,
+      "loss": 1.0949,
+      "step": 2338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003490325568953e-05,
+      "loss": 1.0425,
+      "step": 2339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998568657196925e-05,
+      "loss": 1.0533,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993645835656955e-05,
+      "loss": 1.1326,
+      "step": 2341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988721862368766e-05,
+      "loss": 1.0321,
+      "step": 2342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983796738752401e-05,
+      "loss": 0.9545,
+      "step": 2343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978870466228248e-05,
+      "loss": 1.1121,
+      "step": 2344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973943046217016e-05,
+      "loss": 1.0867,
+      "step": 2345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969014480139748e-05,
+      "loss": 1.0121,
+      "step": 2346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964084769417823e-05,
+      "loss": 1.0008,
+      "step": 2347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959153915472945e-05,
+      "loss": 1.0969,
+      "step": 2348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954221919727146e-05,
+      "loss": 1.0578,
+      "step": 2349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394928878360279e-05,
+      "loss": 1.0512,
+      "step": 2350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944354508522573e-05,
+      "loss": 1.0541,
+      "step": 2351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939419095909513e-05,
+      "loss": 1.0404,
+      "step": 2352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934482547186961e-05,
+      "loss": 1.0673,
+      "step": 2353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929544863778592e-05,
+      "loss": 1.0338,
+      "step": 2354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924606047108412e-05,
+      "loss": 1.0773,
+      "step": 2355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919666098600753e-05,
+      "loss": 1.032,
+      "step": 2356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391472501968027e-05,
+      "loss": 1.0272,
+      "step": 2357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909782811771946e-05,
+      "loss": 1.0696,
+      "step": 2358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3904839476301091e-05,
+      "loss": 1.1286,
+      "step": 2359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899895014693338e-05,
+      "loss": 1.021,
+      "step": 2360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894949428374648e-05,
+      "loss": 0.9933,
+      "step": 2361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890002718771303e-05,
+      "loss": 1.1189,
+      "step": 2362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885054887309912e-05,
+      "loss": 1.0675,
+      "step": 2363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880105935417403e-05,
+      "loss": 1.031,
+      "step": 2364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 1.0671,
+      "step": 2365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870204676048373e-05,
+      "loss": 0.9575,
+      "step": 2366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865252371427328e-05,
+      "loss": 1.0314,
+      "step": 2367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860298952086118e-05,
+      "loss": 1.0422,
+      "step": 2368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385534441945328e-05,
+      "loss": 1.0455,
+      "step": 2369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850388774957683e-05,
+      "loss": 1.1009,
+      "step": 2370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845432020028511e-05,
+      "loss": 0.936,
+      "step": 2371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840474156095263e-05,
+      "loss": 1.0901,
+      "step": 2372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835515184587767e-05,
+      "loss": 1.0339,
+      "step": 2373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830555106936166e-05,
+      "loss": 1.0382,
+      "step": 2374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825593924570924e-05,
+      "loss": 1.0001,
+      "step": 2375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820631638922821e-05,
+      "loss": 1.0624,
+      "step": 2376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815668251422953e-05,
+      "loss": 1.0306,
+      "step": 2377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810703763502744e-05,
+      "loss": 1.0498,
+      "step": 2378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380573817659392e-05,
+      "loss": 1.1017,
+      "step": 2379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800771492128537e-05,
+      "loss": 1.0415,
+      "step": 2380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795803711538966e-05,
+      "loss": 0.9982,
+      "step": 2381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790834836257885e-05,
+      "loss": 1.0579,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785864867718294e-05,
+      "loss": 1.1164,
+      "step": 2383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3780893807353512e-05,
+      "loss": 1.0855,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377592165659717e-05,
+      "loss": 0.947,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770948416883205e-05,
+      "loss": 1.0324,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765974089645883e-05,
+      "loss": 1.0459,
+      "step": 2387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376099867631977e-05,
+      "loss": 1.0498,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756022178339757e-05,
+      "loss": 1.0269,
+      "step": 2389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751044597141036e-05,
+      "loss": 1.0505,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 1.0674,
+      "step": 2391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741086190829838e-05,
+      "loss": 1.0255,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373610536858931e-05,
+      "loss": 1.0993,
+      "step": 2393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731123468873993e-05,
+      "loss": 1.0039,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726140493120639e-05,
+      "loss": 1.0192,
+      "step": 2395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721156442766312e-05,
+      "loss": 1.0824,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3716171319248393e-05,
+      "loss": 1.1325,
+      "step": 2397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711185124004562e-05,
+      "loss": 0.9475,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706197858472821e-05,
+      "loss": 1.0063,
+      "step": 2399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701209524091465e-05,
+      "loss": 1.13,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369622012229911e-05,
+      "loss": 1.0099,
+      "step": 2401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691229654534677e-05,
+      "loss": 1.008,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368623812223739e-05,
+      "loss": 1.0291,
+      "step": 2403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681245526846782e-05,
+      "loss": 1.0257,
+      "step": 2404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676251869802695e-05,
+      "loss": 1.0856,
+      "step": 2405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671257152545277e-05,
+      "loss": 1.0386,
+      "step": 2406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666261376514978e-05,
+      "loss": 1.0894,
+      "step": 2407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366126454315256e-05,
+      "loss": 1.0418,
+      "step": 2408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365626665389908e-05,
+      "loss": 0.9674,
+      "step": 2409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651267710195909e-05,
+      "loss": 1.1016,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646267713484718e-05,
+      "loss": 1.0375,
+      "step": 2411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641266665207477e-05,
+      "loss": 1.041,
+      "step": 2412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636264566806473e-05,
+      "loss": 1.0389,
+      "step": 2413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363126141972428e-05,
+      "loss": 1.0982,
+      "step": 2414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626257225403783e-05,
+      "loss": 1.0206,
+      "step": 2415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362125198528817e-05,
+      "loss": 1.053,
+      "step": 2416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616245700820922e-05,
+      "loss": 1.067,
+      "step": 2417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611238373445839e-05,
+      "loss": 1.0149,
+      "step": 2418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606230004607e-05,
+      "loss": 1.0602,
+      "step": 2419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601220595748798e-05,
+      "loss": 0.9705,
+      "step": 2420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596210148315917e-05,
+      "loss": 1.121,
+      "step": 2421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591198663753358e-05,
+      "loss": 1.0132,
+      "step": 2422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35861861435064e-05,
+      "loss": 1.0659,
+      "step": 2423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358117258902063e-05,
+      "loss": 1.0991,
+      "step": 2424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576158001741932e-05,
+      "loss": 1.038,
+      "step": 2425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571142383116496e-05,
+      "loss": 0.9269,
+      "step": 2426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566125734590794e-05,
+      "loss": 1.033,
+      "step": 2427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561108057611606e-05,
+      "loss": 1.0925,
+      "step": 2428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355608935362601e-05,
+      "loss": 0.9882,
+      "step": 2429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551069624081372e-05,
+      "loss": 1.0365,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 1.0797,
+      "step": 2431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354102709410593e-05,
+      "loss": 0.9963,
+      "step": 2432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536004296571344e-05,
+      "loss": 1.0543,
+      "step": 2433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353098047927015e-05,
+      "loss": 1.0187,
+      "step": 2434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525955643651191e-05,
+      "loss": 1.0929,
+      "step": 2435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520929791163608e-05,
+      "loss": 1.0212,
+      "step": 2436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515902923256832e-05,
+      "loss": 0.9838,
+      "step": 2437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510875041380591e-05,
+      "loss": 1.1308,
+      "step": 2438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505846146984895e-05,
+      "loss": 1.0469,
+      "step": 2439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500816241520059e-05,
+      "loss": 1.0433,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495785326436678e-05,
+      "loss": 1.0588,
+      "step": 2441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349075340318565e-05,
+      "loss": 1.0521,
+      "step": 2442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 0.9856,
+      "step": 2443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480686537985661e-05,
+      "loss": 0.9555,
+      "step": 2444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475651598939936e-05,
+      "loss": 1.1025,
+      "step": 2445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347061565753303e-05,
+      "loss": 1.0749,
+      "step": 2446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465578715217289e-05,
+      "loss": 1.0597,
+      "step": 2447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460540773445332e-05,
+      "loss": 0.983,
+      "step": 2448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455501833670089e-05,
+      "loss": 1.1003,
+      "step": 2449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450461897344757e-05,
+      "loss": 1.054,
+      "step": 2450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445420965922837e-05,
+      "loss": 1.0196,
+      "step": 2451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440379040858105e-05,
+      "loss": 1.0864,
+      "step": 2452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435336123604628e-05,
+      "loss": 1.0252,
+      "step": 2453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343029221561676e-05,
+      "loss": 0.9866,
+      "step": 2454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425247318349137e-05,
+      "loss": 1.0817,
+      "step": 2455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 1.07,
+      "step": 2456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415154561794622e-05,
+      "loss": 1.0037,
+      "step": 2457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410106705418424e-05,
+      "loss": 1.0452,
+      "step": 2458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405057865583877e-05,
+      "loss": 1.0758,
+      "step": 2459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400008043747047e-05,
+      "loss": 1.0379,
+      "step": 2460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394957241364273e-05,
+      "loss": 1.0178,
+      "step": 2461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389905459892182e-05,
+      "loss": 0.947,
+      "step": 2462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3384852700787683e-05,
+      "loss": 1.0855,
+      "step": 2463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379798965507969e-05,
+      "loss": 1.0175,
+      "step": 2464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374744255510512e-05,
+      "loss": 1.0273,
+      "step": 2465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336968857225307e-05,
+      "loss": 1.0957,
+      "step": 2466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364631917193671e-05,
+      "loss": 1.0481,
+      "step": 2467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359574291790633e-05,
+      "loss": 1.0244,
+      "step": 2468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354515697502552e-05,
+      "loss": 1.0766,
+      "step": 2469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349456135788298e-05,
+      "loss": 1.0443,
+      "step": 2470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344395608107032e-05,
+      "loss": 1.0803,
+      "step": 2471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339334115918181e-05,
+      "loss": 1.0132,
+      "step": 2472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334271660681455e-05,
+      "loss": 1.053,
+      "step": 2473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332920824385684e-05,
+      "loss": 1.0719,
+      "step": 2474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324143866904602e-05,
+      "loss": 1.014,
+      "step": 2475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319078531285286e-05,
+      "loss": 1.054,
+      "step": 2476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314012238459708e-05,
+      "loss": 0.9599,
+      "step": 2477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308944989888961e-05,
+      "loss": 1.0359,
+      "step": 2478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330387678703442e-05,
+      "loss": 1.0203,
+      "step": 2479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298807631357725e-05,
+      "loss": 1.0596,
+      "step": 2480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293737524320798e-05,
+      "loss": 1.0018,
+      "step": 2481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288666467385834e-05,
+      "loss": 1.0566,
+      "step": 2482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32835944620153e-05,
+      "loss": 1.0509,
+      "step": 2483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327852150967194e-05,
+      "loss": 1.0615,
+      "step": 2484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273447611818768e-05,
+      "loss": 1.0471,
+      "step": 2485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268372769919073e-05,
+      "loss": 1.0132,
+      "step": 2486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263296985436414e-05,
+      "loss": 1.1246,
+      "step": 2487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258220259834618e-05,
+      "loss": 1.0733,
+      "step": 2488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253142594577795e-05,
+      "loss": 1.0701,
+      "step": 2489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248063991130316e-05,
+      "loss": 1.0665,
+      "step": 2490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242984450956829e-05,
+      "loss": 1.0085,
+      "step": 2491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237903975522244e-05,
+      "loss": 1.0632,
+      "step": 2492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232822566291747e-05,
+      "loss": 1.0494,
+      "step": 2493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227740224730799e-05,
+      "loss": 1.0906,
+      "step": 2494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222656952305113e-05,
+      "loss": 0.9587,
+      "step": 2495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217572750480685e-05,
+      "loss": 1.0614,
+      "step": 2496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321248762072377e-05,
+      "loss": 1.0852,
+      "step": 2497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207401564500904e-05,
+      "loss": 1.0467,
+      "step": 2498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202314583278872e-05,
+      "loss": 1.0587,
+      "step": 2499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197226678524739e-05,
+      "loss": 1.0322,
+      "step": 2500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192137851705833e-05,
+      "loss": 1.0852,
+      "step": 2501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187048104289741e-05,
+      "loss": 1.0116,
+      "step": 2502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181957437744333e-05,
+      "loss": 1.0549,
+      "step": 2503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176865853537723e-05,
+      "loss": 1.0562,
+      "step": 2504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171773353138304e-05,
+      "loss": 1.0492,
+      "step": 2505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166679938014728e-05,
+      "loss": 1.0613,
+      "step": 2506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161585609635908e-05,
+      "loss": 1.0759,
+      "step": 2507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156490369471026e-05,
+      "loss": 1.0965,
+      "step": 2508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151394218989526e-05,
+      "loss": 1.0007,
+      "step": 2509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146297159661115e-05,
+      "loss": 1.0856,
+      "step": 2510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141199192955751e-05,
+      "loss": 1.0235,
+      "step": 2511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136100320343674e-05,
+      "loss": 1.0726,
+      "step": 2512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131000543295369e-05,
+      "loss": 0.993,
+      "step": 2513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125899863281588e-05,
+      "loss": 1.0573,
+      "step": 2514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120798281773346e-05,
+      "loss": 1.0401,
+      "step": 2515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115695800241905e-05,
+      "loss": 1.1143,
+      "step": 2516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110592420158807e-05,
+      "loss": 1.0072,
+      "step": 2517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105488142995837e-05,
+      "loss": 1.0993,
+      "step": 2518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100382970225045e-05,
+      "loss": 1.014,
+      "step": 2519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095276903318736e-05,
+      "loss": 1.0085,
+      "step": 2520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 1.059,
+      "step": 2521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085062092990088e-05,
+      "loss": 1.0562,
+      "step": 2522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3079953352513654e-05,
+      "loss": 1.0609,
+      "step": 2523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074843723793507e-05,
+      "loss": 1.0273,
+      "step": 2524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069733208303242e-05,
+      "loss": 1.0992,
+      "step": 2525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064621807516701e-05,
+      "loss": 1.0309,
+      "step": 2526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059509522907998e-05,
+      "loss": 1.0489,
+      "step": 2527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054396355951482e-05,
+      "loss": 1.041,
+      "step": 2528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304928230812177e-05,
+      "loss": 1.0298,
+      "step": 2529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044167380893726e-05,
+      "loss": 1.0342,
+      "step": 2530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303905157574247e-05,
+      "loss": 1.042,
+      "step": 2531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303393489414338e-05,
+      "loss": 1.1308,
+      "step": 2532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028817337572077e-05,
+      "loss": 0.9878,
+      "step": 2533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023698907504447e-05,
+      "loss": 1.0328,
+      "step": 2534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018579605416613e-05,
+      "loss": 1.037,
+      "step": 2535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301345943278496e-05,
+      "loss": 1.094,
+      "step": 2536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008338391086124e-05,
+      "loss": 0.9996,
+      "step": 2537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003216481796984e-05,
+      "loss": 1.0059,
+      "step": 2538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998093706394676e-05,
+      "loss": 1.126,
+      "step": 2539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992970066356583e-05,
+      "loss": 0.9789,
+      "step": 2540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987845563160339e-05,
+      "loss": 1.0003,
+      "step": 2541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982720198283826e-05,
+      "loss": 1.0859,
+      "step": 2542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977593973205176e-05,
+      "loss": 0.9903,
+      "step": 2543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972466889402764e-05,
+      "loss": 0.998,
+      "step": 2544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2967338948355217e-05,
+      "loss": 1.0198,
+      "step": 2545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962210151541413e-05,
+      "loss": 1.1038,
+      "step": 2546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957080500440469e-05,
+      "loss": 1.067,
+      "step": 2547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295194999653175e-05,
+      "loss": 1.0116,
+      "step": 2548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946818641294869e-05,
+      "loss": 1.091,
+      "step": 2549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941686436209688e-05,
+      "loss": 1.0775,
+      "step": 2550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293655338275631e-05,
+      "loss": 0.9832,
+      "step": 2551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931419482415077e-05,
+      "loss": 1.028,
+      "step": 2552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926284736666585e-05,
+      "loss": 1.0941,
+      "step": 2553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292114914699167e-05,
+      "loss": 1.0111,
+      "step": 2554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291601271487141e-05,
+      "loss": 1.0452,
+      "step": 2555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291087544178713e-05,
+      "loss": 1.0441,
+      "step": 2556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905737329220394e-05,
+      "loss": 1.0173,
+      "step": 2557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900598378653007e-05,
+      "loss": 1.0403,
+      "step": 2558
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895458591567016e-05,
+      "loss": 1.0076,
+      "step": 2559
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890317969444716e-05,
+      "loss": 1.1076,
+      "step": 2560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885176513768638e-05,
+      "loss": 1.0547,
+      "step": 2561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880034226021547e-05,
+      "loss": 1.0136,
+      "step": 2562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2874891107686458e-05,
+      "loss": 1.064,
+      "step": 2563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869747160246618e-05,
+      "loss": 1.0343,
+      "step": 2564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286460238518552e-05,
+      "loss": 1.0367,
+      "step": 2565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859456783986892e-05,
+      "loss": 1.0721,
+      "step": 2566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854310358134702e-05,
+      "loss": 0.9991,
+      "step": 2567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849163109113152e-05,
+      "loss": 1.0223,
+      "step": 2568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844015038406685e-05,
+      "loss": 1.058,
+      "step": 2569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283886614749998e-05,
+      "loss": 1.0469,
+      "step": 2570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833716437877952e-05,
+      "loss": 1.0709,
+      "step": 2571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828565911025751e-05,
+      "loss": 0.9411,
+      "step": 2572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823414568428767e-05,
+      "loss": 1.1223,
+      "step": 2573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281826241157262e-05,
+      "loss": 1.0311,
+      "step": 2574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813109441943166e-05,
+      "loss": 0.9563,
+      "step": 2575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28079556610265e-05,
+      "loss": 1.0218,
+      "step": 2576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802801070308945e-05,
+      "loss": 1.0765,
+      "step": 2577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797645671277062e-05,
+      "loss": 1.03,
+      "step": 2578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792489465417643e-05,
+      "loss": 0.9863,
+      "step": 2579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787332454217714e-05,
+      "loss": 1.0819,
+      "step": 2580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 1.0032,
+      "step": 2581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777016021745579e-05,
+      "loss": 1.0554,
+      "step": 2582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771856603448583e-05,
+      "loss": 0.996,
+      "step": 2583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766696385761494e-05,
+      "loss": 1.0684,
+      "step": 2584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761535370172494e-05,
+      "loss": 0.9995,
+      "step": 2585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.9576,
+      "step": 2586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751210951242636e-05,
+      "loss": 1.0744,
+      "step": 2587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274604755087929e-05,
+      "loss": 1.0614,
+      "step": 2588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740883358569059e-05,
+      "loss": 1.0799,
+      "step": 2589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273571837580127e-05,
+      "loss": 0.9989,
+      "step": 2590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730552604065477e-05,
+      "loss": 1.0255,
+      "step": 2591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725386044851462e-05,
+      "loss": 1.0749,
+      "step": 2592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720218699649243e-05,
+      "loss": 1.0067,
+      "step": 2593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715050569949052e-05,
+      "loss": 1.0627,
+      "step": 2594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709881657241354e-05,
+      "loss": 1.0847,
+      "step": 2595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270471196301684e-05,
+      "loss": 0.9917,
+      "step": 2596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269954148876642e-05,
+      "loss": 1.0779,
+      "step": 2597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694370235981237e-05,
+      "loss": 1.0891,
+      "step": 2598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 1.0526,
+      "step": 2599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684025400772268e-05,
+      "loss": 1.0198,
+      "step": 2600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678851821331883e-05,
+      "loss": 1.0785,
+      "step": 2601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673677469323532e-05,
+      "loss": 0.9907,
+      "step": 2602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668502346239479e-05,
+      "loss": 1.0492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663326453572202e-05,
+      "loss": 1.0007,
+      "step": 2604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658149792814405e-05,
+      "loss": 1.0412,
+      "step": 2605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652972365459008e-05,
+      "loss": 1.0526,
+      "step": 2606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264779417299916e-05,
+      "loss": 1.0233,
+      "step": 2607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642615216928227e-05,
+      "loss": 1.1113,
+      "step": 2608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637435498739794e-05,
+      "loss": 1.0008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632255019927668e-05,
+      "loss": 1.0282,
+      "step": 2610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262707378198587e-05,
+      "loss": 1.0624,
+      "step": 2611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621891786408648e-05,
+      "loss": 1.0382,
+      "step": 2612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616709034690463e-05,
+      "loss": 1.0835,
+      "step": 2613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611525528325999e-05,
+      "loss": 1.0183,
+      "step": 2614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606341268810146e-05,
+      "loss": 1.1284,
+      "step": 2615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601156257638029e-05,
+      "loss": 0.9835,
+      "step": 2616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595970496304975e-05,
+      "loss": 0.9852,
+      "step": 2617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590783986306534e-05,
+      "loss": 0.9874,
+      "step": 2618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585596729138468e-05,
+      "loss": 1.0963,
+      "step": 2619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258040872629676e-05,
+      "loss": 1.0033,
+      "step": 2620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575219979277602e-05,
+      "loss": 1.0057,
+      "step": 2621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570030489577409e-05,
+      "loss": 1.1335,
+      "step": 2622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564840258692803e-05,
+      "loss": 1.0588,
+      "step": 2623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559649288120618e-05,
+      "loss": 1.046,
+      "step": 2624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554457579357906e-05,
+      "loss": 1.0912,
+      "step": 2625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549265133901934e-05,
+      "loss": 1.0617,
+      "step": 2626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544071953250177e-05,
+      "loss": 0.963,
+      "step": 2627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253887803890032e-05,
+      "loss": 1.051,
+      "step": 2628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533683392350264e-05,
+      "loss": 1.0885,
+      "step": 2629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528488015098126e-05,
+      "loss": 0.9977,
+      "step": 2630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523291908642219e-05,
+      "loss": 1.0744,
+      "step": 2631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251809507448108e-05,
+      "loss": 0.9722,
+      "step": 2632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512897514113451e-05,
+      "loss": 1.0884,
+      "step": 2633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507699229038286e-05,
+      "loss": 0.9727,
+      "step": 2634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502500220754736e-05,
+      "loss": 1.0646,
+      "step": 2635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497300490762179e-05,
+      "loss": 1.0153,
+      "step": 2636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492100040560188e-05,
+      "loss": 1.0486,
+      "step": 2637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486898871648552e-05,
+      "loss": 1.0294,
+      "step": 2638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481696985527254e-05,
+      "loss": 0.9914,
+      "step": 2639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476494383696503e-05,
+      "loss": 1.0233,
+      "step": 2640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471291067656696e-05,
+      "loss": 1.0815,
+      "step": 2641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466087038908452e-05,
+      "loss": 1.0465,
+      "step": 2642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460882298952583e-05,
+      "loss": 1.0734,
+      "step": 2643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455676849290113e-05,
+      "loss": 1.005,
+      "step": 2644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450470691422268e-05,
+      "loss": 1.0133,
+      "step": 2645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445263826850476e-05,
+      "loss": 1.0675,
+      "step": 2646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440056257076376e-05,
+      "loss": 1.0413,
+      "step": 2647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434847983601806e-05,
+      "loss": 0.9878,
+      "step": 2648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24296390079288e-05,
+      "loss": 1.0788,
+      "step": 2649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242442933155961e-05,
+      "loss": 1.0722,
+      "step": 2650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 0.9652,
+      "step": 2651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414007882742651e-05,
+      "loss": 1.0702,
+      "step": 2652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240879611330038e-05,
+      "loss": 1.0254,
+      "step": 2653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240358364917291e-05,
+      "loss": 0.9878,
+      "step": 2654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398370491863494e-05,
+      "loss": 1.0718,
+      "step": 2655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 1.0484,
+      "step": 2656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387942103712815e-05,
+      "loss": 1.0438,
+      "step": 2657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382726875879053e-05,
+      "loss": 1.0284,
+      "step": 2658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377510960878331e-05,
+      "loss": 1.0564,
+      "step": 2659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372294360214905e-05,
+      "loss": 1.1251,
+      "step": 2660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236707707539321e-05,
+      "loss": 0.9996,
+      "step": 2661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361859107917887e-05,
+      "loss": 1.0468,
+      "step": 2662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356640459293772e-05,
+      "loss": 0.9362,
+      "step": 2663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23514211310259e-05,
+      "loss": 1.0678,
+      "step": 2664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346201124619502e-05,
+      "loss": 1.0186,
+      "step": 2665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234098044158e-05,
+      "loss": 1.0472,
+      "step": 2666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335759083413015e-05,
+      "loss": 1.0271,
+      "step": 2667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330537051624355e-05,
+      "loss": 1.0409,
+      "step": 2668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232531434772004e-05,
+      "loss": 1.0697,
+      "step": 2669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320090973206269e-05,
+      "loss": 1.105,
+      "step": 2670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314866929589434e-05,
+      "loss": 1.0298,
+      "step": 2671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309642218376124e-05,
+      "loss": 1.037,
+      "step": 2672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230441684107312e-05,
+      "loss": 1.0515,
+      "step": 2673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299190799187405e-05,
+      "loss": 1.1007,
+      "step": 2674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293964094226137e-05,
+      "loss": 1.0232,
+      "step": 2675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288736727696672e-05,
+      "loss": 0.9979,
+      "step": 2676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 1.0411,
+      "step": 2677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278280015963535e-05,
+      "loss": 1.0217,
+      "step": 2678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273050673775529e-05,
+      "loss": 0.9623,
+      "step": 2679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267820676050657e-05,
+      "loss": 0.9991,
+      "step": 2680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262590024297226e-05,
+      "loss": 1.0862,
+      "step": 2681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257358720023724e-05,
+      "loss": 0.9738,
+      "step": 2682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252126764738845e-05,
+      "loss": 0.9654,
+      "step": 2683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246894159951452e-05,
+      "loss": 1.0606,
+      "step": 2684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224166090717061e-05,
+      "loss": 1.035,
+      "step": 2685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236427007905558e-05,
+      "loss": 1.0078,
+      "step": 2686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231192463665726e-05,
+      "loss": 1.0131,
+      "step": 2687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225957275960736e-05,
+      "loss": 1.0915,
+      "step": 2688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222072144630039e-05,
+      "loss": 1.0463,
+      "step": 2689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215484976194675e-05,
+      "loss": 1.0723,
+      "step": 2690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210247867153763e-05,
+      "loss": 1.0217,
+      "step": 2691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205010120688012e-05,
+      "loss": 1.0774,
+      "step": 2692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199771738307964e-05,
+      "loss": 0.9322,
+      "step": 2693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219453272152434e-05,
+      "loss": 1.0204,
+      "step": 2694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189293071848051e-05,
+      "loss": 1.0783,
+      "step": 2695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184052790790184e-05,
+      "loss": 0.9935,
+      "step": 2696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178811879862014e-05,
+      "loss": 1.0049,
+      "step": 2697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173570340574992e-05,
+      "loss": 1.0552,
+      "step": 2698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168328174440754e-05,
+      "loss": 1.0396,
+      "step": 2699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163085382971112e-05,
+      "loss": 1.0345,
+      "step": 2700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157841967678064e-05,
+      "loss": 0.9683,
+      "step": 2701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152597930073787e-05,
+      "loss": 1.1256,
+      "step": 2702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147353271670634e-05,
+      "loss": 1.0287,
+      "step": 2703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214210799398114e-05,
+      "loss": 1.024,
+      "step": 2704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136862098518021e-05,
+      "loss": 1.056,
+      "step": 2705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131615586794162e-05,
+      "loss": 0.9563,
+      "step": 2706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126368460322637e-05,
+      "loss": 1.0763,
+      "step": 2707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212112072061669e-05,
+      "loss": 1.056,
+      "step": 2708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115872369189743e-05,
+      "loss": 1.0575,
+      "step": 2709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110623407555398e-05,
+      "loss": 1.0638,
+      "step": 2710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105373837227425e-05,
+      "loss": 1.0094,
+      "step": 2711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210012365971978e-05,
+      "loss": 1.0075,
+      "step": 2712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094872876546586e-05,
+      "loss": 1.0833,
+      "step": 2713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089621489222147e-05,
+      "loss": 1.0393,
+      "step": 2714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084369499260932e-05,
+      "loss": 0.9282,
+      "step": 2715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 1.0678,
+      "step": 2716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073863717486956e-05,
+      "loss": 1.017,
+      "step": 2717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206860992870401e-05,
+      "loss": 1.0214,
+      "step": 2718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063355543343925e-05,
+      "loss": 1.1077,
+      "step": 2719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058100562922034e-05,
+      "loss": 0.9675,
+      "step": 2720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052844988953861e-05,
+      "loss": 1.0068,
+      "step": 2721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047588822955077e-05,
+      "loss": 1.0696,
+      "step": 2722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204233206644154e-05,
+      "loss": 1.0794,
+      "step": 2723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037074720929273e-05,
+      "loss": 1.0165,
+      "step": 2724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031816787934465e-05,
+      "loss": 1.0065,
+      "step": 2725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026558268973486e-05,
+      "loss": 1.031,
+      "step": 2726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021299165562859e-05,
+      "loss": 1.0462,
+      "step": 2727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016039479219293e-05,
+      "loss": 0.9918,
+      "step": 2728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010779211459649e-05,
+      "loss": 1.053,
+      "step": 2729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005518363800964e-05,
+      "loss": 1.0366,
+      "step": 2730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 0.9505,
+      "step": 2731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199499493485546e-05,
+      "loss": 1.054,
+      "step": 2732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989732356603545e-05,
+      "loss": 1.0858,
+      "step": 2733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984469204522399e-05,
+      "loss": 0.9678,
+      "step": 2734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979205480129897e-05,
+      "loss": 1.0546,
+      "step": 2735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973941184944067e-05,
+      "loss": 1.05,
+      "step": 2736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968676320483103e-05,
+      "loss": 1.0493,
+      "step": 2737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963410888265375e-05,
+      "loss": 1.0511,
+      "step": 2738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958144889809397e-05,
+      "loss": 0.9936,
+      "step": 2739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952878326633872e-05,
+      "loss": 1.0511,
+      "step": 2740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1947611200257641e-05,
+      "loss": 1.0499,
+      "step": 2741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194234351219972e-05,
+      "loss": 1.0141,
+      "step": 2742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193707526397929e-05,
+      "loss": 1.0121,
+      "step": 2743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1931806457115678e-05,
+      "loss": 1.0395,
+      "step": 2744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926537093128392e-05,
+      "loss": 1.0079,
+      "step": 2745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921267173537085e-05,
+      "loss": 1.0036,
+      "step": 2746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191599669986158e-05,
+      "loss": 1.1343,
+      "step": 2747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1910725673621856e-05,
+      "loss": 0.999,
+      "step": 2748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190545409633805e-05,
+      "loss": 0.9983,
+      "step": 2749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900181969530462e-05,
+      "loss": 1.0183,
+      "step": 2750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894909294719547e-05,
+      "loss": 1.0721,
+      "step": 2751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889636073425919e-05,
+      "loss": 1.0334,
+      "step": 2752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884362307170346e-05,
+      "loss": 1.027,
+      "step": 2753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187908799747376e-05,
+      "loss": 1.0697,
+      "step": 2754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 1.0024,
+      "step": 2755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868537753842052e-05,
+      "loss": 1.0624,
+      "step": 2756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863261822949563e-05,
+      "loss": 1.0721,
+      "step": 2757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185798535470134e-05,
+      "loss": 0.9675,
+      "step": 2758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852708350619091e-05,
+      "loss": 1.0001,
+      "step": 2759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847430812224679e-05,
+      "loss": 1.0809,
+      "step": 2760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842152741040117e-05,
+      "loss": 1.0532,
+      "step": 2761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183687413858758e-05,
+      "loss": 1.0269,
+      "step": 2762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831595006389385e-05,
+      "loss": 0.9967,
+      "step": 2763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826315345968014e-05,
+      "loss": 1.0672,
+      "step": 2764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821035158846094e-05,
+      "loss": 1.0134,
+      "step": 2765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815754446546405e-05,
+      "loss": 1.0513,
+      "step": 2766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810473210591882e-05,
+      "loss": 1.028,
+      "step": 2767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805191452505602e-05,
+      "loss": 1.1082,
+      "step": 2768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799909173810802e-05,
+      "loss": 1.0242,
+      "step": 2769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794626376030866e-05,
+      "loss": 0.9625,
+      "step": 2770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178934306068933e-05,
+      "loss": 1.0532,
+      "step": 2771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178405922930987e-05,
+      "loss": 1.0126,
+      "step": 2772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778774883416325e-05,
+      "loss": 1.0475,
+      "step": 2773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177349002453267e-05,
+      "loss": 1.0482,
+      "step": 2774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768204654183033e-05,
+      "loss": 0.992,
+      "step": 2775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762918773891691e-05,
+      "loss": 1.016,
+      "step": 2776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757632385183064e-05,
+      "loss": 1.0953,
+      "step": 2777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752345489581727e-05,
+      "loss": 1.0806,
+      "step": 2778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747058088612389e-05,
+      "loss": 1.03,
+      "step": 2779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741770183799912e-05,
+      "loss": 1.08,
+      "step": 2780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.9834,
+      "step": 2781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731192868745716e-05,
+      "loss": 1.1282,
+      "step": 2782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725903461554444e-05,
+      "loss": 0.9812,
+      "step": 2783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720613556620925e-05,
+      "loss": 1.0173,
+      "step": 2784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715323155470745e-05,
+      "loss": 1.0956,
+      "step": 2785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710032259629631e-05,
+      "loss": 0.9961,
+      "step": 2786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704740870623446e-05,
+      "loss": 1.0241,
+      "step": 2787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699448989978209e-05,
+      "loss": 1.0379,
+      "step": 2788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694156619220065e-05,
+      "loss": 0.9425,
+      "step": 2789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688863759875313e-05,
+      "loss": 1.057,
+      "step": 2790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683570413470384e-05,
+      "loss": 1.024,
+      "step": 2791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678276581531861e-05,
+      "loss": 1.0259,
+      "step": 2792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672982265586456e-05,
+      "loss": 1.0284,
+      "step": 2793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667687467161025e-05,
+      "loss": 1.0562,
+      "step": 2794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662392187782558e-05,
+      "loss": 1.0682,
+      "step": 2795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657096428978193e-05,
+      "loss": 1.0766,
+      "step": 2796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651800192275197e-05,
+      "loss": 1.0254,
+      "step": 2797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646503479200983e-05,
+      "loss": 0.9951,
+      "step": 2798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641206291283098e-05,
+      "loss": 1.0481,
+      "step": 2799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163590863004922e-05,
+      "loss": 1.0537,
+      "step": 2800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630610497027176e-05,
+      "loss": 1.0319,
+      "step": 2801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625311893744912e-05,
+      "loss": 1.0984,
+      "step": 2802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620012821730533e-05,
+      "loss": 0.9705,
+      "step": 2803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161471328251226e-05,
+      "loss": 1.0282,
+      "step": 2804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609413277618448e-05,
+      "loss": 1.0402,
+      "step": 2805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604112808577603e-05,
+      "loss": 1.067,
+      "step": 2806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159881187691835e-05,
+      "loss": 0.9875,
+      "step": 2807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593510484169453e-05,
+      "loss": 1.0371,
+      "step": 2808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588208631859808e-05,
+      "loss": 0.9572,
+      "step": 2809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582906321518438e-05,
+      "loss": 1.061,
+      "step": 2810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577603554674515e-05,
+      "loss": 1.0266,
+      "step": 2811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572300332857322e-05,
+      "loss": 0.9939,
+      "step": 2812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566996657596287e-05,
+      "loss": 1.0347,
+      "step": 2813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561692530420966e-05,
+      "loss": 0.963,
+      "step": 2814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556387952861036e-05,
+      "loss": 1.057,
+      "step": 2815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551082926446322e-05,
+      "loss": 1.0871,
+      "step": 2816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545777452706761e-05,
+      "loss": 0.998,
+      "step": 2817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154047153317243e-05,
+      "loss": 1.0422,
+      "step": 2818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535165169373528e-05,
+      "loss": 1.0359,
+      "step": 2819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529858362840383e-05,
+      "loss": 1.036,
+      "step": 2820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524551115103455e-05,
+      "loss": 0.93,
+      "step": 2821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151924342769333e-05,
+      "loss": 1.0154,
+      "step": 2822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513935302140717e-05,
+      "loss": 1.0731,
+      "step": 2823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508626739976454e-05,
+      "loss": 1.0346,
+      "step": 2824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503317742731508e-05,
+      "loss": 0.937,
+      "step": 2825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498008311936965e-05,
+      "loss": 1.0251,
+      "step": 2826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492698449124042e-05,
+      "loss": 1.0996,
+      "step": 2827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487388155824076e-05,
+      "loss": 1.0206,
+      "step": 2828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482077433568525e-05,
+      "loss": 1.0189,
+      "step": 2829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476766283888986e-05,
+      "loss": 1.0981,
+      "step": 2830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471454708317163e-05,
+      "loss": 1.0099,
+      "step": 2831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466142708384889e-05,
+      "loss": 1.0896,
+      "step": 2832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 1.0921,
+      "step": 2833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455517441566928e-05,
+      "loss": 1.0234,
+      "step": 2834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145020417774552e-05,
+      "loss": 1.0073,
+      "step": 2835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444890495692214e-05,
+      "loss": 1.0464,
+      "step": 2836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439576396939449e-05,
+      "loss": 1.0209,
+      "step": 2837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434261883019784e-05,
+      "loss": 1.0123,
+      "step": 2838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428946955465896e-05,
+      "loss": 0.9919,
+      "step": 2839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423631615810593e-05,
+      "loss": 1.0141,
+      "step": 2840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418315865586789e-05,
+      "loss": 1.0329,
+      "step": 2841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412999706327522e-05,
+      "loss": 1.0082,
+      "step": 2842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1407683139565942e-05,
+      "loss": 0.9923,
+      "step": 2843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402366166835326e-05,
+      "loss": 1.1213,
+      "step": 2844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397048789669061e-05,
+      "loss": 0.9813,
+      "step": 2845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391731009600655e-05,
+      "loss": 0.9595,
+      "step": 2846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386412828163731e-05,
+      "loss": 1.0762,
+      "step": 2847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381094246892022e-05,
+      "loss": 1.0622,
+      "step": 2848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375775267319387e-05,
+      "loss": 1.0496,
+      "step": 2849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1370455890979792e-05,
+      "loss": 0.9879,
+      "step": 2850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365136119407318e-05,
+      "loss": 1.0593,
+      "step": 2851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359815954136164e-05,
+      "loss": 1.0461,
+      "step": 2852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354495396700633e-05,
+      "loss": 0.9735,
+      "step": 2853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349174448635158e-05,
+      "loss": 0.9746,
+      "step": 2854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134385311147427e-05,
+      "loss": 1.0804,
+      "step": 2855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1338531386752618e-05,
+      "loss": 1.0532,
+      "step": 2856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333209276004959e-05,
+      "loss": 1.0534,
+      "step": 2857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1327886780766169e-05,
+      "loss": 1.0425,
+      "step": 2858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322563902571227e-05,
+      "loss": 1.0152,
+      "step": 2859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317240642955226e-05,
+      "loss": 0.989,
+      "step": 2860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311917003453366e-05,
+      "loss": 1.0863,
+      "step": 2861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306592985600962e-05,
+      "loss": 0.9953,
+      "step": 2862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301268590933434e-05,
+      "loss": 1.0318,
+      "step": 2863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295943820986312e-05,
+      "loss": 1.0146,
+      "step": 2864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290618677295235e-05,
+      "loss": 1.0193,
+      "step": 2865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285293161395948e-05,
+      "loss": 1.0744,
+      "step": 2866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279967274824302e-05,
+      "loss": 0.989,
+      "step": 2867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127464101911626e-05,
+      "loss": 0.9594,
+      "step": 2868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269314395807889e-05,
+      "loss": 1.0349,
+      "step": 2869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126398740643536e-05,
+      "loss": 1.0081,
+      "step": 2870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125866005253495e-05,
+      "loss": 1.0255,
+      "step": 2871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253332335643043e-05,
+      "loss": 1.0541,
+      "step": 2872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124800425729613e-05,
+      "loss": 1.0034,
+      "step": 2873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242675819030797e-05,
+      "loss": 1.0031,
+      "step": 2874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237347022383747e-05,
+      "loss": 1.0795,
+      "step": 2875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123201786889177e-05,
+      "loss": 1.0293,
+      "step": 2876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226688360091776e-05,
+      "loss": 1.0284,
+      "step": 2877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221358497520769e-05,
+      "loss": 1.0215,
+      "step": 2878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1216028282715854e-05,
+      "loss": 1.0065,
+      "step": 2879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1210697717214238e-05,
+      "loss": 1.0186,
+      "step": 2880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 1.0624,
+      "step": 2881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200035540270247e-05,
+      "loss": 1.0684,
+      "step": 2882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119470393190279e-05,
+      "loss": 1.0023,
+      "step": 2883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189371978988477e-05,
+      "loss": 1.0281,
+      "step": 2884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184039683065014e-05,
+      "loss": 0.9865,
+      "step": 2885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178707045670205e-05,
+      "loss": 0.9651,
+      "step": 2886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173374068341962e-05,
+      "loss": 1.0368,
+      "step": 2887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168040752618293e-05,
+      "loss": 1.0087,
+      "step": 2888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162707100037296e-05,
+      "loss": 1.1096,
+      "step": 2889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157373112137171e-05,
+      "loss": 1.0084,
+      "step": 2890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152038790456212e-05,
+      "loss": 1.0349,
+      "step": 2891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146704136532818e-05,
+      "loss": 1.0322,
+      "step": 2892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141369151905474e-05,
+      "loss": 1.0267,
+      "step": 2893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136033838112765e-05,
+      "loss": 1.0337,
+      "step": 2894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113069819669337e-05,
+      "loss": 1.0004,
+      "step": 2895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125362229186056e-05,
+      "loss": 1.0401,
+      "step": 2896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11200259371297e-05,
+      "loss": 1.0366,
+      "step": 2897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114689322063255e-05,
+      "loss": 0.918,
+      "step": 2898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1109352385525782e-05,
+      "loss": 1.0555,
+      "step": 2899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110401512905642e-05,
+      "loss": 1.0572,
+      "step": 2900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098677554194418e-05,
+      "loss": 0.9823,
+      "step": 2901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093339662479099e-05,
+      "loss": 1.0321,
+      "step": 2902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088001455449886e-05,
+      "loss": 1.0593,
+      "step": 2903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082662934646296e-05,
+      "loss": 0.9907,
+      "step": 2904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107732410160793e-05,
+      "loss": 1.0161,
+      "step": 2905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107198495787448e-05,
+      "loss": 1.1155,
+      "step": 2906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066645504985734e-05,
+      "loss": 0.9711,
+      "step": 2907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106130574448156e-05,
+      "loss": 1.0551,
+      "step": 2908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055965677901921e-05,
+      "loss": 1.0514,
+      "step": 2909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050625306786865e-05,
+      "loss": 1.0691,
+      "step": 2910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 1.012,
+      "step": 2911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039943657111152e-05,
+      "loss": 0.968,
+      "step": 2912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034602381631026e-05,
+      "loss": 1.0211,
+      "step": 2913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029260807776557e-05,
+      "loss": 1.0627,
+      "step": 2914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023918937088225e-05,
+      "loss": 1.0714,
+      "step": 2915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018576771106605e-05,
+      "loss": 0.9513,
+      "step": 2916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013234311372353e-05,
+      "loss": 1.0064,
+      "step": 2917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007891559426202e-05,
+      "loss": 0.9677,
+      "step": 2918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002548516808984e-05,
+      "loss": 1.0318,
+      "step": 2919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997205185061599e-05,
+      "loss": 1.0499,
+      "step": 2920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991861565725044e-05,
+      "loss": 1.0344,
+      "step": 2921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098651766034039e-05,
+      "loss": 0.9806,
+      "step": 2922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981173470448796e-05,
+      "loss": 1.0218,
+      "step": 2923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975828997591496e-05,
+      "loss": 1.0347,
+      "step": 2924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0970484243309815e-05,
+      "loss": 1.0347,
+      "step": 2925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965139209145153e-05,
+      "loss": 1.0173,
+      "step": 2926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959793896638994e-05,
+      "loss": 1.0451,
+      "step": 2927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954448307332894e-05,
+      "loss": 1.0195,
+      "step": 2928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09491024427685e-05,
+      "loss": 1.0616,
+      "step": 2929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943756304487531e-05,
+      "loss": 1.0375,
+      "step": 2930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938409894031793e-05,
+      "loss": 1.0814,
+      "step": 2931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933063212943162e-05,
+      "loss": 0.9979,
+      "step": 2932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927716262763592e-05,
+      "loss": 0.9404,
+      "step": 2933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922369045035121e-05,
+      "loss": 1.0811,
+      "step": 2934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917021561299864e-05,
+      "loss": 1.0397,
+      "step": 2935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911673813100001e-05,
+      "loss": 1.005,
+      "step": 2936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906325801977804e-05,
+      "loss": 1.1054,
+      "step": 2937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090097752947561e-05,
+      "loss": 1.0566,
+      "step": 2938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895628997135837e-05,
+      "loss": 1.0552,
+      "step": 2939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890280206500971e-05,
+      "loss": 1.0099,
+      "step": 2940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884931159113585e-05,
+      "loss": 1.0482,
+      "step": 2941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879581856516318e-05,
+      "loss": 1.0627,
+      "step": 2942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874232300251874e-05,
+      "loss": 1.0393,
+      "step": 2943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868882491863048e-05,
+      "loss": 0.994,
+      "step": 2944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863532432892698e-05,
+      "loss": 1.0638,
+      "step": 2945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858182124883754e-05,
+      "loss": 0.9966,
+      "step": 2946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852831569379217e-05,
+      "loss": 1.0233,
+      "step": 2947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847480767922163e-05,
+      "loss": 1.1016,
+      "step": 2948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842129722055738e-05,
+      "loss": 0.9586,
+      "step": 2949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083677843332316e-05,
+      "loss": 1.0053,
+      "step": 2950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831426903267706e-05,
+      "loss": 0.9746,
+      "step": 2951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826075133432739e-05,
+      "loss": 1.0829,
+      "step": 2952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820723125361685e-05,
+      "loss": 0.9734,
+      "step": 2953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815370880598035e-05,
+      "loss": 1.0913,
+      "step": 2954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810018400685352e-05,
+      "loss": 1.0533,
+      "step": 2955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804665687167262e-05,
+      "loss": 0.9552,
+      "step": 2956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079931274158746e-05,
+      "loss": 1.0274,
+      "step": 2957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793959565489717e-05,
+      "loss": 1.048,
+      "step": 2958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078860616041786e-05,
+      "loss": 1.0353,
+      "step": 2959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783252527915784e-05,
+      "loss": 0.97,
+      "step": 2960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777898669527449e-05,
+      "loss": 1.0119,
+      "step": 2961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077254458679689e-05,
+      "loss": 0.9623,
+      "step": 2962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767190281268187e-05,
+      "loss": 1.0434,
+      "step": 2963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761835754485506e-05,
+      "loss": 1.0092,
+      "step": 2964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756481007993063e-05,
+      "loss": 1.022,
+      "step": 2965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751126043335141e-05,
+      "loss": 1.0018,
+      "step": 2966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0745770862056082e-05,
+      "loss": 1.0339,
+      "step": 2967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740415465700302e-05,
+      "loss": 1.0822,
+      "step": 2968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735059855812269e-05,
+      "loss": 1.008,
+      "step": 2969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729704033936511e-05,
+      "loss": 1.0528,
+      "step": 2970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724348001617626e-05,
+      "loss": 0.993,
+      "step": 2971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718991760400266e-05,
+      "loss": 1.0052,
+      "step": 2972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713635311829143e-05,
+      "loss": 1.0752,
+      "step": 2973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708278657449035e-05,
+      "loss": 1.0546,
+      "step": 2974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702921798804778e-05,
+      "loss": 1.0235,
+      "step": 2975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697564737441254e-05,
+      "loss": 1.0676,
+      "step": 2976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692207474903421e-05,
+      "loss": 0.9919,
+      "step": 2977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068685001273629e-05,
+      "loss": 1.0055,
+      "step": 2978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681492352484918e-05,
+      "loss": 1.0564,
+      "step": 2979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676134495694439e-05,
+      "loss": 1.0719,
+      "step": 2980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670776443910024e-05,
+      "loss": 0.9935,
+      "step": 2981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0665418198676918e-05,
+      "loss": 1.0716,
+      "step": 2982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660059761540405e-05,
+      "loss": 1.0704,
+      "step": 2983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654701134045835e-05,
+      "loss": 0.9323,
+      "step": 2984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649342317738616e-05,
+      "loss": 1.0242,
+      "step": 2985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643983314164195e-05,
+      "loss": 0.9934,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638624124868092e-05,
+      "loss": 1.1051,
+      "step": 2987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633264751395867e-05,
+      "loss": 0.9926,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 1.041,
+      "step": 2989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062254545810557e-05,
+      "loss": 1.0743,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617185541378895e-05,
+      "loss": 1.041,
+      "step": 2991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611825446658883e-05,
+      "loss": 1.042,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606465175491359e-05,
+      "loss": 1.054,
+      "step": 2993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601104729422196e-05,
+      "loss": 1.0149,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595744109997326e-05,
+      "loss": 1.0239,
+      "step": 2995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590383318762724e-05,
+      "loss": 1.0675,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585022357264419e-05,
+      "loss": 1.0395,
+      "step": 2997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579661227048484e-05,
+      "loss": 1.0427,
+      "step": 2998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057429992966104e-05,
+      "loss": 1.0456,
+      "step": 2999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0568938466648264e-05,
+      "loss": 1.008,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563576839556375e-05,
+      "loss": 1.0112,
+      "step": 3001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055821504993164e-05,
+      "loss": 0.9838,
+      "step": 3002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552853099320375e-05,
+      "loss": 1.0564,
+      "step": 3003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547490989268933e-05,
+      "loss": 1.003,
+      "step": 3004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054212872132373e-05,
+      "loss": 1.0327,
+      "step": 3005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536766297031216e-05,
+      "loss": 1.0315,
+      "step": 3006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531403717937888e-05,
+      "loss": 1.0523,
+      "step": 3007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526040985590287e-05,
+      "loss": 0.9339,
+      "step": 3008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520678101534999e-05,
+      "loss": 1.0974,
+      "step": 3009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515315067318652e-05,
+      "loss": 1.1121,
+      "step": 3010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509951884487926e-05,
+      "loss": 0.9844,
+      "step": 3011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050458855458953e-05,
+      "loss": 1.0455,
+      "step": 3012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499225079170228e-05,
+      "loss": 1.0538,
+      "step": 3013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493861459776814e-05,
+      "loss": 1.0693,
+      "step": 3014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488497697956134e-05,
+      "loss": 1.0167,
+      "step": 3015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483133795255072e-05,
+      "loss": 0.9616,
+      "step": 3016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047776975322055e-05,
+      "loss": 1.0277,
+      "step": 3017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472405573399533e-05,
+      "loss": 1.109,
+      "step": 3018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467041257339023e-05,
+      "loss": 1.0496,
+      "step": 3019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461676806586065e-05,
+      "loss": 1.0269,
+      "step": 3020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456312222687741e-05,
+      "loss": 1.0413,
+      "step": 3021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045094750719117e-05,
+      "loss": 1.0444,
+      "step": 3022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044558266164351e-05,
+      "loss": 0.9746,
+      "step": 3023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440217687591951e-05,
+      "loss": 1.0848,
+      "step": 3024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0434852586583737e-05,
+      "loss": 1.023,
+      "step": 3025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429487360166129e-05,
+      "loss": 0.9635,
+      "step": 3026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424122009886436e-05,
+      "loss": 1.0361,
+      "step": 3027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418756537291996e-05,
+      "loss": 1.0469,
+      "step": 3028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413390943930191e-05,
+      "loss": 1.0169,
+      "step": 3029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040802523134843e-05,
+      "loss": 1.0241,
+      "step": 3030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 1.0419,
+      "step": 3031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397293454714847e-05,
+      "loss": 1.0159,
+      "step": 3032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391927393758021e-05,
+      "loss": 1.0241,
+      "step": 3033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386561219771222e-05,
+      "loss": 0.9932,
+      "step": 3034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381194934302032e-05,
+      "loss": 1.0772,
+      "step": 3035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375828538898055e-05,
+      "loss": 0.9452,
+      "step": 3036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037046203510694e-05,
+      "loss": 1.0383,
+      "step": 3037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365095424476357e-05,
+      "loss": 1.0287,
+      "step": 3038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359728708554013e-05,
+      "loss": 0.9882,
+      "step": 3039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354361888887642e-05,
+      "loss": 0.9749,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 1.0845,
+      "step": 3041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343627944513913e-05,
+      "loss": 1.0327,
+      "step": 3042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338260822902166e-05,
+      "loss": 1.0374,
+      "step": 3043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033289360373763e-05,
+      "loss": 0.9594,
+      "step": 3044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327526288568184e-05,
+      "loss": 1.0446,
+      "step": 3045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322158878941733e-05,
+      "loss": 1.0299,
+      "step": 3046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031679137640621e-05,
+      "loss": 1.0329,
+      "step": 3047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031142378250958e-05,
+      "loss": 1.0236,
+      "step": 3048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306056098799833e-05,
+      "loss": 1.0691,
+      "step": 3049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0300688326824981e-05,
+      "loss": 1.0064,
+      "step": 3050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295320468133066e-05,
+      "loss": 0.9934,
+      "step": 3051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289952524272147e-05,
+      "loss": 1.1247,
+      "step": 3052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284584496790317e-05,
+      "loss": 0.9757,
+      "step": 3053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279216387235691e-05,
+      "loss": 1.0053,
+      "step": 3054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273848197156401e-05,
+      "loss": 1.0232,
+      "step": 3055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268479928100615e-05,
+      "loss": 0.9709,
+      "step": 3056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263111581616505e-05,
+      "loss": 0.9828,
+      "step": 3057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257743159252285e-05,
+      "loss": 1.0107,
+      "step": 3058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252374662556177e-05,
+      "loss": 1.0369,
+      "step": 3059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247006093076434e-05,
+      "loss": 1.0143,
+      "step": 3060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241637452361323e-05,
+      "loss": 1.0391,
+      "step": 3061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236268741959135e-05,
+      "loss": 1.0117,
+      "step": 3062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230899963418182e-05,
+      "loss": 1.107,
+      "step": 3063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022553111828679e-05,
+      "loss": 0.9956,
+      "step": 3064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220162208113309e-05,
+      "loss": 1.0322,
+      "step": 3065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214793234446106e-05,
+      "loss": 1.0564,
+      "step": 3066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 1.0136,
+      "step": 3067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204055102824106e-05,
+      "loss": 0.9994,
+      "step": 3068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198685947966129e-05,
+      "loss": 1.0656,
+      "step": 3069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193316735808085e-05,
+      "loss": 0.9761,
+      "step": 3070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187947467898425e-05,
+      "loss": 0.9675,
+      "step": 3071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018257814578562e-05,
+      "loss": 1.0211,
+      "step": 3072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177208771018159e-05,
+      "loss": 1.0889,
+      "step": 3073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171839345144542e-05,
+      "loss": 0.9979,
+      "step": 3074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166469869713283e-05,
+      "loss": 0.9823,
+      "step": 3075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161100346272913e-05,
+      "loss": 1.0616,
+      "step": 3076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155730776371984e-05,
+      "loss": 1.1099,
+      "step": 3077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150361161559048e-05,
+      "loss": 1.0111,
+      "step": 3078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144991503382676e-05,
+      "loss": 1.0522,
+      "step": 3079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139621803391454e-05,
+      "loss": 1.0856,
+      "step": 3080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134252063133976e-05,
+      "loss": 0.9593,
+      "step": 3081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012888228415885e-05,
+      "loss": 1.1107,
+      "step": 3082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123512468014694e-05,
+      "loss": 1.0424,
+      "step": 3083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118142616250137e-05,
+      "loss": 1.0931,
+      "step": 3084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112772730413816e-05,
+      "loss": 0.9972,
+      "step": 3085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107402812054386e-05,
+      "loss": 1.0224,
+      "step": 3086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102032862720504e-05,
+      "loss": 1.0738,
+      "step": 3087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096662883960833e-05,
+      "loss": 0.9542,
+      "step": 3088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009129287732405e-05,
+      "loss": 0.9747,
+      "step": 3089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085922844358844e-05,
+      "loss": 1.0644,
+      "step": 3090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080552786613899e-05,
+      "loss": 1.0614,
+      "step": 3091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075182705637924e-05,
+      "loss": 0.9673,
+      "step": 3092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069812602979617e-05,
+      "loss": 1.0211,
+      "step": 3093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064442480187692e-05,
+      "loss": 1.0606,
+      "step": 3094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059072338810865e-05,
+      "loss": 1.0534,
+      "step": 3095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053702180397859e-05,
+      "loss": 1.0391,
+      "step": 3096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048332006497406e-05,
+      "loss": 1.1001,
+      "step": 3097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042961818658235e-05,
+      "loss": 1.0444,
+      "step": 3098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037591618429077e-05,
+      "loss": 0.9603,
+      "step": 3099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032221407358683e-05,
+      "loss": 1.0995,
+      "step": 3100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026851186995786e-05,
+      "loss": 0.988,
+      "step": 3101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021480958889138e-05,
+      "loss": 0.9977,
+      "step": 3102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016110724587487e-05,
+      "loss": 0.9743,
+      "step": 3103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001074048563958e-05,
+      "loss": 1.0886,
+      "step": 3104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005370243594165e-05,
+      "loss": 0.9727,
+      "step": 3105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.0761,
+      "step": 3106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994629756405835e-06,
+      "loss": 1.0297,
+      "step": 3107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989259514360426e-06,
+      "loss": 0.972,
+      "step": 3108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983889275412516e-06,
+      "loss": 1.0179,
+      "step": 3109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978519041110864e-06,
+      "loss": 1.0301,
+      "step": 3110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973148813004216e-06,
+      "loss": 1.0912,
+      "step": 3111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96777859264132e-06,
+      "loss": 1.018,
+      "step": 3112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962408381570924e-06,
+      "loss": 1.0238,
+      "step": 3113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95703818134177e-06,
+      "loss": 1.0439,
+      "step": 3114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951667993502599e-06,
+      "loss": 0.9815,
+      "step": 3115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946297819602143e-06,
+      "loss": 1.0607,
+      "step": 3116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940927661189137e-06,
+      "loss": 0.9957,
+      "step": 3117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935557519812313e-06,
+      "loss": 1.0904,
+      "step": 3118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930187397020385e-06,
+      "loss": 1.0157,
+      "step": 3119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924817294362079e-06,
+      "loss": 0.9755,
+      "step": 3120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919447213386103e-06,
+      "loss": 1.0182,
+      "step": 3121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91407715564116e-06,
+      "loss": 1.0804,
+      "step": 3122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908707122675954e-06,
+      "loss": 0.9938,
+      "step": 3123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903337116039172e-06,
+      "loss": 1.0475,
+      "step": 3124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897967137279498e-06,
+      "loss": 1.0871,
+      "step": 3125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892597187945616e-06,
+      "loss": 1.0087,
+      "step": 3126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887227269586184e-06,
+      "loss": 0.9691,
+      "step": 3127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881857383749868e-06,
+      "loss": 1.0888,
+      "step": 3128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87648753198531e-06,
+      "loss": 1.0858,
+      "step": 3129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871117715841151e-06,
+      "loss": 0.9664,
+      "step": 3130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865747936866027e-06,
+      "loss": 1.0255,
+      "step": 3131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860378196608549e-06,
+      "loss": 1.0584,
+      "step": 3132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855008496617326e-06,
+      "loss": 1.0105,
+      "step": 3133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.849638838440954e-06,
+      "loss": 1.0857,
+      "step": 3134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844269223628016e-06,
+      "loss": 0.9539,
+      "step": 3135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838899653727088e-06,
+      "loss": 1.0718,
+      "step": 3136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83353013028672e-06,
+      "loss": 1.0156,
+      "step": 3137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828160654855465e-06,
+      "loss": 1.0369,
+      "step": 3138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822791228981844e-06,
+      "loss": 1.026,
+      "step": 3139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81742185421438e-06,
+      "loss": 1.0182,
+      "step": 3140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812052532101579e-06,
+      "loss": 1.0312,
+      "step": 3141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806683264191916e-06,
+      "loss": 1.0414,
+      "step": 3142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801314052033873e-06,
+      "loss": 1.0645,
+      "step": 3143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795944897175896e-06,
+      "loss": 0.9817,
+      "step": 3144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 1.01,
+      "step": 3145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785206765553896e-06,
+      "loss": 1.062,
+      "step": 3146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779837791886693e-06,
+      "loss": 1.0515,
+      "step": 3147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774468881713216e-06,
+      "loss": 0.9935,
+      "step": 3148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769100036581823e-06,
+      "loss": 1.0562,
+      "step": 3149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763731258040865e-06,
+      "loss": 1.0019,
+      "step": 3150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75836254763868e-06,
+      "loss": 0.9679,
+      "step": 3151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752993906923568e-06,
+      "loss": 0.9633,
+      "step": 3152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747625337443826e-06,
+      "loss": 1.0661,
+      "step": 3153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742256840747718e-06,
+      "loss": 1.0377,
+      "step": 3154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736888418383498e-06,
+      "loss": 0.9954,
+      "step": 3155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73152007189939e-06,
+      "loss": 1.0491,
+      "step": 3156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7261518028436e-06,
+      "loss": 1.0134,
+      "step": 3157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720783612764314e-06,
+      "loss": 0.9536,
+      "step": 3158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715415503209685e-06,
+      "loss": 1.0812,
+      "step": 3159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710047475727854e-06,
+      "loss": 0.9927,
+      "step": 3160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70467953186694e-06,
+      "loss": 1.0618,
+      "step": 3161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69931167317502e-06,
+      "loss": 1.0367,
+      "step": 3162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693943901200169e-06,
+      "loss": 1.035,
+      "step": 3163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688576217490424e-06,
+      "loss": 1.0406,
+      "step": 3164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683208623593793e-06,
+      "loss": 1.0627,
+      "step": 3165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677841121058274e-06,
+      "loss": 0.9606,
+      "step": 3166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672473711431821e-06,
+      "loss": 1.0777,
+      "step": 3167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66710639626237e-06,
+      "loss": 1.0452,
+      "step": 3168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661739177097836e-06,
+      "loss": 0.9508,
+      "step": 3169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65637205548609e-06,
+      "loss": 1.0686,
+      "step": 3170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 1.0241,
+      "step": 3171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64563811111236e-06,
+      "loss": 1.0015,
+      "step": 3172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640271291445987e-06,
+      "loss": 1.0725,
+      "step": 3173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634904575523645e-06,
+      "loss": 1.0309,
+      "step": 3174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629537964893063e-06,
+      "loss": 0.9707,
+      "step": 3175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624171461101947e-06,
+      "loss": 0.9521,
+      "step": 3176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618805065697971e-06,
+      "loss": 1.0992,
+      "step": 3177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613438780228777e-06,
+      "loss": 0.9787,
+      "step": 3178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608072606241982e-06,
+      "loss": 1.075,
+      "step": 3179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602706545285156e-06,
+      "loss": 0.9598,
+      "step": 3180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 1.0072,
+      "step": 3181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591974768651576e-06,
+      "loss": 1.0141,
+      "step": 3182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58660905606981e-06,
+      "loss": 1.0402,
+      "step": 3183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581243462708007e-06,
+      "loss": 1.0576,
+      "step": 3184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575877990113566e-06,
+      "loss": 0.969,
+      "step": 3185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570512639833874e-06,
+      "loss": 1.0045,
+      "step": 3186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565147413416266e-06,
+      "loss": 1.1095,
+      "step": 3187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559782312408049e-06,
+      "loss": 1.0212,
+      "step": 3188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554417338356497e-06,
+      "loss": 1.0022,
+      "step": 3189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549052492808834e-06,
+      "loss": 0.9992,
+      "step": 3190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543687777312264e-06,
+      "loss": 0.9936,
+      "step": 3191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538323193413937e-06,
+      "loss": 1.0439,
+      "step": 3192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532958742660977e-06,
+      "loss": 1.0603,
+      "step": 3193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52759442660047e-06,
+      "loss": 0.9891,
+      "step": 3194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522230246779453e-06,
+      "loss": 1.0272,
+      "step": 3195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516866204744932e-06,
+      "loss": 1.0224,
+      "step": 3196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511502302043867e-06,
+      "loss": 0.98,
+      "step": 3197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50613854022319e-06,
+      "loss": 0.9778,
+      "step": 3198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500774920829779e-06,
+      "loss": 0.9714,
+      "step": 3199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495411445410474e-06,
+      "loss": 1.0117,
+      "step": 3200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490048115512075e-06,
+      "loss": 1.079,
+      "step": 3201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48468493268135e-06,
+      "loss": 1.0136,
+      "step": 3202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479321898465003e-06,
+      "loss": 0.9559,
+      "step": 3203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473959014409718e-06,
+      "loss": 1.062,
+      "step": 3204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468596282062114e-06,
+      "loss": 1.0709,
+      "step": 3205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463233702968784e-06,
+      "loss": 0.9629,
+      "step": 3206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457871278676273e-06,
+      "loss": 1.0016,
+      "step": 3207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45250901073107e-06,
+      "loss": 1.0438,
+      "step": 3208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44714690067963e-06,
+      "loss": 1.0423,
+      "step": 3209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441784950068362e-06,
+      "loss": 1.0243,
+      "step": 3210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436423160443625e-06,
+      "loss": 1.0446,
+      "step": 3211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43106153335174e-06,
+      "loss": 1.0516,
+      "step": 3212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425700070338961e-06,
+      "loss": 1.0319,
+      "step": 3213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420338772951521e-06,
+      "loss": 0.9612,
+      "step": 3214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414977642735583e-06,
+      "loss": 1.0778,
+      "step": 3215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409616681237275e-06,
+      "loss": 0.9938,
+      "step": 3216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404255890002677e-06,
+      "loss": 1.0413,
+      "step": 3217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398895270577808e-06,
+      "loss": 0.987,
+      "step": 3218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393534824508646e-06,
+      "loss": 1.0209,
+      "step": 3219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38817455334112e-06,
+      "loss": 1.0614,
+      "step": 3220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382814458621106e-06,
+      "loss": 0.9901,
+      "step": 3221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377454541894433e-06,
+      "loss": 1.007,
+      "step": 3222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.9923,
+      "step": 3223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36673524860414e-06,
+      "loss": 0.9913,
+      "step": 3224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361375875131912e-06,
+      "loss": 1.0503,
+      "step": 3225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356016685835807e-06,
+      "loss": 1.1132,
+      "step": 3226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35065768226139e-06,
+      "loss": 0.9758,
+      "step": 3227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345298865954167e-06,
+      "loss": 1.0302,
+      "step": 3228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3399402384596e-06,
+      "loss": 1.0225,
+      "step": 3229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334581801323085e-06,
+      "loss": 1.0068,
+      "step": 3230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329223556089976e-06,
+      "loss": 0.9641,
+      "step": 3231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323865504305566e-06,
+      "loss": 1.1019,
+      "step": 3232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318507647515083e-06,
+      "loss": 1.001,
+      "step": 3233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313149987263712e-06,
+      "loss": 1.0299,
+      "step": 3234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307792525096582e-06,
+      "loss": 1.0275,
+      "step": 3235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302435262558748e-06,
+      "loss": 1.0563,
+      "step": 3236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297078201195229e-06,
+      "loss": 1.0186,
+      "step": 3237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291721342550966e-06,
+      "loss": 1.0483,
+      "step": 3238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286364688170857e-06,
+      "loss": 0.9704,
+      "step": 3239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281008239599737e-06,
+      "loss": 1.0781,
+      "step": 3240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275651998382377e-06,
+      "loss": 1.0169,
+      "step": 3241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270295966063494e-06,
+      "loss": 1.016,
+      "step": 3242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264940144187735e-06,
+      "loss": 1.1249,
+      "step": 3243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2595845342997e-06,
+      "loss": 1.0641,
+      "step": 3244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.254229137943921e-06,
+      "loss": 1.0225,
+      "step": 3245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248873956664864e-06,
+      "loss": 1.0815,
+      "step": 3246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243518992006944e-06,
+      "loss": 0.9604,
+      "step": 3247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238164245514498e-06,
+      "loss": 1.0342,
+      "step": 3248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232809718731815e-06,
+      "loss": 0.9658,
+      "step": 3249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227455413203115e-06,
+      "loss": 1.0891,
+      "step": 3250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.222101330472553e-06,
+      "loss": 0.9754,
+      "step": 3251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216747472084221e-06,
+      "loss": 0.9852,
+      "step": 3252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211393839582143e-06,
+      "loss": 1.0424,
+      "step": 3253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206040434510283e-06,
+      "loss": 1.0267,
+      "step": 3254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200687258412541e-06,
+      "loss": 0.9846,
+      "step": 3255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195334312832742e-06,
+      "loss": 1.007,
+      "step": 3256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189981599314655e-06,
+      "loss": 1.0555,
+      "step": 3257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184629119401968e-06,
+      "loss": 0.9802,
+      "step": 3258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179276874638315e-06,
+      "loss": 1.0281,
+      "step": 3259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173924866567263e-06,
+      "loss": 1.0618,
+      "step": 3260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168573096732298e-06,
+      "loss": 0.9685,
+      "step": 3261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163221566676847e-06,
+      "loss": 1.0466,
+      "step": 3262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157870277944265e-06,
+      "loss": 1.0032,
+      "step": 3263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152519232077838e-06,
+      "loss": 1.0168,
+      "step": 3264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147168430620788e-06,
+      "loss": 1.0193,
+      "step": 3265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14181787511625e-06,
+      "loss": 1.003,
+      "step": 3266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136467567107306e-06,
+      "loss": 1.0034,
+      "step": 3267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131117508136953e-06,
+      "loss": 1.019,
+      "step": 3268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125767699748127e-06,
+      "loss": 1.0312,
+      "step": 3269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120418143483689e-06,
+      "loss": 1.0035,
+      "step": 3270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115068840886418e-06,
+      "loss": 1.0693,
+      "step": 3271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10971979349903e-06,
+      "loss": 1.0727,
+      "step": 3272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104371002864167e-06,
+      "loss": 0.9441,
+      "step": 3273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099022470524392e-06,
+      "loss": 1.0478,
+      "step": 3274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093674198022201e-06,
+      "loss": 0.9531,
+      "step": 3275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0883261869e-06,
+      "loss": 0.9952,
+      "step": 3276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082978438700138e-06,
+      "loss": 1.0763,
+      "step": 3277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07763095496488e-06,
+      "loss": 1.0077,
+      "step": 3278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07228373723641e-06,
+      "loss": 1.0075,
+      "step": 3279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066936787056843e-06,
+      "loss": 0.9464,
+      "step": 3280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061590105968208e-06,
+      "loss": 1.0626,
+      "step": 3281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056243695512469e-06,
+      "loss": 0.9816,
+      "step": 3282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050897557231504e-06,
+      "loss": 1.0298,
+      "step": 3283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045551692667108e-06,
+      "loss": 0.9992,
+      "step": 3284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040206103361013e-06,
+      "loss": 1.0813,
+      "step": 3285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034860790854848e-06,
+      "loss": 1.0325,
+      "step": 3286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029515756690187e-06,
+      "loss": 1.0374,
+      "step": 3287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024171002408507e-06,
+      "loss": 1.0571,
+      "step": 3288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018826529551207e-06,
+      "loss": 0.9833,
+      "step": 3289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013482339659616e-06,
+      "loss": 1.0468,
+      "step": 3290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00813843427496e-06,
+      "loss": 0.9991,
+      "step": 3291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002794814938403e-06,
+      "loss": 1.0451,
+      "step": 3292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99745148319102e-06,
+      "loss": 1.0411,
+      "step": 3293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9921084405738e-06,
+      "loss": 0.9335,
+      "step": 3294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986765688627652e-06,
+      "loss": 1.0638,
+      "step": 3295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981423228893396e-06,
+      "loss": 1.0448,
+      "step": 3296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976081062911776e-06,
+      "loss": 0.9539,
+      "step": 3297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970739192223448e-06,
+      "loss": 1.1027,
+      "step": 3298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965397618368977e-06,
+      "loss": 0.9915,
+      "step": 3299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960056342888855e-06,
+      "loss": 0.9979,
+      "step": 3300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 1.0664,
+      "step": 3301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949374693213134e-06,
+      "loss": 1.1032,
+      "step": 3302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944034322098082e-06,
+      "loss": 1.0445,
+      "step": 3303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938694255518444e-06,
+      "loss": 1.0024,
+      "step": 3304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933354495014271e-06,
+      "loss": 1.0002,
+      "step": 3305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928015042125523e-06,
+      "loss": 0.9941,
+      "step": 3306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922675898392072e-06,
+      "loss": 1.0041,
+      "step": 3307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917337065353708e-06,
+      "loss": 0.9939,
+      "step": 3308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911998544550116e-06,
+      "loss": 1.0791,
+      "step": 3309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906660337520903e-06,
+      "loss": 0.9344,
+      "step": 3310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901322445805586e-06,
+      "loss": 1.0425,
+      "step": 3311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89598487094358e-06,
+      "loss": 1.093,
+      "step": 3312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890647614474223e-06,
+      "loss": 0.9781,
+      "step": 3313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885310677936746e-06,
+      "loss": 1.0147,
+      "step": 3314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879974062870302e-06,
+      "loss": 1.0107,
+      "step": 3315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874637770813947e-06,
+      "loss": 1.0665,
+      "step": 3316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869301803306634e-06,
+      "loss": 1.0026,
+      "step": 3317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86396616188724e-06,
+      "loss": 0.97,
+      "step": 3318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858630848094527e-06,
+      "loss": 1.1065,
+      "step": 3319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853295863467182e-06,
+      "loss": 1.013,
+      "step": 3320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84796120954379e-06,
+      "loss": 1.0007,
+      "step": 3321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842626887862832e-06,
+      "loss": 1.0329,
+      "step": 3322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83729289996271e-06,
+      "loss": 1.0291,
+      "step": 3323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831959247381712e-06,
+      "loss": 1.0034,
+      "step": 3324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82662593165804e-06,
+      "loss": 1.007,
+      "step": 3325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821292954329798e-06,
+      "loss": 1.0268,
+      "step": 3326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815960316934991e-06,
+      "loss": 1.0279,
+      "step": 3327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810628021011528e-06,
+      "loss": 0.8767,
+      "step": 3328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805296068097211e-06,
+      "loss": 1.0642,
+      "step": 3329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799964459729755e-06,
+      "loss": 1.0355,
+      "step": 3330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.9957,
+      "step": 3331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789302282785764e-06,
+      "loss": 0.9802,
+      "step": 3332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783971717284152e-06,
+      "loss": 1.0359,
+      "step": 3333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778641502479234e-06,
+      "loss": 1.025,
+      "step": 3334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773311639908226e-06,
+      "loss": 1.0123,
+      "step": 3335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767982131108232e-06,
+      "loss": 1.0666,
+      "step": 3336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762652977616258e-06,
+      "loss": 0.9918,
+      "step": 3337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757324180969208e-06,
+      "loss": 0.9449,
+      "step": 3338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751995742703874e-06,
+      "loss": 0.9203,
+      "step": 3339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746667664356957e-06,
+      "loss": 1.1225,
+      "step": 3340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741339947465055e-06,
+      "loss": 1.0013,
+      "step": 3341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736012593564642e-06,
+      "loss": 0.9914,
+      "step": 3342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730685604192115e-06,
+      "loss": 0.9982,
+      "step": 3343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725358980883743e-06,
+      "loss": 0.9954,
+      "step": 3344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7200327251757e-06,
+      "loss": 1.0458,
+      "step": 3345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714706838604056e-06,
+      "loss": 1.0035,
+      "step": 3346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709381322704768e-06,
+      "loss": 1.0668,
+      "step": 3347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70405617901369e-06,
+      "loss": 1.0219,
+      "step": 3348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69873140906657e-06,
+      "loss": 0.955,
+      "step": 3349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69340701439904e-06,
+      "loss": 1.0359,
+      "step": 3350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688082996546639e-06,
+      "loss": 1.0175,
+      "step": 3351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68275935704478e-06,
+      "loss": 1.0435,
+      "step": 3352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.677436097428775e-06,
+      "loss": 1.0879,
+      "step": 3353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672113219233835e-06,
+      "loss": 1.0476,
+      "step": 3354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666790723995043e-06,
+      "loss": 0.9843,
+      "step": 3355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661468613247387e-06,
+      "loss": 0.9587,
+      "step": 3356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656146888525735e-06,
+      "loss": 1.0661,
+      "step": 3357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650825551364844e-06,
+      "loss": 0.9242,
+      "step": 3358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645504603299369e-06,
+      "loss": 1.0547,
+      "step": 3359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64018404586384e-06,
+      "loss": 1.0252,
+      "step": 3360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634863880592687e-06,
+      "loss": 0.9866,
+      "step": 3361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629544109020211e-06,
+      "loss": 1.0077,
+      "step": 3362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624224732680613e-06,
+      "loss": 1.0669,
+      "step": 3363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61890575310798e-06,
+      "loss": 1.0876,
+      "step": 3364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613587171836272e-06,
+      "loss": 0.9508,
+      "step": 3365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60826899039935e-06,
+      "loss": 0.9771,
+      "step": 3366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602951210330942e-06,
+      "loss": 1.0153,
+      "step": 3367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597633833164677e-06,
+      "loss": 1.0806,
+      "step": 3368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592316860434062e-06,
+      "loss": 1.0446,
+      "step": 3369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587000293672482e-06,
+      "loss": 0.9941,
+      "step": 3370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581684134413216e-06,
+      "loss": 1.0471,
+      "step": 3371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57636838418941e-06,
+      "loss": 0.9968,
+      "step": 3372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571053044534104e-06,
+      "loss": 0.986,
+      "step": 3373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565738116980223e-06,
+      "loss": 1.0594,
+      "step": 3374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560423603060555e-06,
+      "loss": 1.0355,
+      "step": 3375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55510950430779e-06,
+      "loss": 1.0406,
+      "step": 3376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549795822254483e-06,
+      "loss": 1.0233,
+      "step": 3377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544482558433072e-06,
+      "loss": 1.0563,
+      "step": 3378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 1.0859,
+      "step": 3379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533857291615114e-06,
+      "loss": 0.9985,
+      "step": 3380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528545291682839e-06,
+      "loss": 1.0354,
+      "step": 3381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523233716111017e-06,
+      "loss": 1.0104,
+      "step": 3382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517922566431473e-06,
+      "loss": 0.9744,
+      "step": 3383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51261184417593e-06,
+      "loss": 1.0169,
+      "step": 3384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50730155087596e-06,
+      "loss": 0.9838,
+      "step": 3385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501991688063033e-06,
+      "loss": 1.013,
+      "step": 3386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496682257268494e-06,
+      "loss": 1.0113,
+      "step": 3387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491373260023546e-06,
+      "loss": 0.98,
+      "step": 3388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486064697859285e-06,
+      "loss": 1.025,
+      "step": 3389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480756572306674e-06,
+      "loss": 1.0054,
+      "step": 3390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475448884896546e-06,
+      "loss": 1.0669,
+      "step": 3391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47014163715962e-06,
+      "loss": 1.0252,
+      "step": 3392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464834830626477e-06,
+      "loss": 0.983,
+      "step": 3393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459528466827576e-06,
+      "loss": 0.9564,
+      "step": 3394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45422254729324e-06,
+      "loss": 1.0713,
+      "step": 3395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44891707355368e-06,
+      "loss": 1.0022,
+      "step": 3396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443612047138965e-06,
+      "loss": 1.0195,
+      "step": 3397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438307469579038e-06,
+      "loss": 0.9836,
+      "step": 3398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433003342403715e-06,
+      "loss": 1.0216,
+      "step": 3399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427699667142681e-06,
+      "loss": 1.0267,
+      "step": 3400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422396445325488e-06,
+      "loss": 1.0085,
+      "step": 3401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417093678481563e-06,
+      "loss": 1.0636,
+      "step": 3402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411791368140197e-06,
+      "loss": 1.004,
+      "step": 3403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406489515830552e-06,
+      "loss": 0.9404,
+      "step": 3404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401188123081653e-06,
+      "loss": 1.0267,
+      "step": 3405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395887191422397e-06,
+      "loss": 1.0844,
+      "step": 3406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390586722381553e-06,
+      "loss": 1.0201,
+      "step": 3407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385286717487744e-06,
+      "loss": 1.0327,
+      "step": 3408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37998717826947e-06,
+      "loss": 1.0898,
+      "step": 3409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37468810625509e-06,
+      "loss": 0.9354,
+      "step": 3410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36938950297283e-06,
+      "loss": 1.0148,
+      "step": 3411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364091369950783e-06,
+      "loss": 0.9861,
+      "step": 3412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358793708716905e-06,
+      "loss": 1.0489,
+      "step": 3413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353496520799022e-06,
+      "loss": 1.0264,
+      "step": 3414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348199807724806e-06,
+      "loss": 1.0027,
+      "step": 3415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342903571021809e-06,
+      "loss": 1.0009,
+      "step": 3416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337607812217446e-06,
+      "loss": 0.9861,
+      "step": 3417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332312532838978e-06,
+      "loss": 0.9867,
+      "step": 3418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327017734413544e-06,
+      "loss": 1.0512,
+      "step": 3419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32172341846814e-06,
+      "loss": 1.0903,
+      "step": 3420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316429586529616e-06,
+      "loss": 0.974,
+      "step": 3421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31113624012469e-06,
+      "loss": 0.9899,
+      "step": 3422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305843380779938e-06,
+      "loss": 1.0421,
+      "step": 3423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300551010021796e-06,
+      "loss": 0.9893,
+      "step": 3424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295259129376557e-06,
+      "loss": 1.0232,
+      "step": 3425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28996774037037e-06,
+      "loss": 1.0187,
+      "step": 3426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284676844529258e-06,
+      "loss": 1.0835,
+      "step": 3427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279386443379077e-06,
+      "loss": 1.0134,
+      "step": 3428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274096538445558e-06,
+      "loss": 1.0253,
+      "step": 3429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268807131254288e-06,
+      "loss": 1.0301,
+      "step": 3430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 1.0746,
+      "step": 3431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25822981620009e-06,
+      "loss": 0.9805,
+      "step": 3432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252941911387613e-06,
+      "loss": 0.9811,
+      "step": 3433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247654510418276e-06,
+      "loss": 1.0713,
+      "step": 3434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242367614816938e-06,
+      "loss": 1.0367,
+      "step": 3435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23708122610831e-06,
+      "loss": 1.0257,
+      "step": 3436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231795345816972e-06,
+      "loss": 0.9999,
+      "step": 3437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226509975467335e-06,
+      "loss": 1.0043,
+      "step": 3438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221225116583677e-06,
+      "loss": 1.014,
+      "step": 3439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215940770690133e-06,
+      "loss": 1.1019,
+      "step": 3440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210656939310672e-06,
+      "loss": 0.9497,
+      "step": 3441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205373623969137e-06,
+      "loss": 1.0084,
+      "step": 3442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200090826189202e-06,
+      "loss": 0.9606,
+      "step": 3443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194808547494401e-06,
+      "loss": 1.1057,
+      "step": 3444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189526789408123e-06,
+      "loss": 1.0218,
+      "step": 3445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184245553453596e-06,
+      "loss": 0.9879,
+      "step": 3446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17896484115391e-06,
+      "loss": 1.0216,
+      "step": 3447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17368465403199e-06,
+      "loss": 0.9685,
+      "step": 3448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168404993610617e-06,
+      "loss": 1.0082,
+      "step": 3449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163125861412426e-06,
+      "loss": 1.013,
+      "step": 3450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157847258959885e-06,
+      "loss": 1.0865,
+      "step": 3451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152569187775325e-06,
+      "loss": 0.9775,
+      "step": 3452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147291649380912e-06,
+      "loss": 1.0251,
+      "step": 3453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142014645298662e-06,
+      "loss": 1.02,
+      "step": 3454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13673817705044e-06,
+      "loss": 1.0136,
+      "step": 3455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131462246157953e-06,
+      "loss": 1.0069,
+      "step": 3456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 1.0394,
+      "step": 3457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120912002526242e-06,
+      "loss": 1.0114,
+      "step": 3458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115637692829656e-06,
+      "loss": 1.0305,
+      "step": 3459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.110363926574088e-06,
+      "loss": 1.0048,
+      "step": 3460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105090705280456e-06,
+      "loss": 1.0882,
+      "step": 3461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099818030469538e-06,
+      "loss": 0.9377,
+      "step": 3462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094545903661952e-06,
+      "loss": 1.0348,
+      "step": 3463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089274326378145e-06,
+      "loss": 1.0104,
+      "step": 3464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084003300138423e-06,
+      "loss": 1.0126,
+      "step": 3465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078732826462917e-06,
+      "loss": 0.9616,
+      "step": 3466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07346290687161e-06,
+      "loss": 1.0338,
+      "step": 3467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068193542884325e-06,
+      "loss": 1.0322,
+      "step": 3468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062924736020714e-06,
+      "loss": 1.0279,
+      "step": 3469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057656487800283e-06,
+      "loss": 1.0095,
+      "step": 3470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05238879974236e-06,
+      "loss": 1.0001,
+      "step": 3471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04712167336613e-06,
+      "loss": 1.0563,
+      "step": 3472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041855110190604e-06,
+      "loss": 0.9881,
+      "step": 3473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03658911173463e-06,
+      "loss": 0.9573,
+      "step": 3474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0313236795169e-06,
+      "loss": 1.0553,
+      "step": 3475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026058815055937e-06,
+      "loss": 0.9713,
+      "step": 3476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020794519870107e-06,
+      "loss": 0.9912,
+      "step": 3477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015530795477603e-06,
+      "loss": 1.0456,
+      "step": 3478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010267643396457e-06,
+      "loss": 1.0601,
+      "step": 3479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005005065144544e-06,
+      "loss": 0.9982,
+      "step": 3480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 0.9819,
+      "step": 3481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994481636199036e-06,
+      "loss": 1.0231,
+      "step": 3482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989220788540356e-06,
+      "loss": 1.0242,
+      "step": 3483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983960520780712e-06,
+      "loss": 1.0108,
+      "step": 3484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978700834437142e-06,
+      "loss": 0.9354,
+      "step": 3485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973441731026518e-06,
+      "loss": 1.0554,
+      "step": 3486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968183212065537e-06,
+      "loss": 0.9756,
+      "step": 3487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962925279070732e-06,
+      "loss": 1.0273,
+      "step": 3488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957667933558462e-06,
+      "loss": 1.0564,
+      "step": 3489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952411177044923e-06,
+      "loss": 1.0288,
+      "step": 3490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947155011046144e-06,
+      "loss": 0.9685,
+      "step": 3491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941899437077966e-06,
+      "loss": 0.9604,
+      "step": 3492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936644456656082e-06,
+      "loss": 1.0693,
+      "step": 3493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931390071295994e-06,
+      "loss": 1.0411,
+      "step": 3494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926136282513046e-06,
+      "loss": 1.0077,
+      "step": 3495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 1.0018,
+      "step": 3496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915630500739071e-06,
+      "loss": 0.9898,
+      "step": 3497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.910378510777858e-06,
+      "loss": 1.0113,
+      "step": 3498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905127123453416e-06,
+      "loss": 1.0368,
+      "step": 3499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899876340280221e-06,
+      "loss": 0.9544,
+      "step": 3500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894626162772578e-06,
+      "loss": 1.0191,
+      "step": 3501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889376592444605e-06,
+      "loss": 1.0158,
+      "step": 3502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88412763081026e-06,
+      "loss": 1.0385,
+      "step": 3503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878879279383314e-06,
+      "loss": 1.0443,
+      "step": 3504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873631539677364e-06,
+      "loss": 0.9422,
+      "step": 3505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868384413205842e-06,
+      "loss": 1.0479,
+      "step": 3506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863137901481982e-06,
+      "loss": 1.0143,
+      "step": 3507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857892006018863e-06,
+      "loss": 0.9886,
+      "step": 3508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852646728329368e-06,
+      "loss": 1.003,
+      "step": 3509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847402069926215e-06,
+      "loss": 1.0812,
+      "step": 3510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84215803232194e-06,
+      "loss": 1.053,
+      "step": 3511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836914617028892e-06,
+      "loss": 1.0087,
+      "step": 3512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831671825559253e-06,
+      "loss": 1.1018,
+      "step": 3513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826429659425012e-06,
+      "loss": 1.0178,
+      "step": 3514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821188120137987e-06,
+      "loss": 0.8731,
+      "step": 3515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815947209209817e-06,
+      "loss": 1.0143,
+      "step": 3516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81070692815195e-06,
+      "loss": 1.0595,
+      "step": 3517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805467278475663e-06,
+      "loss": 1.0032,
+      "step": 3518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800228261692037e-06,
+      "loss": 0.9656,
+      "step": 3519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794989879311991e-06,
+      "loss": 1.0115,
+      "step": 3520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78975213284624e-06,
+      "loss": 0.9889,
+      "step": 3521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784515023805328e-06,
+      "loss": 1.0283,
+      "step": 3522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779278553699615e-06,
+      "loss": 1.0288,
+      "step": 3523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774042724039268e-06,
+      "loss": 1.108,
+      "step": 3524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768807536334276e-06,
+      "loss": 0.9668,
+      "step": 3525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763572992094448e-06,
+      "loss": 1.0182,
+      "step": 3526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758339092829395e-06,
+      "loss": 1.0718,
+      "step": 3527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753105840048548e-06,
+      "loss": 0.9987,
+      "step": 3528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747873235261157e-06,
+      "loss": 0.9425,
+      "step": 3529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742641279976278e-06,
+      "loss": 1.0136,
+      "step": 3530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73740997570278e-06,
+      "loss": 1.0705,
+      "step": 3531
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732179323949348e-06,
+      "loss": 0.9773,
+      "step": 3532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726949326224471e-06,
+      "loss": 1.0196,
+      "step": 3533
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721719984036469e-06,
+      "loss": 1.0921,
+      "step": 3534
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 0.9164,
+      "step": 3535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711263272303333e-06,
+      "loss": 1.0442,
+      "step": 3536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706035905773865e-06,
+      "loss": 1.0121,
+      "step": 3537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700809200812596e-06,
+      "loss": 1.0176,
+      "step": 3538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69558315892688e-06,
+      "loss": 1.049,
+      "step": 3539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69035778162388e-06,
+      "loss": 0.9625,
+      "step": 3540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685133070410571e-06,
+      "loss": 1.042,
+      "step": 3541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679909026793735e-06,
+      "loss": 0.979,
+      "step": 3542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67468565227996e-06,
+      "loss": 0.9086,
+      "step": 3543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669462948375646e-06,
+      "loss": 0.9865,
+      "step": 3544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66424091658699e-06,
+      "loss": 1.0323,
+      "step": 3545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659019558420004e-06,
+      "loss": 0.9899,
+      "step": 3546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6537988753805e-06,
+      "loss": 0.9564,
+      "step": 3547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6485788689741e-06,
+      "loss": 1.072,
+      "step": 3548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643359540706233e-06,
+      "loss": 0.9995,
+      "step": 3549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638140892082118e-06,
+      "loss": 1.0065,
+      "step": 3550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632922924606794e-06,
+      "loss": 1.0407,
+      "step": 3551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6277056397850975e-06,
+      "loss": 0.9905,
+      "step": 3552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622489039121668e-06,
+      "loss": 1.0135,
+      "step": 3553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6172731241209515e-06,
+      "loss": 1.0158,
+      "step": 3554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612057896287186e-06,
+      "loss": 1.0562,
+      "step": 3555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.9341,
+      "step": 3556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60162950813651e-06,
+      "loss": 0.963,
+      "step": 3557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596416350827091e-06,
+      "loss": 1.0512,
+      "step": 3558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591203886699625e-06,
+      "loss": 1.0003,
+      "step": 3559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585992117257351e-06,
+      "loss": 0.9733,
+      "step": 3560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 1.0251,
+      "step": 3561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575570668440392e-06,
+      "loss": 1.0415,
+      "step": 3562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570360992071199e-06,
+      "loss": 0.9411,
+      "step": 3563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5651520163982e-06,
+      "loss": 1.0109,
+      "step": 3564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559943742923626e-06,
+      "loss": 1.0271,
+      "step": 3565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554736173149524e-06,
+      "loss": 1.0112,
+      "step": 3566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549529308577737e-06,
+      "loss": 0.9691,
+      "step": 3567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544323150709889e-06,
+      "loss": 0.9404,
+      "step": 3568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53911770104742e-06,
+      "loss": 1.0738,
+      "step": 3569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533912961091551e-06,
+      "loss": 1.0318,
+      "step": 3570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5287089323433035e-06,
+      "loss": 0.9739,
+      "step": 3571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5235056163035005e-06,
+      "loss": 1.0074,
+      "step": 3572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.518303014472748e-06,
+      "loss": 1.0192,
+      "step": 3573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513101128351454e-06,
+      "loss": 0.9973,
+      "step": 3574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5078999594398135e-06,
+      "loss": 0.9974,
+      "step": 3575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502699509237822e-06,
+      "loss": 1.0595,
+      "step": 3576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497499779245268e-06,
+      "loss": 0.9996,
+      "step": 3577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492300770961718e-06,
+      "loss": 0.9874,
+      "step": 3578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487102485886552e-06,
+      "loss": 1.0465,
+      "step": 3579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4819049255189215e-06,
+      "loss": 1.0726,
+      "step": 3580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476708091357783e-06,
+      "loss": 0.9885,
+      "step": 3581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471511984901878e-06,
+      "loss": 0.9944,
+      "step": 3582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4663166076497376e-06,
+      "loss": 1.0338,
+      "step": 3583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461121961099685e-06,
+      "loss": 0.9824,
+      "step": 3584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455928046749828e-06,
+      "loss": 1.0006,
+      "step": 3585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450734866098066e-06,
+      "loss": 1.0402,
+      "step": 3586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445542420642097e-06,
+      "loss": 1.0115,
+      "step": 3587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440350711879385e-06,
+      "loss": 1.0414,
+      "step": 3588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435159741307203e-06,
+      "loss": 0.9953,
+      "step": 3589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429969510422592e-06,
+      "loss": 1.0886,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4247800207223975e-06,
+      "loss": 0.9253,
+      "step": 3591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419591273703245e-06,
+      "loss": 1.0437,
+      "step": 3592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414403270861536e-06,
+      "loss": 0.972,
+      "step": 3593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409216013693472e-06,
+      "loss": 1.0739,
+      "step": 3594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404029503695028e-06,
+      "loss": 0.9773,
+      "step": 3595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398843742361972e-06,
+      "loss": 0.9718,
+      "step": 3596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393658731189857e-06,
+      "loss": 1.0551,
+      "step": 3597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388474471674005e-06,
+      "loss": 1.051,
+      "step": 3598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383290965309541e-06,
+      "loss": 0.9474,
+      "step": 3599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378108213591355e-06,
+      "loss": 1.0224,
+      "step": 3600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372926218014131e-06,
+      "loss": 1.0206,
+      "step": 3601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367744980072337e-06,
+      "loss": 0.9716,
+      "step": 3602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362564501260208e-06,
+      "loss": 1.0319,
+      "step": 3603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357384783071773e-06,
+      "loss": 1.0443,
+      "step": 3604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35220582700084e-06,
+      "loss": 1.0477,
+      "step": 3605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347027634540993e-06,
+      "loss": 0.982,
+      "step": 3606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3418502071856004e-06,
+      "loss": 0.9982,
+      "step": 3607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336673546427801e-06,
+      "loss": 0.9935,
+      "step": 3608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331497653760522e-06,
+      "loss": 1.0778,
+      "step": 3609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326322530676471e-06,
+      "loss": 1.0295,
+      "step": 3610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32114817866812e-06,
+      "loss": 1.0334,
+      "step": 3611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315974599227736e-06,
+      "loss": 0.9645,
+      "step": 3612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 1.0427,
+      "step": 3613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305629764018764e-06,
+      "loss": 1.0699,
+      "step": 3614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3004585112335835e-06,
+      "loss": 0.9807,
+      "step": 3615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2952880369831635e-06,
+      "loss": 1.006,
+      "step": 3616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29011834275865e-06,
+      "loss": 1.0449,
+      "step": 3617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28494943005095e-06,
+      "loss": 0.9895,
+      "step": 3618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279781300350758e-06,
+      "loss": 0.9734,
+      "step": 3619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27461395514854e-06,
+      "loss": 0.9973,
+      "step": 3620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269447395934526e-06,
+      "loss": 1.0349,
+      "step": 3621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264281624198735e-06,
+      "loss": 1.0025,
+      "step": 3622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2591166414309435e-06,
+      "loss": 1.0096,
+      "step": 3623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253952449120712e-06,
+      "loss": 1.0188,
+      "step": 3624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248789048757368e-06,
+      "loss": 1.0729,
+      "step": 3625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243626441830009e-06,
+      "loss": 0.9196,
+      "step": 3626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238464629827511e-06,
+      "loss": 1.0131,
+      "step": 3627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23330361423851e-06,
+      "loss": 1.0244,
+      "step": 3628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228143396551419e-06,
+      "loss": 1.0017,
+      "step": 3629
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222983978254426e-06,
+      "loss": 1.0037,
+      "step": 3630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 1.0715,
+      "step": 3631
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212667545782292e-06,
+      "loss": 1.0133,
+      "step": 3632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20751053458236e-06,
+      "loss": 0.9612,
+      "step": 3633
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2023543287229394e-06,
+      "loss": 1.0067,
+      "step": 3634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19719892969106e-06,
+      "loss": 1.0686,
+      "step": 3635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192044338973504e-06,
+      "loss": 0.9854,
+      "step": 3636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186890558056836e-06,
+      "loss": 1.0514,
+      "step": 3637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181737588427385e-06,
+      "loss": 1.0226,
+      "step": 3638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176585431571235e-06,
+      "loss": 0.9888,
+      "step": 3639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171434088974252e-06,
+      "loss": 1.0074,
+      "step": 3640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16628356212205e-06,
+      "loss": 1.0134,
+      "step": 3641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16113385250002e-06,
+      "loss": 0.9932,
+      "step": 3642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155984961593317e-06,
+      "loss": 1.0215,
+      "step": 3643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150836890886848e-06,
+      "loss": 0.9935,
+      "step": 3644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145689641865301e-06,
+      "loss": 1.0424,
+      "step": 3645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140543216013109e-06,
+      "loss": 1.0341,
+      "step": 3646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135397614814481e-06,
+      "loss": 0.9705,
+      "step": 3647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130252839753384e-06,
+      "loss": 0.9528,
+      "step": 3648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.125108892313546e-06,
+      "loss": 1.039,
+      "step": 3649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119965773978459e-06,
+      "loss": 0.9943,
+      "step": 3650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114823486231366e-06,
+      "loss": 1.0127,
+      "step": 3651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109682030555283e-06,
+      "loss": 1.0253,
+      "step": 3652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104541408432985e-06,
+      "loss": 1.0132,
+      "step": 3653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0994016213469955e-06,
+      "loss": 0.9655,
+      "step": 3654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094262670779611e-06,
+      "loss": 1.031,
+      "step": 3655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089124558212872e-06,
+      "loss": 1.0109,
+      "step": 3656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08398728512859e-06,
+      "loss": 0.9537,
+      "step": 3657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078850853008332e-06,
+      "loss": 0.9906,
+      "step": 3658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0737152633334164e-06,
+      "loss": 1.0745,
+      "step": 3659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068580517584928e-06,
+      "loss": 0.9946,
+      "step": 3660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063446617243695e-06,
+      "loss": 1.0334,
+      "step": 3661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058313563790311e-06,
+      "loss": 1.0616,
+      "step": 3662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053181358705132e-06,
+      "loss": 1.0137,
+      "step": 3663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048050003468252e-06,
+      "loss": 1.0032,
+      "step": 3664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042919499559538e-06,
+      "loss": 1.0255,
+      "step": 3665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03778984845859e-06,
+      "loss": 1.032,
+      "step": 3666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0326610516447825e-06,
+      "loss": 1.0192,
+      "step": 3667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027533110597239e-06,
+      "loss": 1.0191,
+      "step": 3668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022406026794828e-06,
+      "loss": 1.0356,
+      "step": 3669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017279801716177e-06,
+      "loss": 1.0536,
+      "step": 3670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012154436839664e-06,
+      "loss": 0.9842,
+      "step": 3671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007029933643419e-06,
+      "loss": 0.9862,
+      "step": 3672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001906293605329e-06,
+      "loss": 1.0217,
+      "step": 3673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996783518203019e-06,
+      "loss": 0.9998,
+      "step": 3674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991661608913877e-06,
+      "loss": 1.0303,
+      "step": 3675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986540567215043e-06,
+      "loss": 1.0134,
+      "step": 3676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9814203945833896e-06,
+      "loss": 1.0023,
+      "step": 3677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976301092495556e-06,
+      "loss": 1.0158,
+      "step": 3678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971182662427925e-06,
+      "loss": 1.0695,
+      "step": 3679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966065105856623e-06,
+      "loss": 1.0507,
+      "step": 3680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960948424257532e-06,
+      "loss": 1.0137,
+      "step": 3681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9558326191062775e-06,
+      "loss": 1.0069,
+      "step": 3682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950717691878236e-06,
+      "loss": 1.0572,
+      "step": 3683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9456036440485216e-06,
+      "loss": 1.028,
+      "step": 3684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940490477092004e-06,
+      "loss": 0.9804,
+      "step": 3685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9353781924832994e-06,
+      "loss": 0.9867,
+      "step": 3686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930266791696761e-06,
+      "loss": 0.9751,
+      "step": 3687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925156276206497e-06,
+      "loss": 0.9952,
+      "step": 3688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9200466474863495e-06,
+      "loss": 1.0113,
+      "step": 3689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914937907009913e-06,
+      "loss": 1.0407,
+      "step": 3690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.9365,
+      "step": 3691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904723096681267e-06,
+      "loss": 1.0369,
+      "step": 3692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899617029774961e-06,
+      "loss": 0.9588,
+      "step": 3693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894511857004166e-06,
+      "loss": 1.0816,
+      "step": 3694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8894075798411945e-06,
+      "loss": 0.9903,
+      "step": 3695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884304199758096e-06,
+      "loss": 0.9601,
+      "step": 3696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879201718226658e-06,
+      "loss": 1.0163,
+      "step": 3697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874100136718416e-06,
+      "loss": 1.0271,
+      "step": 3698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8689994567046335e-06,
+      "loss": 0.9888,
+      "step": 3699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8638996796563275e-06,
+      "loss": 1.0312,
+      "step": 3700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85880080704425e-06,
+      "loss": 0.9401,
+      "step": 3701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85370284033889e-06,
+      "loss": 1.0269,
+      "step": 3702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8486057810104774e-06,
+      "loss": 0.9121,
+      "step": 3703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843509630528977e-06,
+      "loss": 1.069,
+      "step": 3704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838414390364094e-06,
+      "loss": 0.8885,
+      "step": 3705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833320061985278e-06,
+      "loss": 1.0293,
+      "step": 3706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828226646861697e-06,
+      "loss": 1.0184,
+      "step": 3707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82313414646228e-06,
+      "loss": 1.0377,
+      "step": 3708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.81804256225567e-06,
+      "loss": 0.9661,
+      "step": 3709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812951895710258e-06,
+      "loss": 0.9711,
+      "step": 3710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807862148294171e-06,
+      "loss": 1.0742,
+      "step": 3711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802773321475262e-06,
+      "loss": 1.0084,
+      "step": 3712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797685416721131e-06,
+      "loss": 1.0378,
+      "step": 3713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7925984354991e-06,
+      "loss": 1.0692,
+      "step": 3714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78751237927623e-06,
+      "loss": 0.9301,
+      "step": 3715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782427249519321e-06,
+      "loss": 0.9596,
+      "step": 3716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777343047694891e-06,
+      "loss": 1.008,
+      "step": 3717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772259775269203e-06,
+      "loss": 1.0612,
+      "step": 3718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767177433708254e-06,
+      "loss": 0.962,
+      "step": 3719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762096024477758e-06,
+      "loss": 1.036,
+      "step": 3720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757015549043174e-06,
+      "loss": 0.9621,
+      "step": 3721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751936008869686e-06,
+      "loss": 1.0478,
+      "step": 3722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746857405422207e-06,
+      "loss": 1.0058,
+      "step": 3723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741779740165385e-06,
+      "loss": 0.9458,
+      "step": 3724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736703014563591e-06,
+      "loss": 1.0409,
+      "step": 3725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731627230080931e-06,
+      "loss": 1.0158,
+      "step": 3726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7265523881812335e-06,
+      "loss": 0.9717,
+      "step": 3727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7214784903280606e-06,
+      "loss": 1.05,
+      "step": 3728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716405537984703e-06,
+      "loss": 0.979,
+      "step": 3729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711333532614168e-06,
+      "loss": 1.0448,
+      "step": 3730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706262475679205e-06,
+      "loss": 1.0112,
+      "step": 3731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701192368642279e-06,
+      "loss": 1.0136,
+      "step": 3732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696123212965583e-06,
+      "loss": 0.9604,
+      "step": 3733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69105501011104e-06,
+      "loss": 1.0098,
+      "step": 3734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6859877615402935e-06,
+      "loss": 0.981,
+      "step": 3735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680921468714718e-06,
+      "loss": 1.0346,
+      "step": 3736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6758561330953995e-06,
+      "loss": 1.0079,
+      "step": 3737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670791756143162e-06,
+      "loss": 0.9465,
+      "step": 3738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665728339318551e-06,
+      "loss": 1.0249,
+      "step": 3739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6606658840818225e-06,
+      "loss": 1.0209,
+      "step": 3740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655604391892972e-06,
+      "loss": 1.0247,
+      "step": 3741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650543864211702e-06,
+      "loss": 1.0463,
+      "step": 3742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645484302497452e-06,
+      "loss": 0.991,
+      "step": 3743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64042570820937e-06,
+      "loss": 0.9556,
+      "step": 3744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6353680828063306e-06,
+      "loss": 0.9928,
+      "step": 3745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630311427746931e-06,
+      "loss": 1.0614,
+      "step": 3746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62525574448949e-06,
+      "loss": 1.0126,
+      "step": 3747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620201034492032e-06,
+      "loss": 0.9383,
+      "step": 3748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615147299212321e-06,
+      "loss": 1.0097,
+      "step": 3749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610094540107821e-06,
+      "loss": 1.015,
+      "step": 3750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605042758635729e-06,
+      "loss": 1.0204,
+      "step": 3751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599991956252956e-06,
+      "loss": 1.0189,
+      "step": 3752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594942134416123e-06,
+      "loss": 1.0363,
+      "step": 3753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589893294581579e-06,
+      "loss": 0.9632,
+      "step": 3754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5848454382053825e-06,
+      "loss": 1.0114,
+      "step": 3755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 1.0423,
+      "step": 3756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574752681650864e-06,
+      "loss": 1.0268,
+      "step": 3757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569707784383241e-06,
+      "loss": 1.034,
+      "step": 3758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564663876395376e-06,
+      "loss": 1.0516,
+      "step": 3759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559620959141897e-06,
+      "loss": 0.9662,
+      "step": 3760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5545790340771646e-06,
+      "loss": 0.9907,
+      "step": 3761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549538102655245e-06,
+      "loss": 0.9407,
+      "step": 3762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5444981663299135e-06,
+      "loss": 1.0947,
+      "step": 3763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.539459226554669e-06,
+      "loss": 0.9738,
+      "step": 3764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5344212847827145e-06,
+      "loss": 0.9936,
+      "step": 3765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529384342466971e-06,
+      "loss": 0.9749,
+      "step": 3766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5243484010600666e-06,
+      "loss": 1.0523,
+      "step": 3767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519313462014341e-06,
+      "loss": 1.0282,
+      "step": 3768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 1.0277,
+      "step": 3769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509246596814351e-06,
+      "loss": 0.9918,
+      "step": 3770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504214673563321e-06,
+      "loss": 1.0108,
+      "step": 3771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499183758479944e-06,
+      "loss": 1.0367,
+      "step": 3772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494153853015106e-06,
+      "loss": 0.9904,
+      "step": 3773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489124958619413e-06,
+      "loss": 1.0043,
+      "step": 3774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48409707674317e-06,
+      "loss": 1.0188,
+      "step": 3775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479070208836393e-06,
+      "loss": 1.0091,
+      "step": 3776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474044356348812e-06,
+      "loss": 1.075,
+      "step": 3777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4690195207298535e-06,
+      "loss": 1.0174,
+      "step": 3778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463995703428663e-06,
+      "loss": 1.037,
+      "step": 3779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458972905894074e-06,
+      "loss": 0.9966,
+      "step": 3780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 1.0217,
+      "step": 3781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448930375918632e-06,
+      "loss": 0.9436,
+      "step": 3782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443910646373992e-06,
+      "loss": 1.0233,
+      "step": 3783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4388919423883935e-06,
+      "loss": 1.0825,
+      "step": 3784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43387426540921e-06,
+      "loss": 1.039,
+      "step": 3785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428857616883508e-06,
+      "loss": 1.0011,
+      "step": 3786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423841998258069e-06,
+      "loss": 1.0281,
+      "step": 3787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418827410979374e-06,
+      "loss": 1.0029,
+      "step": 3788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4138138564936014e-06,
+      "loss": 0.9716,
+      "step": 3789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408801336246645e-06,
+      "loss": 1.0214,
+      "step": 3790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403789851684081e-06,
+      "loss": 1.044,
+      "step": 3791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398779404251208e-06,
+      "loss": 0.9708,
+      "step": 3792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393769995393003e-06,
+      "loss": 1.0327,
+      "step": 3793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388761626554163e-06,
+      "loss": 1.0486,
+      "step": 3794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383754299179079e-06,
+      "loss": 0.9994,
+      "step": 3795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378748014711834e-06,
+      "loss": 0.9919,
+      "step": 3796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373742774596221e-06,
+      "loss": 0.9861,
+      "step": 3797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368738580275724e-06,
+      "loss": 0.9522,
+      "step": 3798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36373543319353e-06,
+      "loss": 1.0242,
+      "step": 3799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358733334792526e-06,
+      "loss": 0.9616,
+      "step": 3800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.353732286515286e-06,
+      "loss": 1.0819,
+      "step": 3801
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348732289804096e-06,
+      "loss": 0.9876,
+      "step": 3802
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3437333461009235e-06,
+      "loss": 0.986,
+      "step": 3803
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338735456847443e-06,
+      "loss": 1.1055,
+      "step": 3804
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3337386234850255e-06,
+      "loss": 1.0168,
+      "step": 3805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3287428474547256e-06,
+      "loss": 0.992,
+      "step": 3806
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323748130197308e-06,
+      "loss": 0.9746,
+      "step": 3807
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318754473153221e-06,
+      "loss": 1.0693,
+      "step": 3808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313761877762614e-06,
+      "loss": 0.9862,
+      "step": 3809
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308770345465327e-06,
+      "loss": 0.9548,
+      "step": 3810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30377987770089e-06,
+      "loss": 1.0124,
+      "step": 3811
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298790475908539e-06,
+      "loss": 1.0809,
+      "step": 3812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293802141527183e-06,
+      "loss": 1.0088,
+      "step": 3813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288814875995437e-06,
+      "loss": 0.9565,
+      "step": 3814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283828680751611e-06,
+      "loss": 1.0476,
+      "step": 3815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278843557233689e-06,
+      "loss": 0.9547,
+      "step": 3816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273859506879365e-06,
+      "loss": 1.033,
+      "step": 3817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268876531126011e-06,
+      "loss": 1.025,
+      "step": 3818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2638946314106915e-06,
+      "loss": 0.9963,
+      "step": 3819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258913809170169e-06,
+      "loss": 0.9312,
+      "step": 3820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 0.9743,
+      "step": 3821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248955402858965e-06,
+      "loss": 1.0686,
+      "step": 3822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2439778216602475e-06,
+      "loss": 1.0349,
+      "step": 3823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2390013236802316e-06,
+      "loss": 0.9279,
+      "step": 3824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2340259103541226e-06,
+      "loss": 0.9568,
+      "step": 3825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229051583116796e-06,
+      "loss": 1.0549,
+      "step": 3826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224078343402833e-06,
+      "loss": 0.9959,
+      "step": 3827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21910619264649e-06,
+      "loss": 1.0027,
+      "step": 3828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2141351322817065e-06,
+      "loss": 1.0506,
+      "step": 3829
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209165163742119e-06,
+      "loss": 1.0423,
+      "step": 3830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204196288461037e-06,
+      "loss": 0.9496,
+      "step": 3831
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199228507871462e-06,
+      "loss": 1.0583,
+      "step": 3832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194261823406082e-06,
+      "loss": 1.0347,
+      "step": 3833
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18929623649726e-06,
+      "loss": 0.9843,
+      "step": 3834
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18433174857705e-06,
+      "loss": 0.9821,
+      "step": 3835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179368361077183e-06,
+      "loss": 1.0806,
+      "step": 3836
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174406075429075e-06,
+      "loss": 0.9861,
+      "step": 3837
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1694448930638365e-06,
+      "loss": 1.0564,
+      "step": 3838
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164484815412235e-06,
+      "loss": 0.9218,
+      "step": 3839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159525843904741e-06,
+      "loss": 1.0437,
+      "step": 3840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154567979971493e-06,
+      "loss": 1.018,
+      "step": 3841
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149611225042318e-06,
+      "loss": 0.9645,
+      "step": 3842
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144655580546723e-06,
+      "loss": 0.984,
+      "step": 3843
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139701047913885e-06,
+      "loss": 0.9893,
+      "step": 3844
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134747628572677e-06,
+      "loss": 0.9342,
+      "step": 3845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129795323951629e-06,
+      "loss": 1.1025,
+      "step": 3846
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 0.9688,
+      "step": 3847
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119894064582601e-06,
+      "loss": 1.0009,
+      "step": 3848
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114945112690092e-06,
+      "loss": 1.0306,
+      "step": 3849
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1099972812287e-06,
+      "loss": 1.0184,
+      "step": 3850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105050571625354e-06,
+      "loss": 1.0137,
+      "step": 3851
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100104985306665e-06,
+      "loss": 1.0042,
+      "step": 3852
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095160523698913e-06,
+      "loss": 1.0592,
+      "step": 3853
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090217188228058e-06,
+      "loss": 1.036,
+      "step": 3854
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085274980319735e-06,
+      "loss": 1.0121,
+      "step": 3855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080333901399252e-06,
+      "loss": 0.9845,
+      "step": 3856
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075393952891588e-06,
+      "loss": 1.0281,
+      "step": 3857
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0704551362214115e-06,
+      "loss": 1.0092,
+      "step": 3858
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065517452813042e-06,
+      "loss": 0.9373,
+      "step": 3859
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06058090409049e-06,
+      "loss": 1.0439,
+      "step": 3860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.05564549147743e-06,
+      "loss": 0.9867,
+      "step": 3861
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050711216397212e-06,
+      "loss": 1.0343,
+      "step": 3862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04577808027286e-06,
+      "loss": 1.0748,
+      "step": 3863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040846084527059e-06,
+      "loss": 0.942,
+      "step": 3864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.035915230582176e-06,
+      "loss": 0.9533,
+      "step": 3865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.030985519860255e-06,
+      "loss": 1.0532,
+      "step": 3866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026056953782987e-06,
+      "loss": 1.0614,
+      "step": 3867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021129533771756e-06,
+      "loss": 1.0131,
+      "step": 3868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0162032612476e-06,
+      "loss": 0.9853,
+      "step": 3869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011278137631235e-06,
+      "loss": 0.9801,
+      "step": 3870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006354164343047e-06,
+      "loss": 1.0486,
+      "step": 3871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001431342803078e-06,
+      "loss": 1.0292,
+      "step": 3872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996509674431053e-06,
+      "loss": 0.9013,
+      "step": 3873
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991589160646351e-06,
+      "loss": 0.9957,
+      "step": 3874
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986669802868029e-06,
+      "loss": 1.0159,
+      "step": 3875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981751602514809e-06,
+      "loss": 0.9834,
+      "step": 3876
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976834561005069e-06,
+      "loss": 1.0784,
+      "step": 3877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971918679756867e-06,
+      "loss": 1.0274,
+      "step": 3878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967003960187914e-06,
+      "loss": 0.9729,
+      "step": 3879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962090403715592e-06,
+      "loss": 1.004,
+      "step": 3880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957178011756952e-06,
+      "loss": 1.055,
+      "step": 3881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952266785728697e-06,
+      "loss": 0.9567,
+      "step": 3882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947356727047209e-06,
+      "loss": 1.0183,
+      "step": 3883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942447837128517e-06,
+      "loss": 0.9259,
+      "step": 3884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937540117388325e-06,
+      "loss": 1.0724,
+      "step": 3885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.9735,
+      "step": 3886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927728194104558e-06,
+      "loss": 1.005,
+      "step": 3887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922823993390696e-06,
+      "loss": 1.0837,
+      "step": 3888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9179209685147525e-06,
+      "loss": 0.9709,
+      "step": 3889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.913019120890737e-06,
+      "loss": 0.9915,
+      "step": 3890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908118451932328e-06,
+      "loss": 1.0695,
+      "step": 3891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903218963052844e-06,
+      "loss": 0.9047,
+      "step": 3892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898320655665279e-06,
+      "loss": 1.0365,
+      "step": 3893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893423531182283e-06,
+      "loss": 1.0002,
+      "step": 3894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8885275910161574e-06,
+      "loss": 1.0321,
+      "step": 3895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883632836578877e-06,
+      "loss": 0.9922,
+      "step": 3896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878739269282054e-06,
+      "loss": 0.9882,
+      "step": 3897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873846890536976e-06,
+      "loss": 0.9926,
+      "step": 3898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868955701754584e-06,
+      "loss": 1.0476,
+      "step": 3899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864065704345467e-06,
+      "loss": 1.0241,
+      "step": 3900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859176899719883e-06,
+      "loss": 1.0145,
+      "step": 3901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854289289287733e-06,
+      "loss": 0.9586,
+      "step": 3902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849402874458583e-06,
+      "loss": 0.9832,
+      "step": 3903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844517656641655e-06,
+      "loss": 0.9745,
+      "step": 3904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83963363724582e-06,
+      "loss": 1.0429,
+      "step": 3905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834750817679606e-06,
+      "loss": 0.8822,
+      "step": 3906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829869199351188e-06,
+      "loss": 0.9796,
+      "step": 3907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824988783668413e-06,
+      "loss": 1.0249,
+      "step": 3908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820109572038761e-06,
+      "loss": 1.0351,
+      "step": 3909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815231565869377e-06,
+      "loss": 0.9844,
+      "step": 3910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810354766567052e-06,
+      "loss": 1.0073,
+      "step": 3911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8054791755382286e-06,
+      "loss": 1.0592,
+      "step": 3912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800604794189012e-06,
+      "loss": 0.9433,
+      "step": 3913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795731623925146e-06,
+      "loss": 1.0064,
+      "step": 3914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790859666152029e-06,
+      "loss": 0.9731,
+      "step": 3915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785988922274711e-06,
+      "loss": 1.0278,
+      "step": 3916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781119393697891e-06,
+      "loss": 0.9749,
+      "step": 3917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776251081825912e-06,
+      "loss": 1.0406,
+      "step": 3918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771383988062787e-06,
+      "loss": 0.9996,
+      "step": 3919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766518113812141e-06,
+      "loss": 0.9683,
+      "step": 3920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761653460477286e-06,
+      "loss": 0.9924,
+      "step": 3921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756790029461159e-06,
+      "loss": 1.0308,
+      "step": 3922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751927822166346e-06,
+      "loss": 1.0542,
+      "step": 3923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7470668399950936e-06,
+      "loss": 1.0083,
+      "step": 3924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 1.0074,
+      "step": 3925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7373485566304285e-06,
+      "loss": 1.0114,
+      "step": 3926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732491258239723e-06,
+      "loss": 1.0341,
+      "step": 3927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727635190577986e-06,
+      "loss": 0.9625,
+      "step": 3928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722780355045682e-06,
+      "loss": 0.9726,
+      "step": 3929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717926753042922e-06,
+      "loss": 1.0432,
+      "step": 3930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 0.985,
+      "step": 3931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708223255224695e-06,
+      "loss": 1.0375,
+      "step": 3932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703373362207675e-06,
+      "loss": 0.9964,
+      "step": 3933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698524708317082e-06,
+      "loss": 0.987,
+      "step": 3934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6936772949512445e-06,
+      "loss": 0.9986,
+      "step": 3935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688831123508129e-06,
+      "loss": 1.1048,
+      "step": 3936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68398619538536e-06,
+      "loss": 0.9421,
+      "step": 3937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679142511980176e-06,
+      "loss": 1.0494,
+      "step": 3938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674300074689488e-06,
+      "loss": 1.0429,
+      "step": 3939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669458884909815e-06,
+      "loss": 0.9948,
+      "step": 3940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664618944037346e-06,
+      "loss": 1.0095,
+      "step": 3941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659780253467891e-06,
+      "loss": 0.9824,
+      "step": 3942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654942814596902e-06,
+      "loss": 0.9899,
+      "step": 3943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650106628819485e-06,
+      "loss": 0.9846,
+      "step": 3944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645271697530355e-06,
+      "loss": 0.9587,
+      "step": 3945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640438022123898e-06,
+      "loss": 1.0023,
+      "step": 3946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635605603994118e-06,
+      "loss": 0.9964,
+      "step": 3947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6307744445346595e-06,
+      "loss": 1.0016,
+      "step": 3948
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625944545138805e-06,
+      "loss": 0.9463,
+      "step": 3949
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621115907199477e-06,
+      "loss": 1.1132,
+      "step": 3950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 0.9762,
+      "step": 3951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611462421260251e-06,
+      "loss": 0.9438,
+      "step": 3952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606637576044376e-06,
+      "loss": 1.0262,
+      "step": 3953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601813997853063e-06,
+      "loss": 0.9776,
+      "step": 3954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596991688077409e-06,
+      "loss": 1.0306,
+      "step": 3955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5921706481081405e-06,
+      "loss": 1.0013,
+      "step": 3956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5873508793356346e-06,
+      "loss": 0.9962,
+      "step": 3957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582532383149872e-06,
+      "loss": 1.0508,
+      "step": 3958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577715160940502e-06,
+      "loss": 1.032,
+      "step": 3959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572899214096772e-06,
+      "loss": 0.9466,
+      "step": 3960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5680845440075885e-06,
+      "loss": 1.0513,
+      "step": 3961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563271152061476e-06,
+      "loss": 0.9228,
+      "step": 3962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558459039646592e-06,
+      "loss": 0.987,
+      "step": 3963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553648208150728e-06,
+      "loss": 1.0482,
+      "step": 3964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548838658961302e-06,
+      "loss": 0.9785,
+      "step": 3965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5440303934653694e-06,
+      "loss": 1.0216,
+      "step": 3966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53922341304961e-06,
+      "loss": 1.0236,
+      "step": 3967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534417719100332e-06,
+      "loss": 0.9437,
+      "step": 3968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529613313003471e-06,
+      "loss": 1.0335,
+      "step": 3969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5248101961446065e-06,
+      "loss": 0.9473,
+      "step": 3970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.520008369908918e-06,
+      "loss": 1.0414,
+      "step": 3971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515207835681246e-06,
+      "loss": 0.9947,
+      "step": 3972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510408594846024e-06,
+      "loss": 0.9647,
+      "step": 3973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5056106487873425e-06,
+      "loss": 1.0477,
+      "step": 3974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500813998888903e-06,
+      "loss": 1.0411,
+      "step": 3975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496018646534032e-06,
+      "loss": 0.9755,
+      "step": 3976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491224593105695e-06,
+      "loss": 0.9545,
+      "step": 3977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486431839986463e-06,
+      "loss": 1.0661,
+      "step": 3978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481640388558551e-06,
+      "loss": 0.9988,
+      "step": 3979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4768502402037884e-06,
+      "loss": 0.9887,
+      "step": 3980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47206139630363e-06,
+      "loss": 1.0608,
+      "step": 3981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467273858239155e-06,
+      "loss": 1.0138,
+      "step": 3982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4624876273910665e-06,
+      "loss": 0.9612,
+      "step": 3983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4577027051396895e-06,
+      "loss": 0.9648,
+      "step": 3984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452919092864976e-06,
+      "loss": 1.0482,
+      "step": 3985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448136791946495e-06,
+      "loss": 1.0093,
+      "step": 3986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4433558037634386e-06,
+      "loss": 0.9882,
+      "step": 3987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43857612969462e-06,
+      "loss": 1.0582,
+      "step": 3988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433797771118472e-06,
+      "loss": 1.0156,
+      "step": 3989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429020729413062e-06,
+      "loss": 0.9851,
+      "step": 3990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424245005956048e-06,
+      "loss": 1.0181,
+      "step": 3991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419470602124742e-06,
+      "loss": 1.051,
+      "step": 3992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414697519296046e-06,
+      "loss": 1.0044,
+      "step": 3993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409925758846501e-06,
+      "loss": 0.9928,
+      "step": 3994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405155322152262e-06,
+      "loss": 1.0578,
+      "step": 3995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40038621058909e-06,
+      "loss": 0.9823,
+      "step": 3996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39561842553239e-06,
+      "loss": 1.0024,
+      "step": 3997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3908519683571495e-06,
+      "loss": 0.9893,
+      "step": 3998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386086840438004e-06,
+      "loss": 1.0252,
+      "step": 3999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381323043149192e-06,
+      "loss": 1.0247,
+      "step": 4000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3765605778645665e-06,
+      "loss": 0.955,
+      "step": 4001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371799445957598e-06,
+      "loss": 0.9925,
+      "step": 4002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 0.9709,
+      "step": 4003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3622811877686146e-06,
+      "loss": 0.9766,
+      "step": 4004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357524064231616e-06,
+      "loss": 0.9805,
+      "step": 4005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 1.0451,
+      "step": 4006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34801383513226e-06,
+      "loss": 0.9598,
+      "step": 4007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343260732312606e-06,
+      "loss": 1.0149,
+      "step": 4008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338508972474125e-06,
+      "loss": 1.0502,
+      "step": 4009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333758556987211e-06,
+      "loss": 0.9682,
+      "step": 4010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329009487221846e-06,
+      "loss": 0.969,
+      "step": 4011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32426176454765e-06,
+      "loss": 0.9878,
+      "step": 4012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31951539033384e-06,
+      "loss": 0.976,
+      "step": 4013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314770365949249e-06,
+      "loss": 1.0277,
+      "step": 4014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310026692762316e-06,
+      "loss": 0.9761,
+      "step": 4015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305284372141095e-06,
+      "loss": 0.8786,
+      "step": 4016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300543405453245e-06,
+      "loss": 1.0457,
+      "step": 4017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295803794066046e-06,
+      "loss": 0.996,
+      "step": 4018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291065539346374e-06,
+      "loss": 0.9865,
+      "step": 4019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286328642660719e-06,
+      "loss": 0.9995,
+      "step": 4020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28159310537518e-06,
+      "loss": 1.0029,
+      "step": 4021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276858928855459e-06,
+      "loss": 0.9978,
+      "step": 4022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27212611446688e-06,
+      "loss": 1.001,
+      "step": 4023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267394663574351e-06,
+      "loss": 1.0209,
+      "step": 4024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262664577542414e-06,
+      "loss": 0.9435,
+      "step": 4025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257935857735186e-06,
+      "loss": 0.9957,
+      "step": 4026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2532085055164205e-06,
+      "loss": 1.0037,
+      "step": 4027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2484825222494585e-06,
+      "loss": 1.0103,
+      "step": 4028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243757909297247e-06,
+      "loss": 0.9061,
+      "step": 4029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239034668022353e-06,
+      "loss": 1.028,
+      "step": 4030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234312799786921e-06,
+      "loss": 0.9845,
+      "step": 4031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2295923059527285e-06,
+      "loss": 0.9647,
+      "step": 4032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224873187881136e-06,
+      "loss": 1.0147,
+      "step": 4033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220155446933117e-06,
+      "loss": 1.0679,
+      "step": 4034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215439084469245e-06,
+      "loss": 0.9562,
+      "step": 4035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210724101849696e-06,
+      "loss": 1.0087,
+      "step": 4036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2060105004342435e-06,
+      "loss": 1.0638,
+      "step": 4037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201298281582276e-06,
+      "loss": 0.9653,
+      "step": 4038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19658744665277e-06,
+      "loss": 0.9884,
+      "step": 4039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191877997004309e-06,
+      "loss": 1.0006,
+      "step": 4040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187169933995075e-06,
+      "loss": 0.9734,
+      "step": 4041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1824632589828465e-06,
+      "loss": 1.0714,
+      "step": 4042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177757973325017e-06,
+      "loss": 0.9751,
+      "step": 4043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173054078378554e-06,
+      "loss": 1.0273,
+      "step": 4044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168351575500049e-06,
+      "loss": 0.977,
+      "step": 4045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163650466045678e-06,
+      "loss": 1.0016,
+      "step": 4046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.158950751371217e-06,
+      "loss": 1.016,
+      "step": 4047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154252432832043e-06,
+      "loss": 1.0043,
+      "step": 4048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149555511783127e-06,
+      "loss": 0.9431,
+      "step": 4049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144859989579034e-06,
+      "loss": 1.0098,
+      "step": 4050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14016586757394e-06,
+      "loss": 1.0661,
+      "step": 4051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135473147121601e-06,
+      "loss": 0.9917,
+      "step": 4052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130781829575375e-06,
+      "loss": 0.9816,
+      "step": 4053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126091916288218e-06,
+      "loss": 1.0302,
+      "step": 4054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121403408612672e-06,
+      "loss": 0.9887,
+      "step": 4055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116716307900893e-06,
+      "loss": 0.9875,
+      "step": 4056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112030615504601e-06,
+      "loss": 1.0343,
+      "step": 4057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10734633277514e-06,
+      "loss": 1.0755,
+      "step": 4058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102663461063433e-06,
+      "loss": 0.9573,
+      "step": 4059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097982001719994e-06,
+      "loss": 0.9975,
+      "step": 4060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.093301956094934e-06,
+      "loss": 1.0422,
+      "step": 4061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088623325537953e-06,
+      "loss": 1.0643,
+      "step": 4062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083946111398356e-06,
+      "loss": 1.0267,
+      "step": 4063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0792703150250135e-06,
+      "loss": 0.9062,
+      "step": 4064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074595937766415e-06,
+      "loss": 1.0392,
+      "step": 4065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069922980970626e-06,
+      "loss": 0.9374,
+      "step": 4066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065251445985302e-06,
+      "loss": 0.9973,
+      "step": 4067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060581334157693e-06,
+      "loss": 1.0031,
+      "step": 4068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0559126468346354e-06,
+      "loss": 1.0036,
+      "step": 4069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051245385362553e-06,
+      "loss": 1.0102,
+      "step": 4070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046579551087469e-06,
+      "loss": 1.0142,
+      "step": 4071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041915145354983e-06,
+      "loss": 0.9989,
+      "step": 4072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0372521695102896e-06,
+      "loss": 1.0296,
+      "step": 4073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0325906248981655e-06,
+      "loss": 0.9688,
+      "step": 4074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027930512862976e-06,
+      "loss": 1.0552,
+      "step": 4075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0232718347486864e-06,
+      "loss": 0.9575,
+      "step": 4076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01861459189882e-06,
+      "loss": 0.9847,
+      "step": 4077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013958785656516e-06,
+      "loss": 0.9797,
+      "step": 4078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009304417364483e-06,
+      "loss": 1.0775,
+      "step": 4079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004651488365018e-06,
+      "loss": 0.9107,
+      "step": 4080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.0097,
+      "step": 4081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995349953610901e-06,
+      "loss": 0.9896,
+      "step": 4082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99070135053877e-06,
+      "loss": 1.0284,
+      "step": 4083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986054192124242e-06,
+      "loss": 0.9521,
+      "step": 4084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981408479707535e-06,
+      "loss": 1.0509,
+      "step": 4085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97676421462845e-06,
+      "loss": 0.9829,
+      "step": 4086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972121398226371e-06,
+      "loss": 0.9524,
+      "step": 4087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96748003184026e-06,
+      "loss": 1.0323,
+      "step": 4088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962840116808675e-06,
+      "loss": 1.0359,
+      "step": 4089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958201654469731e-06,
+      "loss": 0.9744,
+      "step": 4090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953564646161149e-06,
+      "loss": 1.0288,
+      "step": 4091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9489290932202164e-06,
+      "loss": 0.9551,
+      "step": 4092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944294996983805e-06,
+      "loss": 0.9419,
+      "step": 4093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939662358788364e-06,
+      "loss": 1.0603,
+      "step": 4094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935031179969922e-06,
+      "loss": 0.9692,
+      "step": 4095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930401461864099e-06,
+      "loss": 1.0298,
+      "step": 4096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925773205806069e-06,
+      "loss": 0.9761,
+      "step": 4097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92114641313061e-06,
+      "loss": 0.9617,
+      "step": 4098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9165210851720625e-06,
+      "loss": 1.0423,
+      "step": 4099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911897223264348e-06,
+      "loss": 0.968,
+      "step": 4100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907274828740968e-06,
+      "loss": 1.016,
+      "step": 4101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902653902934998e-06,
+      "loss": 0.9412,
+      "step": 4102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898034447179085e-06,
+      "loss": 1.0801,
+      "step": 4103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893416462805469e-06,
+      "loss": 1.0099,
+      "step": 4104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888799951145948e-06,
+      "loss": 0.9914,
+      "step": 4105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8841849135319015e-06,
+      "loss": 1.0239,
+      "step": 4106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879571351294287e-06,
+      "loss": 1.0481,
+      "step": 4107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874959265763627e-06,
+      "loss": 1.0213,
+      "step": 4108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8703486582700365e-06,
+      "loss": 0.9656,
+      "step": 4109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865739530143179e-06,
+      "loss": 1.0284,
+      "step": 4110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861131882712314e-06,
+      "loss": 0.9946,
+      "step": 4111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856525717306263e-06,
+      "loss": 0.9437,
+      "step": 4112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851921035253421e-06,
+      "loss": 1.0781,
+      "step": 4113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847317837881757e-06,
+      "loss": 1.0205,
+      "step": 4114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842716126518804e-06,
+      "loss": 0.9727,
+      "step": 4115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838115902491685e-06,
+      "loss": 0.9595,
+      "step": 4116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833517167127077e-06,
+      "loss": 1.0347,
+      "step": 4117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828919921751233e-06,
+      "loss": 0.9833,
+      "step": 4118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824324167689975e-06,
+      "loss": 1.0305,
+      "step": 4119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8197299062687e-06,
+      "loss": 1.0022,
+      "step": 4120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815137138812364e-06,
+      "loss": 1.0779,
+      "step": 4121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810545866645512e-06,
+      "loss": 1.0162,
+      "step": 4122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805956091092228e-06,
+      "loss": 0.9856,
+      "step": 4123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801367813476195e-06,
+      "loss": 1.0618,
+      "step": 4124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796781035120642e-06,
+      "loss": 1.0059,
+      "step": 4125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7921957573483756e-06,
+      "loss": 0.9475,
+      "step": 4126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787611981481775e-06,
+      "loss": 1.0034,
+      "step": 4127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783029708842766e-06,
+      "loss": 0.9931,
+      "step": 4128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778448940752869e-06,
+      "loss": 0.9782,
+      "step": 4129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773869678533138e-06,
+      "loss": 1.0816,
+      "step": 4130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769291923504226e-06,
+      "loss": 0.9609,
+      "step": 4131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764715676986327e-06,
+      "loss": 0.9962,
+      "step": 4132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76014094029921e-06,
+      "loss": 0.9464,
+      "step": 4133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75556771476221e-06,
+      "loss": 1.0756,
+      "step": 4134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750996001694215e-06,
+      "loss": 0.983,
+      "step": 4135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7464258024136945e-06,
+      "loss": 0.9549,
+      "step": 4136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7418571182386684e-06,
+      "loss": 0.9823,
+      "step": 4137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7372899504867235e-06,
+      "loss": 1.0676,
+      "step": 4138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732724300475008e-06,
+      "loss": 0.9521,
+      "step": 4139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728160169520235e-06,
+      "loss": 1.0155,
+      "step": 4140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7235975589386715e-06,
+      "loss": 1.034,
+      "step": 4141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719036470046166e-06,
+      "loss": 0.9846,
+      "step": 4142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714476904158098e-06,
+      "loss": 1.0006,
+      "step": 4143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709918862589438e-06,
+      "loss": 1.0195,
+      "step": 4144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705362346654696e-06,
+      "loss": 1.0116,
+      "step": 4145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700807357667953e-06,
+      "loss": 0.9889,
+      "step": 4146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6962538969428416e-06,
+      "loss": 0.9768,
+      "step": 4147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691701965792559e-06,
+      "loss": 0.9995,
+      "step": 4148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687151565529864e-06,
+      "loss": 1.0395,
+      "step": 4149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6826026974670665e-06,
+      "loss": 0.9602,
+      "step": 4150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6780553629160405e-06,
+      "loss": 0.9564,
+      "step": 4151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673509563188214e-06,
+      "loss": 1.0636,
+      "step": 4152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668965299594574e-06,
+      "loss": 0.9566,
+      "step": 4153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664422573445661e-06,
+      "loss": 1.0079,
+      "step": 4154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659881386051586e-06,
+      "loss": 1.0613,
+      "step": 4155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655341738721989e-06,
+      "loss": 1.0138,
+      "step": 4156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6508036327660965e-06,
+      "loss": 1.0041,
+      "step": 4157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6462670694926705e-06,
+      "loss": 0.975,
+      "step": 4158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 1.0193,
+      "step": 4159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637198576226068e-06,
+      "loss": 1.0232,
+      "step": 4160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6326666488481975e-06,
+      "loss": 1.0053,
+      "step": 4161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628136269383421e-06,
+      "loss": 1.0467,
+      "step": 4162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6236074391382625e-06,
+      "loss": 0.9429,
+      "step": 4163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6190801594188265e-06,
+      "loss": 1.0208,
+      "step": 4164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614554431530754e-06,
+      "loss": 1.0006,
+      "step": 4165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6100302567792444e-06,
+      "loss": 0.9796,
+      "step": 4166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605507636469047e-06,
+      "loss": 1.0224,
+      "step": 4167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600986571904461e-06,
+      "loss": 1.0136,
+      "step": 4168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596467064389346e-06,
+      "loss": 0.9644,
+      "step": 4169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5919491152271035e-06,
+      "loss": 0.9906,
+      "step": 4170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587432725720687e-06,
+      "loss": 0.9882,
+      "step": 4171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582917897172603e-06,
+      "loss": 1.0161,
+      "step": 4172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578404630884905e-06,
+      "loss": 0.9931,
+      "step": 4173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573892928159195e-06,
+      "loss": 0.9876,
+      "step": 4174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569382790296636e-06,
+      "loss": 0.9038,
+      "step": 4175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564874218597916e-06,
+      "loss": 1.0068,
+      "step": 4176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560367214363295e-06,
+      "loss": 1.0334,
+      "step": 4177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555861778892569e-06,
+      "loss": 1.0092,
+      "step": 4178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551357913485082e-06,
+      "loss": 0.9968,
+      "step": 4179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546855619439734e-06,
+      "loss": 1.0137,
+      "step": 4180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542354898054953e-06,
+      "loss": 0.9815,
+      "step": 4181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53785575062874e-06,
+      "loss": 0.9754,
+      "step": 4182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533358178458613e-06,
+      "loss": 1.0685,
+      "step": 4183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528862182841659e-06,
+      "loss": 0.9906,
+      "step": 4184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524367765074499e-06,
+      "loss": 1.0733,
+      "step": 4185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519874926453303e-06,
+      "loss": 1.0207,
+      "step": 4186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515383668273778e-06,
+      "loss": 0.9318,
+      "step": 4187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5108939918311886e-06,
+      "loss": 1.0263,
+      "step": 4188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506405898420335e-06,
+      "loss": 1.0814,
+      "step": 4189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501919389335558e-06,
+      "loss": 0.9763,
+      "step": 4190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497434465870749e-06,
+      "loss": 0.9351,
+      "step": 4191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4929511293193315e-06,
+      "loss": 1.0047,
+      "step": 4192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488469380974291e-06,
+      "loss": 1.0321,
+      "step": 4193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483989222128127e-06,
+      "loss": 1.0403,
+      "step": 4194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479510654072909e-06,
+      "loss": 0.9766,
+      "step": 4195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4750336781002225e-06,
+      "loss": 1.0329,
+      "step": 4196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470558295501214e-06,
+      "loss": 1.0024,
+      "step": 4197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46608450756656e-06,
+      "loss": 1.0036,
+      "step": 4198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461612315586481e-06,
+      "loss": 0.9619,
+      "step": 4199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4571417208507326e-06,
+      "loss": 1.0238,
+      "step": 4200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4526727246486116e-06,
+      "loss": 0.9707,
+      "step": 4201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44820532826896e-06,
+      "loss": 0.945,
+      "step": 4202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443739533000151e-06,
+      "loss": 1.0542,
+      "step": 4203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439275340130099e-06,
+      "loss": 1.0156,
+      "step": 4204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434812750946255e-06,
+      "loss": 1.0233,
+      "step": 4205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430351766735609e-06,
+      "loss": 0.9371,
+      "step": 4206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425892388784681e-06,
+      "loss": 1.0701,
+      "step": 4207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42143461837955e-06,
+      "loss": 0.8703,
+      "step": 4208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416978456805796e-06,
+      "loss": 0.9886,
+      "step": 4209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412523905348568e-06,
+      "loss": 1.0327,
+      "step": 4210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 1.022,
+      "step": 4211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403619637921894e-06,
+      "loss": 0.9895,
+      "step": 4212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399169924520403e-06,
+      "loss": 0.9998,
+      "step": 4213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394721826371322e-06,
+      "loss": 1.0471,
+      "step": 4214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390275344757475e-06,
+      "loss": 0.9859,
+      "step": 4215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385830480961192e-06,
+      "loss": 0.9973,
+      "step": 4216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38138723626436e-06,
+      "loss": 1.0209,
+      "step": 4217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3769456119483855e-06,
+      "loss": 0.9773,
+      "step": 4218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372505609294214e-06,
+      "loss": 0.9825,
+      "step": 4219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368067229582319e-06,
+      "loss": 0.9457,
+      "step": 4220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363630474092705e-06,
+      "loss": 1.0834,
+      "step": 4221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359195344104916e-06,
+      "loss": 0.9914,
+      "step": 4222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354761840898022e-06,
+      "loss": 1.0106,
+      "step": 4223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350329965750622e-06,
+      "loss": 0.936,
+      "step": 4224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345899719940844e-06,
+      "loss": 0.9995,
+      "step": 4225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34147110474636e-06,
+      "loss": 0.9895,
+      "step": 4226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337044121444347e-06,
+      "loss": 1.0094,
+      "step": 4227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332618771311542e-06,
+      "loss": 0.9734,
+      "step": 4228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328195055624177e-06,
+      "loss": 0.9914,
+      "step": 4229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323772975658044e-06,
+      "loss": 0.9843,
+      "step": 4230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 1.0702,
+      "step": 4231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314933727990211e-06,
+      "loss": 0.9891,
+      "step": 4232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31051656283771e-06,
+      "loss": 0.9662,
+      "step": 4233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306101038504824e-06,
+      "loss": 0.9411,
+      "step": 4234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3016871562649776e-06,
+      "loss": 1.0477,
+      "step": 4235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297274917391108e-06,
+      "loss": 0.9628,
+      "step": 4236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.9621,
+      "step": 4237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288455374830701e-06,
+      "loss": 1.0644,
+      "step": 4238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2840480736876774e-06,
+      "loss": 0.9826,
+      "step": 4239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279642420997655e-06,
+      "loss": 1.0144,
+      "step": 4240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275238418031209e-06,
+      "loss": 0.9385,
+      "step": 4241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270836066058429e-06,
+      "loss": 1.0735,
+      "step": 4242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266435366348932e-06,
+      "loss": 1.0628,
+      "step": 4243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262036320171859e-06,
+      "loss": 0.9924,
+      "step": 4244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257638928795872e-06,
+      "loss": 1.0142,
+      "step": 4245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2532431934891646e-06,
+      "loss": 0.9204,
+      "step": 4246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248849115519435e-06,
+      "loss": 0.9889,
+      "step": 4247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244456696153925e-06,
+      "loss": 1.0268,
+      "step": 4248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240065936659374e-06,
+      "loss": 1.0241,
+      "step": 4249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235676838302069e-06,
+      "loss": 0.9808,
+      "step": 4250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231289402347798e-06,
+      "loss": 0.908,
+      "step": 4251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.226903630061878e-06,
+      "loss": 0.9828,
+      "step": 4252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222519522709145e-06,
+      "loss": 0.9895,
+      "step": 4253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21813708155395e-06,
+      "loss": 0.9537,
+      "step": 4254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213756307860175e-06,
+      "loss": 1.01,
+      "step": 4255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209377202891212e-06,
+      "loss": 1.0391,
+      "step": 4256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204999767909972e-06,
+      "loss": 0.9057,
+      "step": 4257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200624004178883e-06,
+      "loss": 1.0104,
+      "step": 4258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196249912959905e-06,
+      "loss": 1.0611,
+      "step": 4259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191877495514489e-06,
+      "loss": 0.9994,
+      "step": 4260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187506753103637e-06,
+      "loss": 0.9693,
+      "step": 4261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183137686987834e-06,
+      "loss": 1.0825,
+      "step": 4262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178770298427107e-06,
+      "loss": 0.9603,
+      "step": 4263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174404588680988e-06,
+      "loss": 0.9519,
+      "step": 4264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170040559008522e-06,
+      "loss": 0.9739,
+      "step": 4265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165678210668285e-06,
+      "loss": 0.9924,
+      "step": 4266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161317544918345e-06,
+      "loss": 0.9513,
+      "step": 4267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1569585630163044e-06,
+      "loss": 1.0082,
+      "step": 4268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152601266219271e-06,
+      "loss": 0.9303,
+      "step": 4269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148245655783869e-06,
+      "loss": 1.0516,
+      "step": 4270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143891732966233e-06,
+      "loss": 1.0063,
+      "step": 4271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139539499022015e-06,
+      "loss": 0.9853,
+      "step": 4272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135188955206375e-06,
+      "loss": 1.0072,
+      "step": 4273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130840102773995e-06,
+      "loss": 0.9264,
+      "step": 4274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12649294297906e-06,
+      "loss": 0.9979,
+      "step": 4275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 1.054,
+      "step": 4276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117803706315836e-06,
+      "loss": 1.0193,
+      "step": 4277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113461631953477e-06,
+      "loss": 1.0432,
+      "step": 4278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109121255240438e-06,
+      "loss": 0.8834,
+      "step": 4279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104782577428448e-06,
+      "loss": 1.0228,
+      "step": 4280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100445599768774e-06,
+      "loss": 1.0121,
+      "step": 4281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096110323512167e-06,
+      "loss": 0.9711,
+      "step": 4282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09177674990891e-06,
+      "loss": 0.9955,
+      "step": 4283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0874448802087794e-06,
+      "loss": 0.9721,
+      "step": 4284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083114715661069e-06,
+      "loss": 0.9505,
+      "step": 4285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078786257514573e-06,
+      "loss": 0.9679,
+      "step": 4286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074459507017597e-06,
+      "loss": 1.0425,
+      "step": 4287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070134465417963e-06,
+      "loss": 0.9887,
+      "step": 4288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065811133962987e-06,
+      "loss": 0.9717,
+      "step": 4289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061489513899497e-06,
+      "loss": 1.0258,
+      "step": 4290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057169606473828e-06,
+      "loss": 0.9759,
+      "step": 4291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052851412931819e-06,
+      "loss": 1.0155,
+      "step": 4292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048534934518816e-06,
+      "loss": 0.9845,
+      "step": 4293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044220172479675e-06,
+      "loss": 1.0216,
+      "step": 4294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039907128058749e-06,
+      "loss": 1.0275,
+      "step": 4295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0355958024999e-06,
+      "loss": 0.9886,
+      "step": 4296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031286197046493e-06,
+      "loss": 0.9804,
+      "step": 4297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026978312941396e-06,
+      "loss": 1.0004,
+      "step": 4298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022672151426989e-06,
+      "loss": 0.9929,
+      "step": 4299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018367713745137e-06,
+      "loss": 1.0203,
+      "step": 4300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014065001137229e-06,
+      "loss": 1.0055,
+      "step": 4301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009764014844143e-06,
+      "loss": 1.0335,
+      "step": 4302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0054647561062625e-06,
+      "loss": 0.9591,
+      "step": 4303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001167226163472e-06,
+      "loss": 1.0695,
+      "step": 4304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99687142625516e-06,
+      "loss": 1.0101,
+      "step": 4305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99257735762021e-06,
+      "loss": 0.9631,
+      "step": 4306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98828502149702e-06,
+      "loss": 1.0264,
+      "step": 4307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983994419123472e-06,
+      "loss": 0.9999,
+      "step": 4308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979705551736957e-06,
+      "loss": 0.9369,
+      "step": 4309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975418420574364e-06,
+      "loss": 0.9894,
+      "step": 4310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971133026872077e-06,
+      "loss": 1.0367,
+      "step": 4311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9668493718659924e-06,
+      "loss": 0.9309,
+      "step": 4312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962567456791484e-06,
+      "loss": 0.9516,
+      "step": 4313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958287282883449e-06,
+      "loss": 0.9789,
+      "step": 4314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 1.0554,
+      "step": 4315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949732163503786e-06,
+      "loss": 1.0328,
+      "step": 4316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945457220499422e-06,
+      "loss": 0.9633,
+      "step": 4317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94118402359603e-06,
+      "loss": 1.0322,
+      "step": 4318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936912574025988e-06,
+      "loss": 0.9982,
+      "step": 4319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932642873021151e-06,
+      "loss": 1.0228,
+      "step": 4320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9283749218128885e-06,
+      "loss": 1.0243,
+      "step": 4321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924108721632054e-06,
+      "loss": 0.9636,
+      "step": 4322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919844273708999e-06,
+      "loss": 0.9962,
+      "step": 4323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915581579273569e-06,
+      "loss": 0.987,
+      "step": 4324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911320639555107e-06,
+      "loss": 1.0886,
+      "step": 4325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907061455782441e-06,
+      "loss": 0.9849,
+      "step": 4326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902804029183907e-06,
+      "loss": 0.9694,
+      "step": 4327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898548360987325e-06,
+      "loss": 1.0413,
+      "step": 4328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8942944524200054e-06,
+      "loss": 0.9162,
+      "step": 4329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890042304708758e-06,
+      "loss": 0.9837,
+      "step": 4330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885791919079878e-06,
+      "loss": 0.9839,
+      "step": 4331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881543296759165e-06,
+      "loss": 1.0223,
+      "step": 4332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877296438971888e-06,
+      "loss": 0.9306,
+      "step": 4333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873051346942831e-06,
+      "loss": 1.0185,
+      "step": 4334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868808021896254e-06,
+      "loss": 1.0505,
+      "step": 4335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864566465055913e-06,
+      "loss": 1.0203,
+      "step": 4336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860326677645051e-06,
+      "loss": 0.8985,
+      "step": 4337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856088660886402e-06,
+      "loss": 1.0837,
+      "step": 4338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8518524160021876e-06,
+      "loss": 0.9917,
+      "step": 4339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847617944214127e-06,
+      "loss": 1.0268,
+      "step": 4340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.9762,
+      "step": 4341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839154324810749e-06,
+      "loss": 1.0505,
+      "step": 4342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8349251796363e-06,
+      "loss": 0.96,
+      "step": 4343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830697812439728e-06,
+      "loss": 1.0101,
+      "step": 4344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826472224440202e-06,
+      "loss": 0.9308,
+      "step": 4345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8222484168563426e-06,
+      "loss": 1.0295,
+      "step": 4346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818026390906291e-06,
+      "loss": 0.9488,
+      "step": 4347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813806147807645e-06,
+      "loss": 0.9879,
+      "step": 4348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809587688777514e-06,
+      "loss": 1.0405,
+      "step": 4349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805371015032476e-06,
+      "loss": 0.945,
+      "step": 4350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8011561277885965e-06,
+      "loss": 0.9919,
+      "step": 4351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7969430282614404e-06,
+      "loss": 0.9895,
+      "step": 4352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7927317176660295e-06,
+      "loss": 1.0314,
+      "step": 4353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7885221972168974e-06,
+      "loss": 1.0285,
+      "step": 4354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784314468128044e-06,
+      "loss": 0.9637,
+      "step": 4355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7801085316129615e-06,
+      "loss": 1.0247,
+      "step": 4356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775904388884618e-06,
+      "loss": 1.0318,
+      "step": 4357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7717020411554707e-06,
+      "loss": 0.9684,
+      "step": 4358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7675014896374517e-06,
+      "loss": 0.981,
+      "step": 4359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763302735541987e-06,
+      "loss": 1.0185,
+      "step": 4360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7591057800799747e-06,
+      "loss": 0.8695,
+      "step": 4361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.754910624461795e-06,
+      "loss": 0.9984,
+      "step": 4362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7507172698973105e-06,
+      "loss": 1.0628,
+      "step": 4363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7465257175958623e-06,
+      "loss": 0.9693,
+      "step": 4364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742335968766283e-06,
+      "loss": 0.969,
+      "step": 4365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738148024616863e-06,
+      "loss": 1.0377,
+      "step": 4366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7339618863553983e-06,
+      "loss": 0.9615,
+      "step": 4367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.729777555189139e-06,
+      "loss": 1.0205,
+      "step": 4368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7255950323248335e-06,
+      "loss": 1.0333,
+      "step": 4369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7214143189687e-06,
+      "loss": 0.999,
+      "step": 4370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7172354163264324e-06,
+      "loss": 1.0086,
+      "step": 4371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713058325603213e-06,
+      "loss": 1.0098,
+      "step": 4372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7088830480036898e-06,
+      "loss": 0.9665,
+      "step": 4373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7047095847319935e-06,
+      "loss": 1.0473,
+      "step": 4374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.700537936991733e-06,
+      "loss": 0.9958,
+      "step": 4375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696368105985988e-06,
+      "loss": 0.9594,
+      "step": 4376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692200092917315e-06,
+      "loss": 1.0329,
+      "step": 4377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6880338989877607e-06,
+      "loss": 0.9549,
+      "step": 4378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683869525398821e-06,
+      "loss": 1.0511,
+      "step": 4379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679706973351491e-06,
+      "loss": 0.9879,
+      "step": 4380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6755462440462288e-06,
+      "loss": 1.0111,
+      "step": 4381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6713873386829658e-06,
+      "loss": 0.9733,
+      "step": 4382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.667230258461113e-06,
+      "loss": 0.9768,
+      "step": 4383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6630750045795472e-06,
+      "loss": 1.0268,
+      "step": 4384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658921578236635e-06,
+      "loss": 0.9443,
+      "step": 4385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6547699806301896e-06,
+      "loss": 1.0356,
+      "step": 4386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650620212957524e-06,
+      "loss": 1.0819,
+      "step": 4387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646472276415406e-06,
+      "loss": 0.949,
+      "step": 4388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.64232617220008e-06,
+      "loss": 0.9745,
+      "step": 4389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6381819015072652e-06,
+      "loss": 1.0337,
+      "step": 4390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6340394655321465e-06,
+      "loss": 1.056,
+      "step": 4391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6298988654693814e-06,
+      "loss": 0.9699,
+      "step": 4392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.9772,
+      "step": 4393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62162317785691e-06,
+      "loss": 1.0129,
+      "step": 4394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6174880926938704e-06,
+      "loss": 0.975,
+      "step": 4395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6133548482165225e-06,
+      "loss": 0.9995,
+      "step": 4396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6092234456168707e-06,
+      "loss": 0.9998,
+      "step": 4397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6050938860864025e-06,
+      "loss": 1.0159,
+      "step": 4398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6009661708160504e-06,
+      "loss": 0.9767,
+      "step": 4399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5968403009962383e-06,
+      "loss": 0.9727,
+      "step": 4400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592716277816836e-06,
+      "loss": 1.0335,
+      "step": 4401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5885941024672e-06,
+      "loss": 0.9907,
+      "step": 4402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5844737761361448e-06,
+      "loss": 0.9867,
+      "step": 4403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5803553000119475e-06,
+      "loss": 0.9394,
+      "step": 4404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5762386752823643e-06,
+      "loss": 1.0757,
+      "step": 4405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.9093,
+      "step": 4406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5680109847553548e-06,
+      "loss": 0.9944,
+      "step": 4407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5638999213307544e-06,
+      "loss": 1.0361,
+      "step": 4408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559790714046417e-06,
+      "loss": 0.9577,
+      "step": 4409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5556833640874145e-06,
+      "loss": 0.9995,
+      "step": 4410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5515778726382967e-06,
+      "loss": 0.9477,
+      "step": 4411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5474742408830544e-06,
+      "loss": 1.0534,
+      "step": 4412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543372470005164e-06,
+      "loss": 0.953,
+      "step": 4413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5392725611875556e-06,
+      "loss": 1.006,
+      "step": 4414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5351745156126217e-06,
+      "loss": 1.0329,
+      "step": 4415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5310783344622178e-06,
+      "loss": 0.9056,
+      "step": 4416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5269840189176616e-06,
+      "loss": 0.9878,
+      "step": 4417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522891570159743e-06,
+      "loss": 0.9967,
+      "step": 4418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5188009893686916e-06,
+      "loss": 1.0481,
+      "step": 4419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5147122777242203e-06,
+      "loss": 0.96,
+      "step": 4420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5106254364054915e-06,
+      "loss": 1.0462,
+      "step": 4421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5065404665911297e-06,
+      "loss": 0.9883,
+      "step": 4422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502457369459221e-06,
+      "loss": 1.0166,
+      "step": 4423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4983761461873077e-06,
+      "loss": 1.0307,
+      "step": 4424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494296797952401e-06,
+      "loss": 0.9603,
+      "step": 4425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4902193259309627e-06,
+      "loss": 1.0111,
+      "step": 4426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4861437312989165e-06,
+      "loss": 0.9123,
+      "step": 4427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482070015231642e-06,
+      "loss": 1.0297,
+      "step": 4428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477998178903982e-06,
+      "loss": 1.0547,
+      "step": 4429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4739282234902304e-06,
+      "loss": 1.0067,
+      "step": 4430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4698601501641517e-06,
+      "loss": 0.921,
+      "step": 4431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4657939600989453e-06,
+      "loss": 1.0455,
+      "step": 4432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461729654467293e-06,
+      "loss": 1.0023,
+      "step": 4433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4576672344413142e-06,
+      "loss": 0.9713,
+      "step": 4434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4536067011925945e-06,
+      "loss": 0.9993,
+      "step": 4435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449548055892171e-06,
+      "loss": 1.0444,
+      "step": 4436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4454912997105338e-06,
+      "loss": 0.9568,
+      "step": 4437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.441436433817641e-06,
+      "loss": 0.9735,
+      "step": 4438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437383459382885e-06,
+      "loss": 1.0298,
+      "step": 4439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4333323775751327e-06,
+      "loss": 0.9221,
+      "step": 4440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4292831895626944e-06,
+      "loss": 1.0143,
+      "step": 4441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4252358965133346e-06,
+      "loss": 1.0123,
+      "step": 4442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4211904995942714e-06,
+      "loss": 1.0682,
+      "step": 4443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4171469999721873e-06,
+      "loss": 0.9633,
+      "step": 4444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4131053988131947e-06,
+      "loss": 0.9781,
+      "step": 4445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409065697282883e-06,
+      "loss": 1.044,
+      "step": 4446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405027896546277e-06,
+      "loss": 0.9958,
+      "step": 4447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.400991997767857e-06,
+      "loss": 1.0359,
+      "step": 4448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396958002111568e-06,
+      "loss": 0.9528,
+      "step": 4449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3929259107407785e-06,
+      "loss": 1.045,
+      "step": 4450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3888957248183406e-06,
+      "loss": 0.9535,
+      "step": 4451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3848674455065256e-06,
+      "loss": 0.9576,
+      "step": 4452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3808410739670805e-06,
+      "loss": 1.0466,
+      "step": 4453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3768166113611865e-06,
+      "loss": 1.0117,
+      "step": 4454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372794058849482e-06,
+      "loss": 0.9312,
+      "step": 4455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 0.9636,
+      "step": 4456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3647546887484217e-06,
+      "loss": 1.0074,
+      "step": 4457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360737873477584e-06,
+      "loss": 1.0476,
+      "step": 4458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3567229729379656e-06,
+      "loss": 0.9012,
+      "step": 4459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3527099882874435e-06,
+      "loss": 1.0634,
+      "step": 4460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348698920683343e-06,
+      "loss": 0.9414,
+      "step": 4461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.344689771282438e-06,
+      "loss": 1.0034,
+      "step": 4462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340682541240943e-06,
+      "loss": 0.9714,
+      "step": 4463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336677231714535e-06,
+      "loss": 0.9567,
+      "step": 4464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3326738438583116e-06,
+      "loss": 0.9718,
+      "step": 4465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3286723788268418e-06,
+      "loss": 0.9676,
+      "step": 4466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3246728377741243e-06,
+      "loss": 1.0164,
+      "step": 4467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3206752218536085e-06,
+      "loss": 0.9557,
+      "step": 4468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3166795322181866e-06,
+      "loss": 1.0441,
+      "step": 4469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.312685770020193e-06,
+      "loss": 0.9185,
+      "step": 4470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 1.0541,
+      "step": 4471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3047040325430812e-06,
+      "loss": 0.9683,
+      "step": 4472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3007160595658527e-06,
+      "loss": 0.9794,
+      "step": 4473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2967300186298456e-06,
+      "loss": 1.0414,
+      "step": 4474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292745910884614e-06,
+      "loss": 0.9656,
+      "step": 4475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288763737479155e-06,
+      "loss": 0.9888,
+      "step": 4476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2847834995619067e-06,
+      "loss": 1.0296,
+      "step": 4477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280805198280754e-06,
+      "loss": 1.0023,
+      "step": 4478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27682883478302e-06,
+      "loss": 1.0093,
+      "step": 4479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272854410215467e-06,
+      "loss": 0.9959,
+      "step": 4480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2688819257242963e-06,
+      "loss": 1.0079,
+      "step": 4481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2649113824551636e-06,
+      "loss": 0.9396,
+      "step": 4482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2609427815531426e-06,
+      "loss": 1.0175,
+      "step": 4483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2569761241627694e-06,
+      "loss": 1.0607,
+      "step": 4484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2530114114279977e-06,
+      "loss": 0.9521,
+      "step": 4485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2490486444922396e-06,
+      "loss": 1.0093,
+      "step": 4486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2450878244983365e-06,
+      "loss": 0.9482,
+      "step": 4487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241128952588569e-06,
+      "loss": 1.0801,
+      "step": 4488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237172029904657e-06,
+      "loss": 0.9206,
+      "step": 4489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233217057587754e-06,
+      "loss": 0.9891,
+      "step": 4490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2292640367784624e-06,
+      "loss": 1.0335,
+      "step": 4491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2253129686168105e-06,
+      "loss": 0.9333,
+      "step": 4492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2213638542422677e-06,
+      "loss": 0.9985,
+      "step": 4493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217416694793739e-06,
+      "loss": 1.0214,
+      "step": 4494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213471491409568e-06,
+      "loss": 1.0268,
+      "step": 4495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2095282452275256e-06,
+      "loss": 0.9942,
+      "step": 4496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2055869573848374e-06,
+      "loss": 1.0049,
+      "step": 4497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2016476290181386e-06,
+      "loss": 1.0396,
+      "step": 4498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1977102612635214e-06,
+      "loss": 1.0356,
+      "step": 4499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1937748552565017e-06,
+      "loss": 0.967,
+      "step": 4500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1898414121320277e-06,
+      "loss": 0.981,
+      "step": 4501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1859099330244934e-06,
+      "loss": 1.0046,
+      "step": 4502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1819804190677084e-06,
+      "loss": 0.95,
+      "step": 4503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1780528713949375e-06,
+      "loss": 0.9218,
+      "step": 4504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174127291138853e-06,
+      "loss": 1.0262,
+      "step": 4505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1702036794315837e-06,
+      "loss": 1.0065,
+      "step": 4506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1662820374046776e-06,
+      "loss": 0.963,
+      "step": 4507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162362366189116e-06,
+      "loss": 0.9555,
+      "step": 4508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1584446669153137e-06,
+      "loss": 1.0715,
+      "step": 4509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1545289407131128e-06,
+      "loss": 0.93,
+      "step": 4510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1506151887117974e-06,
+      "loss": 1.0002,
+      "step": 4511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146703412040071e-06,
+      "loss": 1.0483,
+      "step": 4512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1427936118260693e-06,
+      "loss": 1.0232,
+      "step": 4513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13888578919736e-06,
+      "loss": 0.9725,
+      "step": 4514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134979945280948e-06,
+      "loss": 0.9733,
+      "step": 4515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131076081203247e-06,
+      "loss": 0.9983,
+      "step": 4516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127174198090126e-06,
+      "loss": 0.9606,
+      "step": 4517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123274297066856e-06,
+      "loss": 1.0001,
+      "step": 4518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11937637925816e-06,
+      "loss": 1.0377,
+      "step": 4519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1154804457881737e-06,
+      "loss": 0.9466,
+      "step": 4520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1115864977804676e-06,
+      "loss": 1.0294,
+      "step": 4521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107694536358037e-06,
+      "loss": 0.9841,
+      "step": 4522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103804562643302e-06,
+      "loss": 1.0467,
+      "step": 4523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0999165777581174e-06,
+      "loss": 0.8987,
+      "step": 4524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0960305828237568e-06,
+      "loss": 1.0128,
+      "step": 4525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0921465789609227e-06,
+      "loss": 1.0658,
+      "step": 4526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0882645672897425e-06,
+      "loss": 0.9988,
+      "step": 4527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0843845489297698e-06,
+      "loss": 0.9693,
+      "step": 4528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080506524999981e-06,
+      "loss": 0.9949,
+      "step": 4529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0766304966187876e-06,
+      "loss": 0.9932,
+      "step": 4530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0727564649040066e-06,
+      "loss": 0.985,
+      "step": 4531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068884430972897e-06,
+      "loss": 0.9651,
+      "step": 4532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065014395942134e-06,
+      "loss": 1.0296,
+      "step": 4533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0611463609278135e-06,
+      "loss": 0.9691,
+      "step": 4534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0572803270454664e-06,
+      "loss": 0.9418,
+      "step": 4535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0534162954100264e-06,
+      "loss": 1.0251,
+      "step": 4536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0495542671358745e-06,
+      "loss": 1.0187,
+      "step": 4537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0456942433367874e-06,
+      "loss": 0.9845,
+      "step": 4538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0418362251259892e-06,
+      "loss": 1.0051,
+      "step": 4539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0379802136161073e-06,
+      "loss": 1.006,
+      "step": 4540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0341262099191993e-06,
+      "loss": 0.9777,
+      "step": 4541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03027421514674e-06,
+      "loss": 0.9759,
+      "step": 4542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026424230409624e-06,
+      "loss": 1.0311,
+      "step": 4543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0225762568181736e-06,
+      "loss": 0.9739,
+      "step": 4544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.018730295482124e-06,
+      "loss": 0.9953,
+      "step": 4545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0148863475106315e-06,
+      "loss": 0.9415,
+      "step": 4546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0110444140122706e-06,
+      "loss": 1.0294,
+      "step": 4547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0072044960950388e-06,
+      "loss": 0.9613,
+      "step": 4548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 1.001,
+      "step": 4549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999530711433032e-06,
+      "loss": 1.0571,
+      "step": 4550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995696846901337e-06,
+      "loss": 1.018,
+      "step": 4551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.991865002376937e-06,
+      "loss": 0.8721,
+      "step": 4552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9880351789649153e-06,
+      "loss": 0.9316,
+      "step": 4553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984207377769771e-06,
+      "loss": 1.0603,
+      "step": 4554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9803815998954334e-06,
+      "loss": 0.9609,
+      "step": 4555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.976557846445225e-06,
+      "loss": 0.9345,
+      "step": 4556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9727361185219085e-06,
+      "loss": 1.008,
+      "step": 4557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9689164172276463e-06,
+      "loss": 1.0437,
+      "step": 4558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9650987436640243e-06,
+      "loss": 0.9511,
+      "step": 4559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9612830989320406e-06,
+      "loss": 1.0123,
+      "step": 4560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9574694841321082e-06,
+      "loss": 1.024,
+      "step": 4561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953657900364053e-06,
+      "loss": 0.9466,
+      "step": 4562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9498483487271225e-06,
+      "loss": 0.9319,
+      "step": 4563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9460408303199696e-06,
+      "loss": 0.9532,
+      "step": 4564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9422353462406663e-06,
+      "loss": 0.9917,
+      "step": 4565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9384318975866945e-06,
+      "loss": 1.0235,
+      "step": 4566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934630485454948e-06,
+      "loss": 0.9434,
+      "step": 4567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.930831110941743e-06,
+      "loss": 0.9711,
+      "step": 4568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9270337751427914e-06,
+      "loss": 0.9887,
+      "step": 4569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9232384791532377e-06,
+      "loss": 0.9949,
+      "step": 4570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.919445224067614e-06,
+      "loss": 0.959,
+      "step": 4571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.915654010979887e-06,
+      "loss": 0.9763,
+      "step": 4572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9118648409834205e-06,
+      "loss": 1.0834,
+      "step": 4573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9080777151709926e-06,
+      "loss": 0.96,
+      "step": 4574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9042926346347932e-06,
+      "loss": 1.0546,
+      "step": 4575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.900509600466418e-06,
+      "loss": 0.9957,
+      "step": 4576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8967286137568808e-06,
+      "loss": 0.9994,
+      "step": 4577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892949675596599e-06,
+      "loss": 0.9813,
+      "step": 4578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889172787075398e-06,
+      "loss": 0.9882,
+      "step": 4579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8853979492825156e-06,
+      "loss": 0.9932,
+      "step": 4580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8816251633065963e-06,
+      "loss": 0.9963,
+      "step": 4581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8778544302356904e-06,
+      "loss": 1.0202,
+      "step": 4582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8740857511572684e-06,
+      "loss": 0.9531,
+      "step": 4583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8703191271581875e-06,
+      "loss": 1.0229,
+      "step": 4584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866554559324731e-06,
+      "loss": 0.99,
+      "step": 4585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862792048742583e-06,
+      "loss": 0.959,
+      "step": 4586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8590315964968262e-06,
+      "loss": 0.9739,
+      "step": 4587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855273203671969e-06,
+      "loss": 1.0025,
+      "step": 4588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8515168713518995e-06,
+      "loss": 0.9996,
+      "step": 4589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8477626006199366e-06,
+      "loss": 0.9817,
+      "step": 4590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8440103925587904e-06,
+      "loss": 0.9768,
+      "step": 4591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.840260248250578e-06,
+      "loss": 1.0616,
+      "step": 4592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.836512168776826e-06,
+      "loss": 0.9337,
+      "step": 4593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8327661552184603e-06,
+      "loss": 1.0047,
+      "step": 4594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8290222086558106e-06,
+      "loss": 1.0172,
+      "step": 4595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825280330168619e-06,
+      "loss": 0.9622,
+      "step": 4596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8215405208360237e-06,
+      "loss": 0.9278,
+      "step": 4597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8178027817365648e-06,
+      "loss": 0.9943,
+      "step": 4598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.814067113948191e-06,
+      "loss": 1.0187,
+      "step": 4599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810333518548246e-06,
+      "loss": 0.9926,
+      "step": 4600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 0.9196,
+      "step": 4601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802872549220066e-06,
+      "loss": 1.0253,
+      "step": 4602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7991451774435385e-06,
+      "loss": 1.0023,
+      "step": 4603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7954198823588517e-06,
+      "loss": 0.991,
+      "step": 4604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7916966650403734e-06,
+      "loss": 0.96,
+      "step": 4605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7879755265618558e-06,
+      "loss": 1.0662,
+      "step": 4606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7842564679964567e-06,
+      "loss": 0.9733,
+      "step": 4607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.780539490416744e-06,
+      "loss": 0.9807,
+      "step": 4608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7768245948946615e-06,
+      "loss": 0.9793,
+      "step": 4609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7731117825015773e-06,
+      "loss": 0.9996,
+      "step": 4610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769401054308247e-06,
+      "loss": 0.9842,
+      "step": 4611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765692411384826e-06,
+      "loss": 0.9149,
+      "step": 4612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761985854800868e-06,
+      "loss": 1.063,
+      "step": 4613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7582813856253276e-06,
+      "loss": 0.9177,
+      "step": 4614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7545790049265506e-06,
+      "loss": 0.9618,
+      "step": 4615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750878713772295e-06,
+      "loss": 1.0196,
+      "step": 4616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7471805132297014e-06,
+      "loss": 0.9689,
+      "step": 4617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743484404365314e-06,
+      "loss": 0.9526,
+      "step": 4618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7397903882450728e-06,
+      "loss": 0.9464,
+      "step": 4619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7360984659343106e-06,
+      "loss": 1.0383,
+      "step": 4620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73240863849777e-06,
+      "loss": 1.0128,
+      "step": 4621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7287209069995667e-06,
+      "loss": 1.0132,
+      "step": 4622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725035272503238e-06,
+      "loss": 0.9744,
+      "step": 4623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7213517360716892e-06,
+      "loss": 1.0163,
+      "step": 4624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7176702987672444e-06,
+      "loss": 0.9821,
+      "step": 4625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7139909616516092e-06,
+      "loss": 0.9884,
+      "step": 4626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 1.0541,
+      "step": 4627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7066385922305715e-06,
+      "loss": 0.914,
+      "step": 4628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7029655620455606e-06,
+      "loss": 0.9475,
+      "step": 4629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699294636290134e-06,
+      "loss": 1.0165,
+      "step": 4630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69562581602297e-06,
+      "loss": 0.9856,
+      "step": 4631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6919591023021387e-06,
+      "loss": 0.9881,
+      "step": 4632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688294496185099e-06,
+      "loss": 0.937,
+      "step": 4633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684631998728715e-06,
+      "loss": 1.0107,
+      "step": 4634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6809716109892214e-06,
+      "loss": 0.9728,
+      "step": 4635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6773133340222677e-06,
+      "loss": 0.9988,
+      "step": 4636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6736571688828726e-06,
+      "loss": 1.0675,
+      "step": 4637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6700031166254658e-06,
+      "loss": 0.9435,
+      "step": 4638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6663511783038533e-06,
+      "loss": 0.9647,
+      "step": 4639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6627013549712355e-06,
+      "loss": 1.0419,
+      "step": 4640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659053647680212e-06,
+      "loss": 0.9798,
+      "step": 4641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655408057482751e-06,
+      "loss": 1.0027,
+      "step": 4642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6517645854302342e-06,
+      "loss": 0.9687,
+      "step": 4643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6481232325734173e-06,
+      "loss": 1.0198,
+      "step": 4644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6444839999624496e-06,
+      "loss": 0.9777,
+      "step": 4645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6408468886468677e-06,
+      "loss": 0.9337,
+      "step": 4646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6372118996755956e-06,
+      "loss": 1.0693,
+      "step": 4647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633579034096946e-06,
+      "loss": 0.9356,
+      "step": 4648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629948292958625e-06,
+      "loss": 1.0158,
+      "step": 4649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626319677307718e-06,
+      "loss": 0.9521,
+      "step": 4650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622693188190699e-06,
+      "loss": 1.0373,
+      "step": 4651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6190688266534315e-06,
+      "loss": 0.9658,
+      "step": 4652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615446593741161e-06,
+      "loss": 1.047,
+      "step": 4653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61182649049853e-06,
+      "loss": 1.0419,
+      "step": 4654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608208517969547e-06,
+      "loss": 0.982,
+      "step": 4655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6045926771976306e-06,
+      "loss": 0.8912,
+      "step": 4656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600978969225558e-06,
+      "loss": 1.0257,
+      "step": 4657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597367395095517e-06,
+      "loss": 1.0495,
+      "step": 4658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5937579558490634e-06,
+      "loss": 0.9783,
+      "step": 4659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5901506525271424e-06,
+      "loss": 1.0075,
+      "step": 4660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58654548617008e-06,
+      "loss": 1.0497,
+      "step": 4661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5829424578175944e-06,
+      "loss": 0.9224,
+      "step": 4662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5793415685087797e-06,
+      "loss": 0.9851,
+      "step": 4663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5757428192821134e-06,
+      "loss": 0.9713,
+      "step": 4664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572146211175458e-06,
+      "loss": 1.0688,
+      "step": 4665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.9648,
+      "step": 4666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5649594224705433e-06,
+      "loss": 0.9423,
+      "step": 4667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5613692439449145e-06,
+      "loss": 1.0272,
+      "step": 4668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55778121068457e-06,
+      "loss": 0.9931,
+      "step": 4669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5541953237242777e-06,
+      "loss": 0.9615,
+      "step": 4670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5506115840981905e-06,
+      "loss": 0.9716,
+      "step": 4671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5470299928398424e-06,
+      "loss": 1.0689,
+      "step": 4672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5434505509821426e-06,
+      "loss": 0.9742,
+      "step": 4673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.539873259557396e-06,
+      "loss": 0.9661,
+      "step": 4674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5362981195972627e-06,
+      "loss": 0.9952,
+      "step": 4675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5327251321328038e-06,
+      "loss": 0.9583,
+      "step": 4676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5291542981944507e-06,
+      "loss": 0.9582,
+      "step": 4677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.525585618812013e-06,
+      "loss": 0.944,
+      "step": 4678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.522019095014683e-06,
+      "loss": 1.0718,
+      "step": 4679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518454727831027e-06,
+      "loss": 0.9852,
+      "step": 4680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.9805,
+      "step": 4681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511332467415899e-06,
+      "loss": 0.9614,
+      "step": 4682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5077745762384542e-06,
+      "loss": 1.0185,
+      "step": 4683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5042188457827365e-06,
+      "loss": 0.8991,
+      "step": 4684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5006652770741977e-06,
+      "loss": 1.0924,
+      "step": 4685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4971138711376698e-06,
+      "loss": 0.9739,
+      "step": 4686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493564628997369e-06,
+      "loss": 0.9598,
+      "step": 4687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4900175516768675e-06,
+      "loss": 0.9714,
+      "step": 4688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486472640199138e-06,
+      "loss": 1.0476,
+      "step": 4689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4829298955865022e-06,
+      "loss": 1.003,
+      "step": 4690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479389318860682e-06,
+      "loss": 0.928,
+      "step": 4691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4758509110427576e-06,
+      "loss": 1.0102,
+      "step": 4692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472314673153188e-06,
+      "loss": 1.0385,
+      "step": 4693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4687806062118134e-06,
+      "loss": 1.0007,
+      "step": 4694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465248711237831e-06,
+      "loss": 0.9742,
+      "step": 4695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4617189892498326e-06,
+      "loss": 0.9956,
+      "step": 4696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4581914412657682e-06,
+      "loss": 0.9912,
+      "step": 4697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4546660683029656e-06,
+      "loss": 0.9752,
+      "step": 4698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451142871378124e-06,
+      "loss": 1.0148,
+      "step": 4699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4476218515073234e-06,
+      "loss": 1.0044,
+      "step": 4700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4441030097059992e-06,
+      "loss": 1.0094,
+      "step": 4701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4405863469889756e-06,
+      "loss": 1.0032,
+      "step": 4702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437071864370439e-06,
+      "loss": 1.1111,
+      "step": 4703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43355956286395e-06,
+      "loss": 0.9986,
+      "step": 4704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 0.9648,
+      "step": 4705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4265415072382016e-06,
+      "loss": 0.9414,
+      "step": 4706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4230357551429217e-06,
+      "loss": 1.0257,
+      "step": 4707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4195321882076295e-06,
+      "loss": 0.9503,
+      "step": 4708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4160308074427453e-06,
+      "loss": 1.0156,
+      "step": 4709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412531613858047e-06,
+      "loss": 1.028,
+      "step": 4710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4090346084626857e-06,
+      "loss": 1.021,
+      "step": 4711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40553979226518e-06,
+      "loss": 0.9634,
+      "step": 4712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4020471662734204e-06,
+      "loss": 1.0093,
+      "step": 4713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398556731494658e-06,
+      "loss": 0.9393,
+      "step": 4714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3950684889355235e-06,
+      "loss": 1.0323,
+      "step": 4715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391582439602007e-06,
+      "loss": 0.9891,
+      "step": 4716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3880985844994674e-06,
+      "loss": 1.0332,
+      "step": 4717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3846169246326345e-06,
+      "loss": 1.0374,
+      "step": 4718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381137461005596e-06,
+      "loss": 0.9574,
+      "step": 4719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3776601946218225e-06,
+      "loss": 1.0081,
+      "step": 4720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37418512648413e-06,
+      "loss": 1.008,
+      "step": 4721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3707122575947195e-06,
+      "loss": 0.9502,
+      "step": 4722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3672415889551457e-06,
+      "loss": 1.0049,
+      "step": 4723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3637731215663338e-06,
+      "loss": 0.9754,
+      "step": 4724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360306856428575e-06,
+      "loss": 0.9841,
+      "step": 4725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3568427945415163e-06,
+      "loss": 0.9841,
+      "step": 4726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3533809369041883e-06,
+      "loss": 0.954,
+      "step": 4727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349921284514961e-06,
+      "loss": 1.0236,
+      "step": 4728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346463838371591e-06,
+      "loss": 0.9774,
+      "step": 4729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3430085994711873e-06,
+      "loss": 1.0464,
+      "step": 4730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 1.003,
+      "step": 4731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336104747384532e-06,
+      "loss": 1.0046,
+      "step": 4732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3326561361893206e-06,
+      "loss": 0.8957,
+      "step": 4733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329209736219146e-06,
+      "loss": 1.0326,
+      "step": 4734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3257655484679376e-06,
+      "loss": 0.9835,
+      "step": 4735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322323573928983e-06,
+      "loss": 1.0278,
+      "step": 4736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3188838135949254e-06,
+      "loss": 0.9727,
+      "step": 4737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315446268457786e-06,
+      "loss": 0.9235,
+      "step": 4738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312010939508923e-06,
+      "loss": 0.9922,
+      "step": 4739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3085778277390802e-06,
+      "loss": 1.016,
+      "step": 4740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3051469341383403e-06,
+      "loss": 1.0498,
+      "step": 4741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301718259696165e-06,
+      "loss": 0.9856,
+      "step": 4742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2982918054013635e-06,
+      "loss": 0.9495,
+      "step": 4743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2948675722421086e-06,
+      "loss": 1.0278,
+      "step": 4744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291445561205934e-06,
+      "loss": 0.9646,
+      "step": 4745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2880257732797286e-06,
+      "loss": 0.996,
+      "step": 4746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284608209449746e-06,
+      "loss": 0.9466,
+      "step": 4747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281192870701594e-06,
+      "loss": 1.0058,
+      "step": 4748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2777797580202408e-06,
+      "loss": 0.9491,
+      "step": 4749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274368872390009e-06,
+      "loss": 1.0065,
+      "step": 4750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270960214794584e-06,
+      "loss": 0.9427,
+      "step": 4751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2675537862170026e-06,
+      "loss": 0.9899,
+      "step": 4752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2641495876396713e-06,
+      "loss": 0.9756,
+      "step": 4753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260747620044332e-06,
+      "loss": 0.9739,
+      "step": 4754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2573478844121057e-06,
+      "loss": 1.0004,
+      "step": 4755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2539503817234553e-06,
+      "loss": 1.028,
+      "step": 4756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2505551129582047e-06,
+      "loss": 0.936,
+      "step": 4757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247162079095535e-06,
+      "loss": 0.9879,
+      "step": 4758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2437712811139767e-06,
+      "loss": 1.0045,
+      "step": 4759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2403827199914265e-06,
+      "loss": 0.9695,
+      "step": 4760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23699639670512e-06,
+      "loss": 1.0166,
+      "step": 4761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2336123122316642e-06,
+      "loss": 1.0169,
+      "step": 4762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23023046754701e-06,
+      "loss": 0.9628,
+      "step": 4763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226850863626465e-06,
+      "loss": 0.9879,
+      "step": 4764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2234735014446905e-06,
+      "loss": 1.0133,
+      "step": 4765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220098381975703e-06,
+      "loss": 0.9593,
+      "step": 4766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216725506192865e-06,
+      "loss": 0.9381,
+      "step": 4767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213354875068906e-06,
+      "loss": 1.0049,
+      "step": 4768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2099864895758948e-06,
+      "loss": 1.0112,
+      "step": 4769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206620350685257e-06,
+      "loss": 0.9384,
+      "step": 4770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2032564593677773e-06,
+      "loss": 0.974,
+      "step": 4771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1998948165935763e-06,
+      "loss": 1.015,
+      "step": 4772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1965354233321444e-06,
+      "loss": 0.9683,
+      "step": 4773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193178280552306e-06,
+      "loss": 0.9754,
+      "step": 4774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189823389222251e-06,
+      "loss": 0.978,
+      "step": 4775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1864707503095128e-06,
+      "loss": 0.9934,
+      "step": 4776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183120364780975e-06,
+      "loss": 0.9904,
+      "step": 4777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1797722336028726e-06,
+      "loss": 0.9829,
+      "step": 4778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1764263577407894e-06,
+      "loss": 1.0433,
+      "step": 4779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1730827381596643e-06,
+      "loss": 0.9967,
+      "step": 4780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1697413758237785e-06,
+      "loss": 0.9738,
+      "step": 4781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166402271696764e-06,
+      "loss": 0.9896,
+      "step": 4782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 0.981,
+      "step": 4783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1597308419206264e-06,
+      "loss": 1.0263,
+      "step": 4784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1563985181955085e-06,
+      "loss": 0.9546,
+      "step": 4785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153068456527283e-06,
+      "loss": 0.9904,
+      "step": 4786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149740657876315e-06,
+      "loss": 0.9859,
+      "step": 4787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146415123202331e-06,
+      "loss": 0.933,
+      "step": 4788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1430918534643996e-06,
+      "loss": 1.0559,
+      "step": 4789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13977084962093e-06,
+      "loss": 0.9668,
+      "step": 4790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1364521126296933e-06,
+      "loss": 1.0131,
+      "step": 4791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1331356434477867e-06,
+      "loss": 0.9928,
+      "step": 4792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1298214430316743e-06,
+      "loss": 1.0291,
+      "step": 4793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126509512337145e-06,
+      "loss": 0.9812,
+      "step": 4794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.123199852319352e-06,
+      "loss": 1.0222,
+      "step": 4795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119892463932781e-06,
+      "loss": 1.0,
+      "step": 4796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1165873481312694e-06,
+      "loss": 1.0063,
+      "step": 4797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1132845058679942e-06,
+      "loss": 1.0291,
+      "step": 4798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109983938095479e-06,
+      "loss": 0.9379,
+      "step": 4799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106685645765595e-06,
+      "loss": 1.0153,
+      "step": 4800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103389629829551e-06,
+      "loss": 0.9885,
+      "step": 4801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1000958912379032e-06,
+      "loss": 0.936,
+      "step": 4802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096804430940551e-06,
+      "loss": 1.0314,
+      "step": 4803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0935152498867326e-06,
+      "loss": 0.9295,
+      "step": 4804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090228349025032e-06,
+      "loss": 1.043,
+      "step": 4805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0869437293033835e-06,
+      "loss": 0.9168,
+      "step": 4806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083661391669043e-06,
+      "loss": 1.0039,
+      "step": 4807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08038133706863e-06,
+      "loss": 0.9409,
+      "step": 4808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0771035664480944e-06,
+      "loss": 1.0306,
+      "step": 4809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073828080752728e-06,
+      "loss": 0.9351,
+      "step": 4810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070554880927166e-06,
+      "loss": 1.0684,
+      "step": 4811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067283967915381e-06,
+      "loss": 0.9428,
+      "step": 4812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0640153426606933e-06,
+      "loss": 1.0059,
+      "step": 4813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.060749006105757e-06,
+      "loss": 1.0274,
+      "step": 4814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0574849591925674e-06,
+      "loss": 0.9555,
+      "step": 4815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0542232028624585e-06,
+      "loss": 0.9933,
+      "step": 4816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0509637380561065e-06,
+      "loss": 0.9593,
+      "step": 4817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047706565713523e-06,
+      "loss": 0.9302,
+      "step": 4818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044451686774068e-06,
+      "loss": 0.9734,
+      "step": 4819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041199102176422e-06,
+      "loss": 0.9922,
+      "step": 4820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0379488128586246e-06,
+      "loss": 0.9893,
+      "step": 4821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0347008197580376e-06,
+      "loss": 1.0152,
+      "step": 4822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0314551238113677e-06,
+      "loss": 1.0431,
+      "step": 4823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.028211725954663e-06,
+      "loss": 0.9717,
+      "step": 4824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024970627123295e-06,
+      "loss": 0.9713,
+      "step": 4825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0217318282519905e-06,
+      "loss": 0.9273,
+      "step": 4826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018495330274793e-06,
+      "loss": 1.0034,
+      "step": 4827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0152611341251015e-06,
+      "loss": 0.9984,
+      "step": 4828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0120292407356402e-06,
+      "loss": 0.9888,
+      "step": 4829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00879965103847e-06,
+      "loss": 0.9442,
+      "step": 4830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0055723659649907e-06,
+      "loss": 1.0513,
+      "step": 4831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002347386445932e-06,
+      "loss": 0.9812,
+      "step": 4832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9991247134113677e-06,
+      "loss": 0.989,
+      "step": 4833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9959043477907e-06,
+      "loss": 0.9826,
+      "step": 4834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9926862905126663e-06,
+      "loss": 0.9893,
+      "step": 4835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989470542505337e-06,
+      "loss": 0.9093,
+      "step": 4836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986257104696121e-06,
+      "loss": 0.9951,
+      "step": 4837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983045978011754e-06,
+      "loss": 1.0035,
+      "step": 4838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9798371633783174e-06,
+      "loss": 0.9507,
+      "step": 4839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976630661721207e-06,
+      "loss": 1.0277,
+      "step": 4840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734264739651723e-06,
+      "loss": 0.9711,
+      "step": 4841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970224601034282e-06,
+      "loss": 1.0183,
+      "step": 4842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967025043851939e-06,
+      "loss": 0.9973,
+      "step": 4843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9638278033408832e-06,
+      "loss": 0.9676,
+      "step": 4844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960632880423178e-06,
+      "loss": 1.0219,
+      "step": 4845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9574402760202315e-06,
+      "loss": 0.9653,
+      "step": 4846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95424999105277e-06,
+      "loss": 0.9306,
+      "step": 4847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95106202644086e-06,
+      "loss": 1.0933,
+      "step": 4848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9478763831038916e-06,
+      "loss": 0.941,
+      "step": 4849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9446930619605908e-06,
+      "loss": 0.996,
+      "step": 4850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9415120639290086e-06,
+      "loss": 1.0015,
+      "step": 4851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9383333899265368e-06,
+      "loss": 1.0381,
+      "step": 4852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.935157040869884e-06,
+      "loss": 0.8954,
+      "step": 4853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9319830176750963e-06,
+      "loss": 0.9573,
+      "step": 4854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9288113212575454e-06,
+      "loss": 1.0052,
+      "step": 4855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9256419525319316e-06,
+      "loss": 1.0661,
+      "step": 4856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9224749124122923e-06,
+      "loss": 0.9844,
+      "step": 4857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919310201811977e-06,
+      "loss": 0.9505,
+      "step": 4858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9161478216436835e-06,
+      "loss": 1.0367,
+      "step": 4859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912987772819417e-06,
+      "loss": 1.0295,
+      "step": 4860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.9993,
+      "step": 4861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9066746728476815e-06,
+      "loss": 1.0122,
+      "step": 4862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9035216235208775e-06,
+      "loss": 0.9792,
+      "step": 4863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9003709091794408e-06,
+      "loss": 0.9152,
+      "step": 4864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897222530732018e-06,
+      "loss": 0.9659,
+      "step": 4865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8940764890865915e-06,
+      "loss": 0.9604,
+      "step": 4866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8909327851504633e-06,
+      "loss": 0.9887,
+      "step": 4867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8877914198302628e-06,
+      "loss": 0.9694,
+      "step": 4868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8846523940319416e-06,
+      "loss": 0.9675,
+      "step": 4869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8815157086607826e-06,
+      "loss": 1.013,
+      "step": 4870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8783813646213866e-06,
+      "loss": 1.0412,
+      "step": 4871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8752493628176925e-06,
+      "loss": 0.9251,
+      "step": 4872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8721197041529425e-06,
+      "loss": 1.0565,
+      "step": 4873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8689923895297247e-06,
+      "loss": 1.0129,
+      "step": 4874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865867419849937e-06,
+      "loss": 0.9646,
+      "step": 4875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8627447960148036e-06,
+      "loss": 1.0336,
+      "step": 4876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8596245189248828e-06,
+      "loss": 0.9964,
+      "step": 4877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.856506589480037e-06,
+      "loss": 0.9491,
+      "step": 4878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8533910085794714e-06,
+      "loss": 0.9001,
+      "step": 4879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8502777771216962e-06,
+      "loss": 1.0698,
+      "step": 4880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8471668960045575e-06,
+      "loss": 0.9491,
+      "step": 4881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8440583661252165e-06,
+      "loss": 0.9432,
+      "step": 4882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8409521883801596e-06,
+      "loss": 1.0191,
+      "step": 4883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.83784836366519e-06,
+      "loss": 1.0542,
+      "step": 4884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8347468928754408e-06,
+      "loss": 0.9237,
+      "step": 4885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8316477769053597e-06,
+      "loss": 1.0286,
+      "step": 4886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8285510166487154e-06,
+      "loss": 1.0494,
+      "step": 4887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8254566129985996e-06,
+      "loss": 0.9836,
+      "step": 4888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82236456684742e-06,
+      "loss": 0.9642,
+      "step": 4889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8192748790869163e-06,
+      "loss": 1.0161,
+      "step": 4890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8161875506081294e-06,
+      "loss": 1.0005,
+      "step": 4891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8131025823014403e-06,
+      "loss": 0.9757,
+      "step": 4892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8100199750565273e-06,
+      "loss": 0.9723,
+      "step": 4893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806939729762409e-06,
+      "loss": 0.9887,
+      "step": 4894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8038618473074088e-06,
+      "loss": 1.0301,
+      "step": 4895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.800786328579176e-06,
+      "loss": 0.9702,
+      "step": 4896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7977131744646724e-06,
+      "loss": 1.0466,
+      "step": 4897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7946423858501793e-06,
+      "loss": 0.9793,
+      "step": 4898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7915739636213036e-06,
+      "loss": 1.0277,
+      "step": 4899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7885079086629598e-06,
+      "loss": 0.9524,
+      "step": 4900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7854442218593836e-06,
+      "loss": 0.9646,
+      "step": 4901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7823829040941288e-06,
+      "loss": 0.992,
+      "step": 4902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7793239562500618e-06,
+      "loss": 0.9532,
+      "step": 4903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7762673792093687e-06,
+      "loss": 1.0085,
+      "step": 4904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.773213173853554e-06,
+      "loss": 0.9939,
+      "step": 4905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.971,
+      "step": 4906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767111881719148e-06,
+      "loss": 1.0107,
+      "step": 4907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.764064796700138e-06,
+      "loss": 1.0257,
+      "step": 4908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761020086885169e-06,
+      "loss": 0.9698,
+      "step": 4909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7579777531523279e-06,
+      "loss": 1.0289,
+      "step": 4910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7549377963789994e-06,
+      "loss": 0.9958,
+      "step": 4911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751900217441902e-06,
+      "loss": 0.919,
+      "step": 4912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7488650172170496e-06,
+      "loss": 0.975,
+      "step": 4913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7458321965797854e-06,
+      "loss": 0.966,
+      "step": 4914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7428017564047594e-06,
+      "loss": 1.0297,
+      "step": 4915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7397736975659351e-06,
+      "loss": 0.9313,
+      "step": 4916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7367480209365871e-06,
+      "loss": 0.9983,
+      "step": 4917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7337247273893131e-06,
+      "loss": 1.0595,
+      "step": 4918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730703817796011e-06,
+      "loss": 0.967,
+      "step": 4919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7276852930278986e-06,
+      "loss": 0.9903,
+      "step": 4920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7246691539555027e-06,
+      "loss": 1.0392,
+      "step": 4921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.721655401448662e-06,
+      "loss": 0.965,
+      "step": 4922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7186440363765356e-06,
+      "loss": 0.9675,
+      "step": 4923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7156350596075743e-06,
+      "loss": 1.0205,
+      "step": 4924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7126284720095654e-06,
+      "loss": 0.9418,
+      "step": 4925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709624274449584e-06,
+      "loss": 1.0058,
+      "step": 4926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066224677940313e-06,
+      "loss": 1.0093,
+      "step": 4927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703623052908614e-06,
+      "loss": 0.9901,
+      "step": 4928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7006260306583456e-06,
+      "loss": 0.9429,
+      "step": 4929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697631401907559e-06,
+      "loss": 0.8983,
+      "step": 4930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6946391675198838e-06,
+      "loss": 1.0137,
+      "step": 4931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6916493283582702e-06,
+      "loss": 1.0142,
+      "step": 4932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6886618852849723e-06,
+      "loss": 0.9856,
+      "step": 4933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6856768391615541e-06,
+      "loss": 0.9334,
+      "step": 4934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826941908488891e-06,
+      "loss": 1.0875,
+      "step": 4935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6797139412071583e-06,
+      "loss": 0.9424,
+      "step": 4936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6767360910958497e-06,
+      "loss": 0.9565,
+      "step": 4937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673760641373764e-06,
+      "loss": 1.0481,
+      "step": 4938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 1.0045,
+      "step": 4939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6678169465289863e-06,
+      "loss": 0.9864,
+      "step": 4940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6648487031204274e-06,
+      "loss": 0.9252,
+      "step": 4941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6618828635293538e-06,
+      "loss": 1.0626,
+      "step": 4942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6589194286111043e-06,
+      "loss": 0.947,
+      "step": 4943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655958399220312e-06,
+      "loss": 0.9539,
+      "step": 4944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6529997762109319e-06,
+      "loss": 0.987,
+      "step": 4945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6500435604362075e-06,
+      "loss": 1.0273,
+      "step": 4946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.647089752748704e-06,
+      "loss": 1.0321,
+      "step": 4947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6441383540002842e-06,
+      "loss": 0.965,
+      "step": 4948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641189365042115e-06,
+      "loss": 1.0025,
+      "step": 4949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6382427867246719e-06,
+      "loss": 1.0093,
+      "step": 4950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6352986198977327e-06,
+      "loss": 0.9417,
+      "step": 4951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6323568654103838e-06,
+      "loss": 1.024,
+      "step": 4952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6294175241110111e-06,
+      "loss": 0.9818,
+      "step": 4953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6264805968473063e-06,
+      "loss": 0.9713,
+      "step": 4954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6235460844662642e-06,
+      "loss": 0.9091,
+      "step": 4955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620613987814189e-06,
+      "loss": 1.0158,
+      "step": 4956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6176843077366755e-06,
+      "loss": 0.9911,
+      "step": 4957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6147570450786387e-06,
+      "loss": 0.9733,
+      "step": 4958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6118322006842757e-06,
+      "loss": 0.9878,
+      "step": 4959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6089097753971061e-06,
+      "loss": 1.0004,
+      "step": 4960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6059897700599415e-06,
+      "loss": 0.9231,
+      "step": 4961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6030721855148935e-06,
+      "loss": 0.9428,
+      "step": 4962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600157022603388e-06,
+      "loss": 0.9797,
+      "step": 4963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5972442821661337e-06,
+      "loss": 1.0006,
+      "step": 4964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5943339650431578e-06,
+      "loss": 0.9521,
+      "step": 4965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5914260720737796e-06,
+      "loss": 1.0434,
+      "step": 4966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5885206040966216e-06,
+      "loss": 0.9821,
+      "step": 4967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585617561949606e-06,
+      "loss": 0.9761,
+      "step": 4968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5827169464699576e-06,
+      "loss": 1.0069,
+      "step": 4969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579818758494197e-06,
+      "loss": 1.0073,
+      "step": 4970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769229988581514e-06,
+      "loss": 0.9838,
+      "step": 4971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5740296683969424e-06,
+      "loss": 0.9827,
+      "step": 4972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571138767944993e-06,
+      "loss": 0.939,
+      "step": 4973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5682502983360247e-06,
+      "loss": 0.9412,
+      "step": 4974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565364260403055e-06,
+      "loss": 1.0168,
+      "step": 4975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5624806549784111e-06,
+      "loss": 0.939,
+      "step": 4976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5595994828937012e-06,
+      "loss": 1.0719,
+      "step": 4977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5567207449798517e-06,
+      "loss": 0.9404,
+      "step": 4978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5538444420670663e-06,
+      "loss": 0.9963,
+      "step": 4979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5509705749848647e-06,
+      "loss": 1.0112,
+      "step": 4980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5480991445620541e-06,
+      "loss": 1.0051,
+      "step": 4981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5452301516267377e-06,
+      "loss": 0.9595,
+      "step": 4982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5423635970063267e-06,
+      "loss": 1.0263,
+      "step": 4983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5394994815275132e-06,
+      "loss": 1.0012,
+      "step": 4984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.536637806016301e-06,
+      "loss": 0.9895,
+      "step": 4985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5337785712979802e-06,
+      "loss": 0.9598,
+      "step": 4986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5309217781971419e-06,
+      "loss": 0.9743,
+      "step": 4987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5280674275376696e-06,
+      "loss": 0.9966,
+      "step": 4988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5252155201427454e-06,
+      "loss": 0.8882,
+      "step": 4989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.522366056834844e-06,
+      "loss": 1.0003,
+      "step": 4990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 1.0298,
+      "step": 4991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.516674465766499e-06,
+      "loss": 0.9423,
+      "step": 4992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5138323396474808e-06,
+      "loss": 0.9846,
+      "step": 4993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5109926608983416e-06,
+      "loss": 1.0208,
+      "step": 4994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5081554303380285e-06,
+      "loss": 0.9115,
+      "step": 4995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5053206487847916e-06,
+      "loss": 1.029,
+      "step": 4996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5024883170561599e-06,
+      "loss": 1.0322,
+      "step": 4997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4996584359689702e-06,
+      "loss": 0.9809,
+      "step": 4998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4968310063393454e-06,
+      "loss": 0.9512,
+      "step": 4999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940060289827019e-06,
+      "loss": 0.9438,
+      "step": 5000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49118350471375e-06,
+      "loss": 1.0893,
+      "step": 5001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4883634343464915e-06,
+      "loss": 1.002,
+      "step": 5002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855458186942184e-06,
+      "loss": 1.0063,
+      "step": 5003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4827306585695234e-06,
+      "loss": 0.9666,
+      "step": 5004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4799179547842823e-06,
+      "loss": 0.9647,
+      "step": 5005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4771077081496654e-06,
+      "loss": 0.9993,
+      "step": 5006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4742999194761342e-06,
+      "loss": 0.9375,
+      "step": 5007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4714945895734379e-06,
+      "loss": 0.9527,
+      "step": 5008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468691719250629e-06,
+      "loss": 1.0531,
+      "step": 5009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4658913093160322e-06,
+      "loss": 0.9324,
+      "step": 5010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4630933605772801e-06,
+      "loss": 0.9926,
+      "step": 5011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.46029787384128e-06,
+      "loss": 0.993,
+      "step": 5012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457504849914242e-06,
+      "loss": 1.002,
+      "step": 5013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454714289601661e-06,
+      "loss": 1.0245,
+      "step": 5014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4519261937083161e-06,
+      "loss": 1.0072,
+      "step": 5015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491405630382894e-06,
+      "loss": 0.9821,
+      "step": 5016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 0.9594,
+      "step": 5017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435767005809077e-06,
+      "loss": 1.0051,
+      "step": 5018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.440798470398147e-06,
+      "loss": 1.0259,
+      "step": 5019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4380227086478816e-06,
+      "loss": 0.9312,
+      "step": 5020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4352494161306285e-06,
+      "loss": 0.9718,
+      "step": 5021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4324785936461894e-06,
+      "loss": 1.0123,
+      "step": 5022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4297102419936559e-06,
+      "loss": 0.9548,
+      "step": 5023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4269443619714118e-06,
+      "loss": 0.9676,
+      "step": 5024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.42418095437712e-06,
+      "loss": 1.0127,
+      "step": 5025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4214200200077343e-06,
+      "loss": 0.9922,
+      "step": 5026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4186615596594965e-06,
+      "loss": 0.9936,
+      "step": 5027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4159055741279282e-06,
+      "loss": 0.9783,
+      "step": 5028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4131520642078522e-06,
+      "loss": 1.0218,
+      "step": 5029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4104010306933558e-06,
+      "loss": 0.9734,
+      "step": 5030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407652474377832e-06,
+      "loss": 0.9409,
+      "step": 5031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4049063960539488e-06,
+      "loss": 1.024,
+      "step": 5032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402162796513662e-06,
+      "loss": 0.9297,
+      "step": 5033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3994216765482126e-06,
+      "loss": 0.9954,
+      "step": 5034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3966830369481231e-06,
+      "loss": 0.9148,
+      "step": 5035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3939468785032107e-06,
+      "loss": 1.0234,
+      "step": 5036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3912132020025648e-06,
+      "loss": 0.934,
+      "step": 5037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3884820082345674e-06,
+      "loss": 0.9603,
+      "step": 5038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3857532979868804e-06,
+      "loss": 0.9834,
+      "step": 5039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3830270720464512e-06,
+      "loss": 0.9841,
+      "step": 5040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3803033311995072e-06,
+      "loss": 0.9229,
+      "step": 5041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3775820762315694e-06,
+      "loss": 1.0878,
+      "step": 5042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3748633079274254e-06,
+      "loss": 0.9534,
+      "step": 5043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3721470270711624e-06,
+      "loss": 0.9789,
+      "step": 5044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3694332344461393e-06,
+      "loss": 0.9518,
+      "step": 5045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366721930835001e-06,
+      "loss": 0.9491,
+      "step": 5046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3640131170196758e-06,
+      "loss": 1.0074,
+      "step": 5047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3613067937813684e-06,
+      "loss": 0.9873,
+      "step": 5048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358602961900578e-06,
+      "loss": 0.9949,
+      "step": 5049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3559016221570663e-06,
+      "loss": 1.0317,
+      "step": 5050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3532027753298937e-06,
+      "loss": 0.8933,
+      "step": 5051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350506422197393e-06,
+      "loss": 0.9377,
+      "step": 5052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3478125635371785e-06,
+      "loss": 1.0046,
+      "step": 5053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.345121200126146e-06,
+      "loss": 0.997,
+      "step": 5054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342432332740473e-06,
+      "loss": 0.978,
+      "step": 5055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 1.0188,
+      "step": 5056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.337062089146307e-06,
+      "loss": 1.0244,
+      "step": 5057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3343807144865683e-06,
+      "loss": 0.9734,
+      "step": 5058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3317018389496927e-06,
+      "loss": 1.0022,
+      "step": 5059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3290254633082554e-06,
+      "loss": 1.0358,
+      "step": 5060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3263515883341071e-06,
+      "loss": 0.9497,
+      "step": 5061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3236802147983874e-06,
+      "loss": 0.989,
+      "step": 5062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3210113434715e-06,
+      "loss": 0.944,
+      "step": 5063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3183449751231392e-06,
+      "loss": 1.0005,
+      "step": 5064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3156811105222723e-06,
+      "loss": 0.9781,
+      "step": 5065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3130197504371434e-06,
+      "loss": 0.9768,
+      "step": 5066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3103608956352765e-06,
+      "loss": 1.0149,
+      "step": 5067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3077045468834714e-06,
+      "loss": 0.9888,
+      "step": 5068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30505070494781e-06,
+      "loss": 0.9842,
+      "step": 5069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3023993705936444e-06,
+      "loss": 1.037,
+      "step": 5070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2997505445856085e-06,
+      "loss": 0.9003,
+      "step": 5071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2971042276876089e-06,
+      "loss": 1.0226,
+      "step": 5072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.294460420662832e-06,
+      "loss": 1.0055,
+      "step": 5073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2918191242737365e-06,
+      "loss": 0.9819,
+      "step": 5074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2891803392820669e-06,
+      "loss": 0.9803,
+      "step": 5075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2865440664488248e-06,
+      "loss": 0.9657,
+      "step": 5076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2839103065343084e-06,
+      "loss": 0.9625,
+      "step": 5077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2812790602980774e-06,
+      "loss": 1.0041,
+      "step": 5078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2786503284989703e-06,
+      "loss": 1.0003,
+      "step": 5079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276024111895101e-06,
+      "loss": 0.9021,
+      "step": 5080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273400411243857e-06,
+      "loss": 1.0665,
+      "step": 5081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2707792273019049e-06,
+      "loss": 0.9919,
+      "step": 5082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2681605608251745e-06,
+      "loss": 0.7878,
+      "step": 5083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2655444125688832e-06,
+      "loss": 1.0508,
+      "step": 5084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2629307832875126e-06,
+      "loss": 0.9697,
+      "step": 5085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603196737348211e-06,
+      "loss": 0.9844,
+      "step": 5086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2577110846638407e-06,
+      "loss": 0.9938,
+      "step": 5087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255105016826873e-06,
+      "loss": 1.0062,
+      "step": 5088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252501470975499e-06,
+      "loss": 0.945,
+      "step": 5089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249900447860568e-06,
+      "loss": 0.9892,
+      "step": 5090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2473019482322024e-06,
+      "loss": 0.959,
+      "step": 5091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244705972839797e-06,
+      "loss": 1.0046,
+      "step": 5092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2421125224320186e-06,
+      "loss": 0.9559,
+      "step": 5093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.239521597756802e-06,
+      "loss": 0.9867,
+      "step": 5094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 1.0375,
+      "step": 5095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2343473285921825e-06,
+      "loss": 0.9753,
+      "step": 5096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2317639855950113e-06,
+      "loss": 0.9565,
+      "step": 5097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2291831713148727e-06,
+      "loss": 1.0426,
+      "step": 5098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2266048864960633e-06,
+      "loss": 0.9725,
+      "step": 5099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2240291318821462e-06,
+      "loss": 0.9775,
+      "step": 5100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2214559082159538e-06,
+      "loss": 0.9723,
+      "step": 5101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2188852162395981e-06,
+      "loss": 0.9902,
+      "step": 5102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2163170566944505e-06,
+      "loss": 0.9646,
+      "step": 5103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213751430321156e-06,
+      "loss": 0.9415,
+      "step": 5104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2111883378596301e-06,
+      "loss": 1.0287,
+      "step": 5105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2086277800490554e-06,
+      "loss": 0.9518,
+      "step": 5106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2060697576278812e-06,
+      "loss": 1.0085,
+      "step": 5107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2035142713338366e-06,
+      "loss": 0.9494,
+      "step": 5108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.200961321903903e-06,
+      "loss": 1.0321,
+      "step": 5109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1984109100743445e-06,
+      "loss": 0.8976,
+      "step": 5110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1958630365806866e-06,
+      "loss": 0.9804,
+      "step": 5111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1933177021577202e-06,
+      "loss": 1.0426,
+      "step": 5112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1907749075395147e-06,
+      "loss": 1.0314,
+      "step": 5113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1882346534593904e-06,
+      "loss": 0.887,
+      "step": 5114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185696940649954e-06,
+      "loss": 1.011,
+      "step": 5115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1831617698430609e-06,
+      "loss": 0.9664,
+      "step": 5116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180629141769849e-06,
+      "loss": 0.961,
+      "step": 5117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.178099057160712e-06,
+      "loss": 1.0235,
+      "step": 5118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1755715167453151e-06,
+      "loss": 1.0657,
+      "step": 5119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1730465212525898e-06,
+      "loss": 0.9501,
+      "step": 5120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.9437,
+      "step": 5121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168004167947202e-06,
+      "loss": 0.9717,
+      "step": 5122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.165486811588732e-06,
+      "loss": 0.9908,
+      "step": 5123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1629720030613124e-06,
+      "loss": 1.0028,
+      "step": 5124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1604597430902032e-06,
+      "loss": 0.9702,
+      "step": 5125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157950032399927e-06,
+      "loss": 0.9906,
+      "step": 5126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1554428717142707e-06,
+      "loss": 1.0039,
+      "step": 5127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1529382617562945e-06,
+      "loss": 0.9907,
+      "step": 5128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1504362032483064e-06,
+      "loss": 1.0494,
+      "step": 5129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1479366969118943e-06,
+      "loss": 1.0165,
+      "step": 5130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 0.9879,
+      "step": 5131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1429453436364391e-06,
+      "loss": 0.9362,
+      "step": 5132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1404534981368776e-06,
+      "loss": 1.0393,
+      "step": 5133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1379642076878528e-06,
+      "loss": 0.9538,
+      "step": 5134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1354774730072693e-06,
+      "loss": 0.9503,
+      "step": 5135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1329932948122824e-06,
+      "loss": 0.9548,
+      "step": 5136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1305116738193211e-06,
+      "loss": 1.0275,
+      "step": 5137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.128032610744073e-06,
+      "loss": 0.9509,
+      "step": 5138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1255561063014875e-06,
+      "loss": 0.9513,
+      "step": 5139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.123082161205775e-06,
+      "loss": 1.0492,
+      "step": 5140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1206107761704133e-06,
+      "loss": 0.9644,
+      "step": 5141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1181419519081315e-06,
+      "loss": 0.9836,
+      "step": 5142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1156756891309328e-06,
+      "loss": 1.0476,
+      "step": 5143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1132119885500736e-06,
+      "loss": 0.9446,
+      "step": 5144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1107508508760701e-06,
+      "loss": 0.9826,
+      "step": 5145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1082922768187098e-06,
+      "loss": 0.9742,
+      "step": 5146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1058362670870248e-06,
+      "loss": 0.9689,
+      "step": 5147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103382822389324e-06,
+      "loss": 0.9726,
+      "step": 5148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1009319434331623e-06,
+      "loss": 1.0064,
+      "step": 5149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0984836309253666e-06,
+      "loss": 0.9347,
+      "step": 5150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0960378855720177e-06,
+      "loss": 1.0248,
+      "step": 5151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0935947080784548e-06,
+      "loss": 1.0337,
+      "step": 5152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0911540991492797e-06,
+      "loss": 0.9684,
+      "step": 5153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0887160594883507e-06,
+      "loss": 0.96,
+      "step": 5154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862805897987894e-06,
+      "loss": 1.0177,
+      "step": 5155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083847690782972e-06,
+      "loss": 0.9448,
+      "step": 5156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.081417363142535e-06,
+      "loss": 1.002,
+      "step": 5157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0789896075783734e-06,
+      "loss": 0.9845,
+      "step": 5158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0765644247906404e-06,
+      "loss": 0.9221,
+      "step": 5159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0741418154787443e-06,
+      "loss": 0.9724,
+      "step": 5160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0717217803413605e-06,
+      "loss": 0.9972,
+      "step": 5161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0693043200764075e-06,
+      "loss": 0.9782,
+      "step": 5162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0668894353810744e-06,
+      "loss": 1.0451,
+      "step": 5163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0644771269518018e-06,
+      "loss": 1.0188,
+      "step": 5164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0620673954842843e-06,
+      "loss": 0.9424,
+      "step": 5165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0596602416734836e-06,
+      "loss": 0.9597,
+      "step": 5166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0572556662136036e-06,
+      "loss": 0.9578,
+      "step": 5167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0548536697981194e-06,
+      "loss": 0.9969,
+      "step": 5168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0524542531197468e-06,
+      "loss": 0.964,
+      "step": 5169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0500574168704746e-06,
+      "loss": 0.9686,
+      "step": 5170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476631617415333e-06,
+      "loss": 1.0484,
+      "step": 5171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0452714884234173e-06,
+      "loss": 0.9777,
+      "step": 5172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 1.0238,
+      "step": 5173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0404958899778994e-06,
+      "loss": 1.0006,
+      "step": 5174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0381119662277594e-06,
+      "loss": 0.9823,
+      "step": 5175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0357306270429623e-06,
+      "loss": 0.9745,
+      "step": 5176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0333518731102754e-06,
+      "loss": 1.0156,
+      "step": 5177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0309757051157176e-06,
+      "loss": 0.9993,
+      "step": 5178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0286021237445699e-06,
+      "loss": 0.9293,
+      "step": 5179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0262311296813543e-06,
+      "loss": 1.0456,
+      "step": 5180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0238627236098619e-06,
+      "loss": 0.9266,
+      "step": 5181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0214969062131208e-06,
+      "loss": 1.063,
+      "step": 5182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0191336781734284e-06,
+      "loss": 0.9331,
+      "step": 5183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167730401723263e-06,
+      "loss": 0.9596,
+      "step": 5184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.014414992890611e-06,
+      "loss": 1.0276,
+      "step": 5185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012059537008332e-06,
+      "loss": 0.9842,
+      "step": 5186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097066732047878e-06,
+      "loss": 0.9696,
+      "step": 5187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007356402158539e-06,
+      "loss": 1.0442,
+      "step": 5188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0050087245473906e-06,
+      "loss": 0.9715,
+      "step": 5189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0026636410484004e-06,
+      "loss": 0.9933,
+      "step": 5190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0003211523378798e-06,
+      "loss": 0.9432,
+      "step": 5191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.979812590913907e-07,
+      "loss": 1.0678,
+      "step": 5192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.956439619837455e-07,
+      "loss": 0.9172,
+      "step": 5193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93309261689015e-07,
+      "loss": 0.9329,
+      "step": 5194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.909771588805073e-07,
+      "loss": 1.0023,
+      "step": 5195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.886476542307966e-07,
+      "loss": 0.9687,
+      "step": 5196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.863207484116987e-07,
+      "loss": 1.0243,
+      "step": 5197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.839964420942795e-07,
+      "loss": 0.9512,
+      "step": 5198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.816747359488632e-07,
+      "loss": 0.9819,
+      "step": 5199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.793556306450125e-07,
+      "loss": 1.0081,
+      "step": 5200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.770391268515523e-07,
+      "loss": 1.016,
+      "step": 5201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.747252252365446e-07,
+      "loss": 1.0153,
+      "step": 5202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.724139264673116e-07,
+      "loss": 0.9281,
+      "step": 5203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.701052312104208e-07,
+      "loss": 0.9878,
+      "step": 5204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.677991401316877e-07,
+      "loss": 0.9671,
+      "step": 5205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65495653896179e-07,
+      "loss": 1.044,
+      "step": 5206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.631947731682056e-07,
+      "loss": 0.9604,
+      "step": 5207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.608964986113346e-07,
+      "loss": 1.0405,
+      "step": 5208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586008308883754e-07,
+      "loss": 0.9983,
+      "step": 5209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.563077706613878e-07,
+      "loss": 0.9624,
+      "step": 5210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.54017318591678e-07,
+      "loss": 0.9067,
+      "step": 5211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517294753398066e-07,
+      "loss": 0.9912,
+      "step": 5212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494442415655681e-07,
+      "loss": 0.9468,
+      "step": 5213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471616179280207e-07,
+      "loss": 0.9436,
+      "step": 5214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448816050854559e-07,
+      "loss": 1.01,
+      "step": 5215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.42604203695423e-07,
+      "loss": 1.008,
+      "step": 5216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.403294144147124e-07,
+      "loss": 0.9595,
+      "step": 5217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.380572378993591e-07,
+      "loss": 0.9229,
+      "step": 5218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.357876748046546e-07,
+      "loss": 1.0655,
+      "step": 5219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.335207257851231e-07,
+      "loss": 0.9871,
+      "step": 5220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312563914945461e-07,
+      "loss": 1.0467,
+      "step": 5221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.289946725859444e-07,
+      "loss": 0.9542,
+      "step": 5222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267355697115887e-07,
+      "loss": 0.9993,
+      "step": 5223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.244790835229922e-07,
+      "loss": 0.9194,
+      "step": 5224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.222252146709143e-07,
+      "loss": 0.9607,
+      "step": 5225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.199739638053595e-07,
+      "loss": 0.9713,
+      "step": 5226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177253315755796e-07,
+      "loss": 0.9759,
+      "step": 5227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.154793186300693e-07,
+      "loss": 1.0229,
+      "step": 5228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.132359256165668e-07,
+      "loss": 0.9695,
+      "step": 5229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.10995153182056e-07,
+      "loss": 1.0833,
+      "step": 5230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.08757001972762e-07,
+      "loss": 0.9645,
+      "step": 5231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.065214726341653e-07,
+      "loss": 0.9473,
+      "step": 5232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.042885658109712e-07,
+      "loss": 1.0233,
+      "step": 5233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.020582821471479e-07,
+      "loss": 0.978,
+      "step": 5234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.998306222858921e-07,
+      "loss": 0.9757,
+      "step": 5235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.976055868696543e-07,
+      "loss": 0.9595,
+      "step": 5236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.953831765401233e-07,
+      "loss": 1.0456,
+      "step": 5237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.931633919382299e-07,
+      "loss": 0.999,
+      "step": 5238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.909462337041508e-07,
+      "loss": 0.9549,
+      "step": 5239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.887317024773012e-07,
+      "loss": 0.9862,
+      "step": 5240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.865197988963459e-07,
+      "loss": 0.9327,
+      "step": 5241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.843105235991834e-07,
+      "loss": 1.0017,
+      "step": 5242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.821038772229596e-07,
+      "loss": 0.9242,
+      "step": 5243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.798998604040609e-07,
+      "loss": 1.0384,
+      "step": 5244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.776984737781135e-07,
+      "loss": 0.9456,
+      "step": 5245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754997179799873e-07,
+      "loss": 1.0035,
+      "step": 5246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.733035936437962e-07,
+      "loss": 1.0172,
+      "step": 5247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.711101014028855e-07,
+      "loss": 0.9637,
+      "step": 5248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.689192418898529e-07,
+      "loss": 1.025,
+      "step": 5249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.667310157365305e-07,
+      "loss": 0.9417,
+      "step": 5250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 1.0671,
+      "step": 5251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623624660325514e-07,
+      "loss": 0.9729,
+      "step": 5252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601821437417612e-07,
+      "loss": 1.0018,
+      "step": 5253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.580044573304203e-07,
+      "loss": 1.0278,
+      "step": 5254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.558294074265605e-07,
+      "loss": 0.9883,
+      "step": 5255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.536569946574546e-07,
+      "loss": 0.9568,
+      "step": 5256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.514872196496182e-07,
+      "loss": 1.0147,
+      "step": 5257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493200830288029e-07,
+      "loss": 1.0076,
+      "step": 5258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.471555854199987e-07,
+      "loss": 0.9715,
+      "step": 5259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.449937274474396e-07,
+      "loss": 0.9846,
+      "step": 5260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.428345097345936e-07,
+      "loss": 1.0134,
+      "step": 5261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.406779329041681e-07,
+      "loss": 0.9811,
+      "step": 5262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385239975781112e-07,
+      "loss": 0.9739,
+      "step": 5263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363727043776037e-07,
+      "loss": 0.9108,
+      "step": 5264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34224053923074e-07,
+      "loss": 1.0478,
+      "step": 5265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320780468341761e-07,
+      "loss": 0.9851,
+      "step": 5266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.299346837298139e-07,
+      "loss": 0.994,
+      "step": 5267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.277939652281186e-07,
+      "loss": 1.0621,
+      "step": 5268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.256558919464653e-07,
+      "loss": 0.9391,
+      "step": 5269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.235204645014639e-07,
+      "loss": 0.91,
+      "step": 5270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.213876835089607e-07,
+      "loss": 1.0157,
+      "step": 5271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.192575495840404e-07,
+      "loss": 1.0254,
+      "step": 5272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.171300633410217e-07,
+      "loss": 1.014,
+      "step": 5273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.150052253934637e-07,
+      "loss": 0.9417,
+      "step": 5274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128830363541574e-07,
+      "loss": 0.9698,
+      "step": 5275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.107634968351341e-07,
+      "loss": 1.0073,
+      "step": 5276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.086466074476562e-07,
+      "loss": 1.0556,
+      "step": 5277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.065323688022264e-07,
+      "loss": 0.9885,
+      "step": 5278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.044207815085792e-07,
+      "loss": 0.9411,
+      "step": 5279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.023118461756874e-07,
+      "loss": 0.9771,
+      "step": 5280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.002055634117578e-07,
+      "loss": 0.9305,
+      "step": 5281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.981019338242324e-07,
+      "loss": 1.0505,
+      "step": 5282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.96000958019788e-07,
+      "loss": 1.0328,
+      "step": 5283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.939026366043323e-07,
+      "loss": 0.9679,
+      "step": 5284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918069701830189e-07,
+      "loss": 0.9512,
+      "step": 5285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.897139593602188e-07,
+      "loss": 1.0241,
+      "step": 5286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876236047395525e-07,
+      "loss": 1.0126,
+      "step": 5287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.855359069238666e-07,
+      "loss": 0.9495,
+      "step": 5288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.834508665152418e-07,
+      "loss": 1.0086,
+      "step": 5289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.81368484114996e-07,
+      "loss": 0.9589,
+      "step": 5290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.792887603236754e-07,
+      "loss": 0.9628,
+      "step": 5291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.772116957410636e-07,
+      "loss": 1.0384,
+      "step": 5292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.75137290966177e-07,
+      "loss": 0.9438,
+      "step": 5293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.730655465972636e-07,
+      "loss": 0.9461,
+      "step": 5294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.709964632318035e-07,
+      "loss": 1.0029,
+      "step": 5295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689300414665124e-07,
+      "loss": 1.0476,
+      "step": 5296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668662818973316e-07,
+      "loss": 0.9271,
+      "step": 5297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.648051851194471e-07,
+      "loss": 0.9688,
+      "step": 5298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.627467517272613e-07,
+      "loss": 0.9649,
+      "step": 5299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606909823144237e-07,
+      "loss": 1.0105,
+      "step": 5300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.586378774738012e-07,
+      "loss": 1.0032,
+      "step": 5301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.565874377975046e-07,
+      "loss": 0.9603,
+      "step": 5302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.545396638768698e-07,
+      "loss": 1.0329,
+      "step": 5303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.524945563024622e-07,
+      "loss": 1.0281,
+      "step": 5304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.504521156640854e-07,
+      "loss": 0.9648,
+      "step": 5305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.48412342550765e-07,
+      "loss": 1.0013,
+      "step": 5306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463752375507649e-07,
+      "loss": 1.0079,
+      "step": 5307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.443408012515751e-07,
+      "loss": 0.9915,
+      "step": 5308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.423090342399176e-07,
+      "loss": 0.955,
+      "step": 5309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.402799371017444e-07,
+      "loss": 1.0477,
+      "step": 5310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.382535104222366e-07,
+      "loss": 0.9771,
+      "step": 5311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.362297547858044e-07,
+      "loss": 1.0135,
+      "step": 5312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.342086707760932e-07,
+      "loss": 0.9329,
+      "step": 5313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32190258975971e-07,
+      "loss": 1.0068,
+      "step": 5314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.301745199675392e-07,
+      "loss": 0.964,
+      "step": 5315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281614543321269e-07,
+      "loss": 1.0186,
+      "step": 5316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.26151062650291e-07,
+      "loss": 1.001,
+      "step": 5317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.241433455018232e-07,
+      "loss": 0.9099,
+      "step": 5318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.221383034657326e-07,
+      "loss": 1.0216,
+      "step": 5319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.201359371202698e-07,
+      "loss": 1.0454,
+      "step": 5320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.181362470429054e-07,
+      "loss": 0.9251,
+      "step": 5321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.161392338103401e-07,
+      "loss": 0.939,
+      "step": 5322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.141448979985033e-07,
+      "loss": 1.0135,
+      "step": 5323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.121532401825515e-07,
+      "loss": 1.0272,
+      "step": 5324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.101642609368675e-07,
+      "loss": 0.9932,
+      "step": 5325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.08177960835068e-07,
+      "loss": 0.9774,
+      "step": 5326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.061943404499883e-07,
+      "loss": 0.9752,
+      "step": 5327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.042134003536971e-07,
+      "loss": 0.9367,
+      "step": 5328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.9932,
+      "step": 5329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.00259563311877e-07,
+      "loss": 0.9832,
+      "step": 5330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.98286667506618e-07,
+      "loss": 0.9941,
+      "step": 5331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.963164542706791e-07,
+      "loss": 0.9339,
+      "step": 5332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.943489241722646e-07,
+      "loss": 0.9107,
+      "step": 5333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.923840777787983e-07,
+      "loss": 1.0719,
+      "step": 5334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.904219156569325e-07,
+      "loss": 0.9757,
+      "step": 5335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.884624383725469e-07,
+      "loss": 0.9718,
+      "step": 5336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.865056464907416e-07,
+      "loss": 0.9811,
+      "step": 5337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.845515405758518e-07,
+      "loss": 1.0148,
+      "step": 5338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.826001211914257e-07,
+      "loss": 0.9704,
+      "step": 5339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.806513889002486e-07,
+      "loss": 0.9592,
+      "step": 5340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.787053442643233e-07,
+      "loss": 1.0327,
+      "step": 5341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.767619878448783e-07,
+      "loss": 0.9688,
+      "step": 5342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748213202023712e-07,
+      "loss": 1.003,
+      "step": 5343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.728833418964797e-07,
+      "loss": 0.9746,
+      "step": 5344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709480534861046e-07,
+      "loss": 0.9839,
+      "step": 5345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.690154555293793e-07,
+      "loss": 0.9173,
+      "step": 5346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.670855485836525e-07,
+      "loss": 1.0034,
+      "step": 5347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.651583332055011e-07,
+      "loss": 1.0503,
+      "step": 5348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.632338099507241e-07,
+      "loss": 0.9636,
+      "step": 5349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.613119793743428e-07,
+      "loss": 0.9716,
+      "step": 5350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593928420306083e-07,
+      "loss": 1.0102,
+      "step": 5351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.574763984729859e-07,
+      "loss": 0.965,
+      "step": 5352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.555626492541744e-07,
+      "loss": 1.0031,
+      "step": 5353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.536515949260835e-07,
+      "loss": 0.9674,
+      "step": 5354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.517432360398556e-07,
+      "loss": 1.0287,
+      "step": 5355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 0.9672,
+      "step": 5356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.479346067936554e-07,
+      "loss": 0.9653,
+      "step": 5357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460343375320743e-07,
+      "loss": 1.0132,
+      "step": 5358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.441367659091358e-07,
+      "loss": 0.9738,
+      "step": 5359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.422418924720907e-07,
+      "loss": 0.9411,
+      "step": 5360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403497177674112e-07,
+      "loss": 1.0169,
+      "step": 5361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.384602423407904e-07,
+      "loss": 0.995,
+      "step": 5362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.365734667371448e-07,
+      "loss": 1.0013,
+      "step": 5363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346893915006136e-07,
+      "loss": 1.0061,
+      "step": 5364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.32808017174551e-07,
+      "loss": 1.0099,
+      "step": 5365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.309293443015385e-07,
+      "loss": 0.9116,
+      "step": 5366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.290533734233762e-07,
+      "loss": 0.9912,
+      "step": 5367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.271801050810856e-07,
+      "loss": 0.9894,
+      "step": 5368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.253095398149067e-07,
+      "loss": 1.0178,
+      "step": 5369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.234416781643015e-07,
+      "loss": 1.0074,
+      "step": 5370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.215765206679569e-07,
+      "loss": 0.9793,
+      "step": 5371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.19714067863767e-07,
+      "loss": 1.0473,
+      "step": 5372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.178543202888621e-07,
+      "loss": 0.9618,
+      "step": 5373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.159972784795798e-07,
+      "loss": 0.9843,
+      "step": 5374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.14142942971484e-07,
+      "loss": 0.9709,
+      "step": 5375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.122913142993548e-07,
+      "loss": 0.9673,
+      "step": 5376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.104423929971948e-07,
+      "loss": 0.9832,
+      "step": 5377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.085961795982209e-07,
+      "loss": 0.9745,
+      "step": 5378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.06752674634875e-07,
+      "loss": 1.0428,
+      "step": 5379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.049118786388153e-07,
+      "loss": 0.949,
+      "step": 5380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 1.0095,
+      "step": 5381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.012384156712758e-07,
+      "loss": 0.9144,
+      "step": 5382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.994057497592032e-07,
+      "loss": 1.0193,
+      "step": 5383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.975757949332373e-07,
+      "loss": 0.9537,
+      "step": 5384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.957485517211214e-07,
+      "loss": 0.9516,
+      "step": 5385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.939240206498287e-07,
+      "loss": 1.0365,
+      "step": 5386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921022022455425e-07,
+      "loss": 0.9506,
+      "step": 5387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90283097033667e-07,
+      "loss": 0.9764,
+      "step": 5388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884667055388226e-07,
+      "loss": 0.9537,
+      "step": 5389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.866530282848492e-07,
+      "loss": 1.0428,
+      "step": 5390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84842065794804e-07,
+      "loss": 1.0031,
+      "step": 5391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.830338185909545e-07,
+      "loss": 0.9847,
+      "step": 5392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.81228287194796e-07,
+      "loss": 1.0101,
+      "step": 5393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794254721270331e-07,
+      "loss": 0.9554,
+      "step": 5394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.776253739075887e-07,
+      "loss": 0.9566,
+      "step": 5395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.758279930556021e-07,
+      "loss": 1.0169,
+      "step": 5396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.740333300894319e-07,
+      "loss": 0.9549,
+      "step": 5397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.722413855266451e-07,
+      "loss": 0.9647,
+      "step": 5398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.70452159884034e-07,
+      "loss": 0.9761,
+      "step": 5399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686656536776025e-07,
+      "loss": 1.0084,
+      "step": 5400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.668818674225684e-07,
+      "loss": 0.9878,
+      "step": 5401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.651008016333703e-07,
+      "loss": 0.994,
+      "step": 5402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.633224568236539e-07,
+      "loss": 0.9617,
+      "step": 5403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.615468335062913e-07,
+      "loss": 1.0124,
+      "step": 5404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.597739321933582e-07,
+      "loss": 0.988,
+      "step": 5405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.580037533961546e-07,
+      "loss": 0.9205,
+      "step": 5406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 1.0233,
+      "step": 5407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5447156539019e-07,
+      "loss": 0.9802,
+      "step": 5408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.527095572000962e-07,
+      "loss": 0.9754,
+      "step": 5409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509502735630601e-07,
+      "loss": 1.0012,
+      "step": 5410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.491937149864546e-07,
+      "loss": 0.9711,
+      "step": 5411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4743988197686e-07,
+      "loss": 0.9423,
+      "step": 5412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.456887750400752e-07,
+      "loss": 0.9461,
+      "step": 5413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.439403946811094e-07,
+      "loss": 0.946,
+      "step": 5414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.421947414041884e-07,
+      "loss": 0.9509,
+      "step": 5415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404518157127481e-07,
+      "loss": 1.0121,
+      "step": 5416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.387116181094431e-07,
+      "loss": 0.9687,
+      "step": 5417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369741490961345e-07,
+      "loss": 0.9697,
+      "step": 5418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.352394091739022e-07,
+      "loss": 0.9196,
+      "step": 5419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.335073988430373e-07,
+      "loss": 0.9645,
+      "step": 5420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317781186030413e-07,
+      "loss": 1.0271,
+      "step": 5421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30051568952632e-07,
+      "loss": 0.9421,
+      "step": 5422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283277503897355e-07,
+      "loss": 1.018,
+      "step": 5423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266066634114975e-07,
+      "loss": 1.0254,
+      "step": 5424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.248883085142653e-07,
+      "loss": 0.9677,
+      "step": 5425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.231726861936082e-07,
+      "loss": 0.9372,
+      "step": 5426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.214597969443025e-07,
+      "loss": 0.9816,
+      "step": 5427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.197496412603365e-07,
+      "loss": 0.962,
+      "step": 5428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.180422196349111e-07,
+      "loss": 0.988,
+      "step": 5429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.163375325604414e-07,
+      "loss": 0.9527,
+      "step": 5430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146355805285452e-07,
+      "loss": 1.0516,
+      "step": 5431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.129363640300611e-07,
+      "loss": 0.9661,
+      "step": 5432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.112398835550348e-07,
+      "loss": 1.0193,
+      "step": 5433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095461395927215e-07,
+      "loss": 0.9818,
+      "step": 5434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.078551326315917e-07,
+      "loss": 1.0571,
+      "step": 5435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.061668631593197e-07,
+      "loss": 0.9888,
+      "step": 5436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.044813316627994e-07,
+      "loss": 0.9487,
+      "step": 5437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027985386281231e-07,
+      "loss": 1.0343,
+      "step": 5438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01118484540607e-07,
+      "loss": 0.9969,
+      "step": 5439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.994411698847668e-07,
+      "loss": 0.9607,
+      "step": 5440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977665951443323e-07,
+      "loss": 0.9317,
+      "step": 5441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.960947608022437e-07,
+      "loss": 1.0382,
+      "step": 5442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.944256673406478e-07,
+      "loss": 0.964,
+      "step": 5443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.927593152409072e-07,
+      "loss": 1.0046,
+      "step": 5444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.910957049835862e-07,
+      "loss": 1.037,
+      "step": 5445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.9654,
+      "step": 5446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.877767119145271e-07,
+      "loss": 0.9516,
+      "step": 5447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.861213300599687e-07,
+      "loss": 0.982,
+      "step": 5448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.844686919621932e-07,
+      "loss": 1.035,
+      "step": 5449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.828187980978171e-07,
+      "loss": 1.0149,
+      "step": 5450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.811716489426565e-07,
+      "loss": 0.9442,
+      "step": 5451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795272449717458e-07,
+      "loss": 1.0168,
+      "step": 5452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.778855866593202e-07,
+      "loss": 0.9645,
+      "step": 5453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7624667447882857e-07,
+      "loss": 1.0303,
+      "step": 5454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.746105089029229e-07,
+      "loss": 1.0223,
+      "step": 5455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7297709040346474e-07,
+      "loss": 0.9898,
+      "step": 5456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7134641945152936e-07,
+      "loss": 0.9544,
+      "step": 5457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.697184965173884e-07,
+      "loss": 0.9914,
+      "step": 5458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6809332207053083e-07,
+      "loss": 0.9967,
+      "step": 5459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.664708965796472e-07,
+      "loss": 0.8817,
+      "step": 5460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6485122051263764e-07,
+      "loss": 0.9961,
+      "step": 5461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.632342943366097e-07,
+      "loss": 1.0567,
+      "step": 5462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.616201185178748e-07,
+      "loss": 0.983,
+      "step": 5463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6000869352195607e-07,
+      "loss": 0.9539,
+      "step": 5464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5840001981358075e-07,
+      "loss": 1.0,
+      "step": 5465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5679409785668096e-07,
+      "loss": 1.0074,
+      "step": 5466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5519092811439627e-07,
+      "loss": 0.9349,
+      "step": 5467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.535905110490779e-07,
+      "loss": 0.9846,
+      "step": 5468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.519928471222712e-07,
+      "loss": 1.0872,
+      "step": 5469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.50397936794742e-07,
+      "loss": 0.9363,
+      "step": 5470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4880578052644786e-07,
+      "loss": 0.989,
+      "step": 5471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4721637877656377e-07,
+      "loss": 1.0286,
+      "step": 5472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4562973200346413e-07,
+      "loss": 0.9615,
+      "step": 5473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4404584066472965e-07,
+      "loss": 0.9918,
+      "step": 5474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4246470521714825e-07,
+      "loss": 0.976,
+      "step": 5475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408863261167096e-07,
+      "loss": 0.9986,
+      "step": 5476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.39310703818614e-07,
+      "loss": 0.9914,
+      "step": 5477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.377378387772602e-07,
+      "loss": 0.9611,
+      "step": 5478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3616773144625644e-07,
+      "loss": 0.9921,
+      "step": 5479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3460038227841396e-07,
+      "loss": 1.0201,
+      "step": 5480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3303579172574884e-07,
+      "loss": 0.9959,
+      "step": 5481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.314739602394791e-07,
+      "loss": 0.986,
+      "step": 5482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.299148882700355e-07,
+      "loss": 0.995,
+      "step": 5483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.283585762670384e-07,
+      "loss": 1.0107,
+      "step": 5484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 0.9127,
+      "step": 5485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2525423395493703e-07,
+      "loss": 1.0019,
+      "step": 5486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.237062045411067e-07,
+      "loss": 1.0285,
+      "step": 5487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2216093688428384e-07,
+      "loss": 0.9701,
+      "step": 5488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2061843143011204e-07,
+      "loss": 0.9113,
+      "step": 5489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190786886234466e-07,
+      "loss": 1.0169,
+      "step": 5490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1754170890833777e-07,
+      "loss": 0.9359,
+      "step": 5491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.160074927280466e-07,
+      "loss": 0.9754,
+      "step": 5492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1447604052503255e-07,
+      "loss": 0.9636,
+      "step": 5493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1294735274095777e-07,
+      "loss": 1.0355,
+      "step": 5494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1142142981668964e-07,
+      "loss": 0.9762,
+      "step": 5495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098982721922962e-07,
+      "loss": 1.0161,
+      "step": 5496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.083778803070504e-07,
+      "loss": 1.0343,
+      "step": 5497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.068602545994249e-07,
+      "loss": 0.9223,
+      "step": 5498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0534539550709516e-07,
+      "loss": 0.9532,
+      "step": 5499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.038333034669406e-07,
+      "loss": 0.9746,
+      "step": 5500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0232397891504016e-07,
+      "loss": 1.0219,
+      "step": 5501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.008174222866745e-07,
+      "loss": 0.9835,
+      "step": 5502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9931363401633174e-07,
+      "loss": 0.9593,
+      "step": 5503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.978126145376915e-07,
+      "loss": 1.006,
+      "step": 5504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.963143642836442e-07,
+      "loss": 0.9859,
+      "step": 5505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9481888368627764e-07,
+      "loss": 1.0203,
+      "step": 5506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9332617317688004e-07,
+      "loss": 1.0348,
+      "step": 5507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.918362331859438e-07,
+      "loss": 1.0108,
+      "step": 5508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.903490641431573e-07,
+      "loss": 0.9169,
+      "step": 5509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.888646664774154e-07,
+      "loss": 0.9772,
+      "step": 5510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 0.9846,
+      "step": 5511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.859041869886382e-07,
+      "loss": 0.986,
+      "step": 5512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.844281060193911e-07,
+      "loss": 0.9828,
+      "step": 5513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8295479813476253e-07,
+      "loss": 1.0401,
+      "step": 5514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.814842637596483e-07,
+      "loss": 0.9615,
+      "step": 5515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.800165033181447e-07,
+      "loss": 0.9852,
+      "step": 5516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785515172335463e-07,
+      "loss": 1.0061,
+      "step": 5517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.770893059283465e-07,
+      "loss": 1.0243,
+      "step": 5518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.756298698242422e-07,
+      "loss": 0.9502,
+      "step": 5519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7417320934212573e-07,
+      "loss": 0.9657,
+      "step": 5520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7271932490209327e-07,
+      "loss": 1.0114,
+      "step": 5521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7126821692343606e-07,
+      "loss": 0.9625,
+      "step": 5522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.698198858246482e-07,
+      "loss": 0.9492,
+      "step": 5523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.68374332023419e-07,
+      "loss": 1.006,
+      "step": 5524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6693155593664156e-07,
+      "loss": 0.9793,
+      "step": 5525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.654915579804041e-07,
+      "loss": 0.9751,
+      "step": 5526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6405433856999684e-07,
+      "loss": 0.988,
+      "step": 5527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.626198981199047e-07,
+      "loss": 1.0338,
+      "step": 5528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6118823704381466e-07,
+      "loss": 0.9813,
+      "step": 5529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5975935575461083e-07,
+      "loss": 1.0007,
+      "step": 5530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5833325466437697e-07,
+      "loss": 0.9068,
+      "step": 5531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5690993418439067e-07,
+      "loss": 1.0372,
+      "step": 5532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5548939472513367e-07,
+      "loss": 0.9479,
+      "step": 5533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.540716366962815e-07,
+      "loss": 0.9638,
+      "step": 5534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.526566605067072e-07,
+      "loss": 1.0224,
+      "step": 5535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5124446656448654e-07,
+      "loss": 0.9724,
+      "step": 5536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498350552768859e-07,
+      "loss": 0.9356,
+      "step": 5537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.484284270503746e-07,
+      "loss": 0.9543,
+      "step": 5538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470245822906182e-07,
+      "loss": 1.0508,
+      "step": 5539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4562352140247613e-07,
+      "loss": 1.0149,
+      "step": 5540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4422524479001187e-07,
+      "loss": 1.0135,
+      "step": 5541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4282975285647615e-07,
+      "loss": 1.0143,
+      "step": 5542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.414370460043259e-07,
+      "loss": 0.9537,
+      "step": 5543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4004712463521105e-07,
+      "loss": 0.962,
+      "step": 5544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3865998914997645e-07,
+      "loss": 1.006,
+      "step": 5545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3727563994866655e-07,
+      "loss": 0.9988,
+      "step": 5546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3589407743052085e-07,
+      "loss": 0.9122,
+      "step": 5547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3451530199397395e-07,
+      "loss": 0.9694,
+      "step": 5548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.331393140366601e-07,
+      "loss": 1.0102,
+      "step": 5549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3176611395540625e-07,
+      "loss": 0.978,
+      "step": 5550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.303957021462378e-07,
+      "loss": 0.9713,
+      "step": 5551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2902807900437317e-07,
+      "loss": 0.9275,
+      "step": 5552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2766324492422896e-07,
+      "loss": 0.9734,
+      "step": 5553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2630120029942034e-07,
+      "loss": 0.9998,
+      "step": 5554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.249419455227476e-07,
+      "loss": 0.9676,
+      "step": 5555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.235854809862193e-07,
+      "loss": 0.9694,
+      "step": 5556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2223180708102933e-07,
+      "loss": 0.9891,
+      "step": 5557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.20880924197573e-07,
+      "loss": 0.9594,
+      "step": 5558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1953283272543767e-07,
+      "loss": 0.9886,
+      "step": 5559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1818753305340566e-07,
+      "loss": 0.9849,
+      "step": 5560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1684502556945664e-07,
+      "loss": 0.9936,
+      "step": 5561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1550531066076085e-07,
+      "loss": 0.985,
+      "step": 5562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 1.0131,
+      "step": 5563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.128342601138001e-07,
+      "loss": 0.9358,
+      "step": 5564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1150292524585234e-07,
+      "loss": 0.9713,
+      "step": 5565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.101743844937943e-07,
+      "loss": 0.9584,
+      "step": 5566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.088486382407718e-07,
+      "loss": 0.9933,
+      "step": 5567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0752568686912343e-07,
+      "loss": 1.0013,
+      "step": 5568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0620553076038394e-07,
+      "loss": 0.9999,
+      "step": 5569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0488817029527553e-07,
+      "loss": 0.9234,
+      "step": 5570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.035736058537231e-07,
+      "loss": 0.9315,
+      "step": 5571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0226183781483897e-07,
+      "loss": 0.9869,
+      "step": 5572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.009528665569295e-07,
+      "loss": 1.0731,
+      "step": 5573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.996466924574992e-07,
+      "loss": 0.9554,
+      "step": 5574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9834331589323697e-07,
+      "loss": 0.9414,
+      "step": 5575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.970427372400353e-07,
+      "loss": 0.942,
+      "step": 5576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.957449568729731e-07,
+      "loss": 1.0304,
+      "step": 5577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9444997516632324e-07,
+      "loss": 1.0163,
+      "step": 5578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.931577924935536e-07,
+      "loss": 0.9632,
+      "step": 5579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9186840922732165e-07,
+      "loss": 0.9202,
+      "step": 5580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 0.972,
+      "step": 5581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8929804240107383e-07,
+      "loss": 0.9857,
+      "step": 5582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8801705958233947e-07,
+      "loss": 0.9938,
+      "step": 5583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.867388776527069e-07,
+      "loss": 1.0424,
+      "step": 5584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.854634969807979e-07,
+      "loss": 0.9923,
+      "step": 5585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8419091793442266e-07,
+      "loss": 0.9345,
+      "step": 5586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.829211408805932e-07,
+      "loss": 0.9989,
+      "step": 5587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.816541661854999e-07,
+      "loss": 0.9809,
+      "step": 5588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8038999421453827e-07,
+      "loss": 1.0122,
+      "step": 5589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.791286253322856e-07,
+      "loss": 0.8981,
+      "step": 5590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7787005990251635e-07,
+      "loss": 1.0054,
+      "step": 5591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7661429828819366e-07,
+      "loss": 0.9905,
+      "step": 5592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7536134085147324e-07,
+      "loss": 0.996,
+      "step": 5593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7411118795370396e-07,
+      "loss": 1.0437,
+      "step": 5594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7286383995542063e-07,
+      "loss": 0.9373,
+      "step": 5595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.716192972163556e-07,
+      "loss": 0.9826,
+      "step": 5596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7037756009542616e-07,
+      "loss": 0.9564,
+      "step": 5597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.691386289507458e-07,
+      "loss": 1.0203,
+      "step": 5598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.679025041396155e-07,
+      "loss": 0.9359,
+      "step": 5599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.666691860185266e-07,
+      "loss": 1.0188,
+      "step": 5600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6543867494316254e-07,
+      "loss": 1.0068,
+      "step": 5601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6421097126839714e-07,
+      "loss": 0.9615,
+      "step": 5602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.629860753482949e-07,
+      "loss": 1.0141,
+      "step": 5603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6176398753610885e-07,
+      "loss": 0.9388,
+      "step": 5604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605447081842838e-07,
+      "loss": 1.0688,
+      "step": 5605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.593282376444539e-07,
+      "loss": 0.9738,
+      "step": 5606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.581145762674442e-07,
+      "loss": 0.9721,
+      "step": 5607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.569037244032657e-07,
+      "loss": 1.0031,
+      "step": 5608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5569568240112696e-07,
+      "loss": 0.9282,
+      "step": 5609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.54490450609417e-07,
+      "loss": 1.018,
+      "step": 5610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532880293757223e-07,
+      "loss": 1.0512,
+      "step": 5611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5208841904681314e-07,
+      "loss": 1.0172,
+      "step": 5612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508916199686517e-07,
+      "loss": 0.963,
+      "step": 5613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.496976324863898e-07,
+      "loss": 0.9463,
+      "step": 5614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4850645694436736e-07,
+      "loss": 1.0651,
+      "step": 5615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4731809368611414e-07,
+      "loss": 1.0246,
+      "step": 5616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.461325430543482e-07,
+      "loss": 0.9976,
+      "step": 5617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.44949805390976e-07,
+      "loss": 0.9082,
+      "step": 5618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4376988103709254e-07,
+      "loss": 1.0159,
+      "step": 5619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.425927703329856e-07,
+      "loss": 0.9067,
+      "step": 5620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4141847361812464e-07,
+      "loss": 1.0123,
+      "step": 5621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4024699123117444e-07,
+      "loss": 1.0016,
+      "step": 5622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.390783235099814e-07,
+      "loss": 0.9395,
+      "step": 5623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3791247079158586e-07,
+      "loss": 0.9337,
+      "step": 5624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3674943341221447e-07,
+      "loss": 1.0268,
+      "step": 5625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.355892117072789e-07,
+      "loss": 0.9699,
+      "step": 5626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3443180601138592e-07,
+      "loss": 0.9818,
+      "step": 5627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.332772166583208e-07,
+      "loss": 1.0066,
+      "step": 5628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3212544398106497e-07,
+      "loss": 1.0025,
+      "step": 5629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3097648831178376e-07,
+      "loss": 0.9748,
+      "step": 5630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2983034998182997e-07,
+      "loss": 0.9494,
+      "step": 5631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2868702932174357e-07,
+      "loss": 1.0298,
+      "step": 5632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.275465266612542e-07,
+      "loss": 0.9035,
+      "step": 5633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264088423292743e-07,
+      "loss": 0.9962,
+      "step": 5634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2527397665391026e-07,
+      "loss": 0.9717,
+      "step": 5635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2414192996245143e-07,
+      "loss": 1.0496,
+      "step": 5636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.230127025813722e-07,
+      "loss": 0.9964,
+      "step": 5637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2188629483633872e-07,
+      "loss": 0.9469,
+      "step": 5638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2076270705219893e-07,
+      "loss": 0.9908,
+      "step": 5639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1964193955299363e-07,
+      "loss": 0.953,
+      "step": 5640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.9467,
+      "step": 5641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1740886670146176e-07,
+      "loss": 0.9661,
+      "step": 5642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.162965619931423e-07,
+      "loss": 1.0333,
+      "step": 5643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1518707885777147e-07,
+      "loss": 0.9717,
+      "step": 5644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1408041761531771e-07,
+      "loss": 1.0184,
+      "step": 5645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.129765785849358e-07,
+      "loss": 1.0272,
+      "step": 5646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1187556208496885e-07,
+      "loss": 0.8962,
+      "step": 5647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1077736843294394e-07,
+      "loss": 0.9625,
+      "step": 5648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0968199794557554e-07,
+      "loss": 0.9846,
+      "step": 5649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0858945093876315e-07,
+      "loss": 0.9812,
+      "step": 5650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0749972772759142e-07,
+      "loss": 0.9893,
+      "step": 5651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0641282862633118e-07,
+      "loss": 0.9107,
+      "step": 5652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0532875394844053e-07,
+      "loss": 0.9872,
+      "step": 5653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0424750400655947e-07,
+      "loss": 0.9857,
+      "step": 5654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0316907911251626e-07,
+      "loss": 0.9727,
+      "step": 5655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0209347957732328e-07,
+      "loss": 0.9587,
+      "step": 5656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0102070571117794e-07,
+      "loss": 0.9656,
+      "step": 5657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9995075782346384e-07,
+      "loss": 0.9934,
+      "step": 5658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9888363622274642e-07,
+      "loss": 0.9591,
+      "step": 5659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.978193412167828e-07,
+      "loss": 1.0706,
+      "step": 5660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.967578731125064e-07,
+      "loss": 0.9737,
+      "step": 5661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9569923221604224e-07,
+      "loss": 0.9699,
+      "step": 5662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.946434188326951e-07,
+      "loss": 0.951,
+      "step": 5663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9359043326695803e-07,
+      "loss": 1.0641,
+      "step": 5664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9254027582250588e-07,
+      "loss": 0.946,
+      "step": 5665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9149294680220087e-07,
+      "loss": 0.9756,
+      "step": 5666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9044844650808468e-07,
+      "loss": 0.993,
+      "step": 5667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8940677524138861e-07,
+      "loss": 0.9604,
+      "step": 5668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8836793330252457e-07,
+      "loss": 0.9802,
+      "step": 5669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733192099108954e-07,
+      "loss": 1.0344,
+      "step": 5670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8629873860586567e-07,
+      "loss": 0.9355,
+      "step": 5671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.852683864448157e-07,
+      "loss": 0.993,
+      "step": 5672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8424086480509195e-07,
+      "loss": 0.9757,
+      "step": 5673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8321617398302182e-07,
+      "loss": 1.0109,
+      "step": 5674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.821943142741256e-07,
+      "loss": 0.9728,
+      "step": 5675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8117528597309864e-07,
+      "loss": 0.9634,
+      "step": 5676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8015908937382587e-07,
+      "loss": 0.9533,
+      "step": 5677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7914572476937507e-07,
+      "loss": 1.0201,
+      "step": 5678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.781351924519925e-07,
+      "loss": 0.9515,
+      "step": 5679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7712749271311392e-07,
+      "loss": 0.9377,
+      "step": 5680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.761226258433524e-07,
+      "loss": 1.0109,
+      "step": 5681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7512059213250832e-07,
+      "loss": 0.9347,
+      "step": 5682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741213918695628e-07,
+      "loss": 1.0161,
+      "step": 5683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7312502534268083e-07,
+      "loss": 1.02,
+      "step": 5684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7213149283920928e-07,
+      "loss": 0.9685,
+      "step": 5685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.711407946456789e-07,
+      "loss": 0.9974,
+      "step": 5686
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7015293104780007e-07,
+      "loss": 0.9709,
+      "step": 5687
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6916790233047042e-07,
+      "loss": 1.0372,
+      "step": 5688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.681857087777672e-07,
+      "loss": 0.9649,
+      "step": 5689
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6720635067294933e-07,
+      "loss": 0.9446,
+      "step": 5690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.662298282984587e-07,
+      "loss": 1.0309,
+      "step": 5691
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6525614193592e-07,
+      "loss": 1.0008,
+      "step": 5692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6428529186614195e-07,
+      "loss": 0.9646,
+      "step": 5693
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6331727836910838e-07,
+      "loss": 0.9692,
+      "step": 5694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6235210172399373e-07,
+      "loss": 0.9899,
+      "step": 5695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6138976220914872e-07,
+      "loss": 0.9757,
+      "step": 5696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6043026010210795e-07,
+      "loss": 0.9925,
+      "step": 5697
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5947359567958677e-07,
+      "loss": 1.0179,
+      "step": 5698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5851976921748225e-07,
+      "loss": 0.9781,
+      "step": 5699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5756878099087436e-07,
+      "loss": 0.9822,
+      "step": 5700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5662063127402262e-07,
+      "loss": 0.9484,
+      "step": 5701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5567532034036936e-07,
+      "loss": 1.023,
+      "step": 5702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5473284846253766e-07,
+      "loss": 0.9704,
+      "step": 5703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.537932159123312e-07,
+      "loss": 1.0148,
+      "step": 5704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5285642296073656e-07,
+      "loss": 0.9938,
+      "step": 5705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.968,
+      "step": 5706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5099135693322776e-07,
+      "loss": 1.011,
+      "step": 5707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5006308439519002e-07,
+      "loss": 0.9582,
+      "step": 5708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.491376525315158e-07,
+      "loss": 0.9621,
+      "step": 5709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4821506160909492e-07,
+      "loss": 1.0074,
+      "step": 5710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4729531189399903e-07,
+      "loss": 0.9418,
+      "step": 5711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4637840365147704e-07,
+      "loss": 1.015,
+      "step": 5712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4546433714596298e-07,
+      "loss": 0.9956,
+      "step": 5713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.445531126410682e-07,
+      "loss": 0.9115,
+      "step": 5714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4364473039958693e-07,
+      "loss": 1.044,
+      "step": 5715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4273919068349184e-07,
+      "loss": 0.9575,
+      "step": 5716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4183649375393404e-07,
+      "loss": 0.9803,
+      "step": 5717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4093663987124973e-07,
+      "loss": 1.0006,
+      "step": 5718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 0.9902,
+      "step": 5719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3914546228373182e-07,
+      "loss": 0.9385,
+      "step": 5720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3825413909546502e-07,
+      "loss": 1.0144,
+      "step": 5721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.373656599872053e-07,
+      "loss": 0.984,
+      "step": 5722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3648002521518321e-07,
+      "loss": 0.9569,
+      "step": 5723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.355972350348156e-07,
+      "loss": 0.9955,
+      "step": 5724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3471728970068986e-07,
+      "loss": 1.0251,
+      "step": 5725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3384018946658305e-07,
+      "loss": 1.0098,
+      "step": 5726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3296593458544284e-07,
+      "loss": 1.0054,
+      "step": 5727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3209452530940194e-07,
+      "loss": 0.92,
+      "step": 5728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3122596188977045e-07,
+      "loss": 0.9921,
+      "step": 5729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.30360244577038e-07,
+      "loss": 0.9948,
+      "step": 5730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2949737362087156e-07,
+      "loss": 0.9572,
+      "step": 5731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2863734927012094e-07,
+      "loss": 0.9103,
+      "step": 5732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2778017177281222e-07,
+      "loss": 1.0252,
+      "step": 5733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2692584137615205e-07,
+      "loss": 0.9898,
+      "step": 5734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2607435832652447e-07,
+      "loss": 0.9522,
+      "step": 5735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2522572286949198e-07,
+      "loss": 1.038,
+      "step": 5736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2437993524979984e-07,
+      "loss": 1.0061,
+      "step": 5737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2353699571136634e-07,
+      "loss": 0.9591,
+      "step": 5738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2269690449729254e-07,
+      "loss": 0.9602,
+      "step": 5739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2185966184985687e-07,
+      "loss": 1.0163,
+      "step": 5740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2102526801051507e-07,
+      "loss": 0.993,
+      "step": 5741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2019372321990352e-07,
+      "loss": 0.9885,
+      "step": 5742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1936502771783488e-07,
+      "loss": 0.985,
+      "step": 5743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1853918174330125e-07,
+      "loss": 0.9786,
+      "step": 5744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1771618553447217e-07,
+      "loss": 0.9861,
+      "step": 5745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1689603932869664e-07,
+      "loss": 0.9651,
+      "step": 5746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1607874336249991e-07,
+      "loss": 1.0176,
+      "step": 5747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1526429787158678e-07,
+      "loss": 0.9614,
+      "step": 5748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1445270309083934e-07,
+      "loss": 0.9877,
+      "step": 5749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1364395925431815e-07,
+      "loss": 0.969,
+      "step": 5750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1283806659525998e-07,
+      "loss": 1.0319,
+      "step": 5751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1203502534608113e-07,
+      "loss": 0.9631,
+      "step": 5752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1123483573837413e-07,
+      "loss": 0.9667,
+      "step": 5753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1043749800291103e-07,
+      "loss": 0.9888,
+      "step": 5754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0964301236963904e-07,
+      "loss": 1.007,
+      "step": 5755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0885137906768373e-07,
+      "loss": 0.9546,
+      "step": 5756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0806259832535027e-07,
+      "loss": 0.9709,
+      "step": 5757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0727667037011668e-07,
+      "loss": 1.0236,
+      "step": 5758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.064935954286428e-07,
+      "loss": 0.9448,
+      "step": 5759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0571337372676239e-07,
+      "loss": 1.0204,
+      "step": 5760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0493600548948879e-07,
+      "loss": 0.963,
+      "step": 5761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0416149094101046e-07,
+      "loss": 0.9435,
+      "step": 5762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0338983030469429e-07,
+      "loss": 0.9262,
+      "step": 5763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0262102380308225e-07,
+      "loss": 1.0528,
+      "step": 5764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0185507165789588e-07,
+      "loss": 0.9503,
+      "step": 5765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0109197409003068e-07,
+      "loss": 0.9854,
+      "step": 5766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0033173131956175e-07,
+      "loss": 0.9603,
+      "step": 5767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.957434356573814e-08,
+      "loss": 1.0546,
+      "step": 5768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.881981104698846e-08,
+      "loss": 0.9741,
+      "step": 5769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.806813398091419e-08,
+      "loss": 1.0118,
+      "step": 5770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 0.9885,
+      "step": 5771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.65733470730934e-08,
+      "loss": 0.9817,
+      "step": 5772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.583023766243427e-08,
+      "loss": 0.9214,
+      "step": 5773
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.508998456663088e-08,
+      "loss": 1.0348,
+      "step": 5774
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.435258799916803e-08,
+      "loss": 0.9349,
+      "step": 5775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.36180481727067e-08,
+      "loss": 1.0041,
+      "step": 5776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.288636529908634e-08,
+      "loss": 0.9297,
+      "step": 5777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.215753958931816e-08,
+      "loss": 0.9963,
+      "step": 5778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.143157125359514e-08,
+      "loss": 0.9565,
+      "step": 5779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.07084605012798e-08,
+      "loss": 0.9832,
+      "step": 5780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.99882075409153e-08,
+      "loss": 0.9379,
+      "step": 5781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.927081258021996e-08,
+      "loss": 1.0191,
+      "step": 5782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.855627582608605e-08,
+      "loss": 0.9786,
+      "step": 5783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.784459748458318e-08,
+      "loss": 0.9981,
+      "step": 5784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.713577776095494e-08,
+      "loss": 1.0027,
+      "step": 5785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.642981685962226e-08,
+      "loss": 0.9688,
+      "step": 5786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.572671498418005e-08,
+      "loss": 0.9735,
+      "step": 5787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.502647233740169e-08,
+      "loss": 1.017,
+      "step": 5788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.432908912123117e-08,
+      "loss": 0.9808,
+      "step": 5789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.363456553679206e-08,
+      "loss": 0.9915,
+      "step": 5790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.29429017843797e-08,
+      "loss": 0.887,
+      "step": 5791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.225409806347007e-08,
+      "loss": 1.0626,
+      "step": 5792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.15681545727065e-08,
+      "loss": 0.9272,
+      "step": 5793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.08850715099152e-08,
+      "loss": 0.9951,
+      "step": 5794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.020484907209303e-08,
+      "loss": 0.9207,
+      "step": 5795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.952748745541305e-08,
+      "loss": 1.0351,
+      "step": 5796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.9137,
+      "step": 5797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.818134746604312e-08,
+      "loss": 1.0183,
+      "step": 5798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.751256948157482e-08,
+      "loss": 0.9949,
+      "step": 5799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.684665309468875e-08,
+      "loss": 0.9928,
+      "step": 5800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.618359849743128e-08,
+      "loss": 0.9766,
+      "step": 5801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.552340588102613e-08,
+      "loss": 1.0145,
+      "step": 5802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.486607543586765e-08,
+      "loss": 0.9793,
+      "step": 5803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.421160735152755e-08,
+      "loss": 0.9898,
+      "step": 5804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.35600018167526e-08,
+      "loss": 0.9224,
+      "step": 5805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 1.0279,
+      "step": 5806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.226537914674648e-08,
+      "loss": 0.958,
+      "step": 5807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.162236238487885e-08,
+      "loss": 0.9935,
+      "step": 5808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.09822089193013e-08,
+      "loss": 1.0182,
+      "step": 5809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.034491893463059e-08,
+      "loss": 1.008,
+      "step": 5810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.971049261465746e-08,
+      "loss": 0.9471,
+      "step": 5811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.907893014234889e-08,
+      "loss": 1.0116,
+      "step": 5812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.845023169984366e-08,
+      "loss": 1.0368,
+      "step": 5813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.782439746845448e-08,
+      "loss": 0.9961,
+      "step": 5814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.720142762867032e-08,
+      "loss": 0.9187,
+      "step": 5815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.658132236015191e-08,
+      "loss": 1.044,
+      "step": 5816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.596408184173508e-08,
+      "loss": 0.9729,
+      "step": 5817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.534970625142855e-08,
+      "loss": 0.9852,
+      "step": 5818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.473819576641505e-08,
+      "loss": 1.008,
+      "step": 5819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.412955056305237e-08,
+      "loss": 0.9397,
+      "step": 5820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.352377081687011e-08,
+      "loss": 1.0077,
+      "step": 5821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.292085670257187e-08,
+      "loss": 0.9879,
+      "step": 5822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.232080839403631e-08,
+      "loss": 1.0661,
+      "step": 5823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.172362606431281e-08,
+      "loss": 0.9669,
+      "step": 5824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.112930988562804e-08,
+      "loss": 0.9908,
+      "step": 5825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.053786002937823e-08,
+      "loss": 0.9408,
+      "step": 5826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.99492766661347e-08,
+      "loss": 1.0193,
+      "step": 5827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.93635599656428e-08,
+      "loss": 1.0072,
+      "step": 5828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.878071009682074e-08,
+      "loss": 0.9384,
+      "step": 5829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.820072722775849e-08,
+      "loss": 1.0325,
+      "step": 5830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7623611525721155e-08,
+      "loss": 0.9839,
+      "step": 5831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.704936315714449e-08,
+      "loss": 1.0255,
+      "step": 5832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.647798228764156e-08,
+      "loss": 0.9702,
+      "step": 5833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5909469081995015e-08,
+      "loss": 0.9559,
+      "step": 5834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.534382370416036e-08,
+      "loss": 0.9999,
+      "step": 5835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.9561,
+      "step": 5836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.422113708361765e-08,
+      "loss": 0.9839,
+      "step": 5837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.366409616468837e-08,
+      "loss": 0.9695,
+      "step": 5838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.310992372112522e-08,
+      "loss": 0.9745,
+      "step": 5839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.255861991275035e-08,
+      "loss": 1.0021,
+      "step": 5840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2010184898556585e-08,
+      "loss": 1.0311,
+      "step": 5841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.146461883671072e-08,
+      "loss": 0.966,
+      "step": 5842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.092192188455025e-08,
+      "loss": 0.9069,
+      "step": 5843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.038209419858664e-08,
+      "loss": 1.0313,
+      "step": 5844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.984513593450424e-08,
+      "loss": 0.9601,
+      "step": 5845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9311047247156964e-08,
+      "loss": 0.9922,
+      "step": 5846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.877982829057715e-08,
+      "loss": 0.9905,
+      "step": 5847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.825147921796225e-08,
+      "loss": 0.9601,
+      "step": 5848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.772600018168816e-08,
+      "loss": 0.9398,
+      "step": 5849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.720339133329921e-08,
+      "loss": 1.0043,
+      "step": 5850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6683652823513725e-08,
+      "loss": 1.0286,
+      "step": 5851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.616678480222181e-08,
+      "loss": 0.9968,
+      "step": 5852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5652787418485336e-08,
+      "loss": 0.9502,
+      "step": 5853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.514166082053795e-08,
+      "loss": 0.9061,
+      "step": 5854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.46334051557884e-08,
+      "loss": 1.0542,
+      "step": 5855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.412802057081278e-08,
+      "loss": 1.0078,
+      "step": 5856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.362550721136338e-08,
+      "loss": 1.0306,
+      "step": 5857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.312586522236206e-08,
+      "loss": 0.9798,
+      "step": 5858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2629094747901334e-08,
+      "loss": 0.9429,
+      "step": 5859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2135195931249925e-08,
+      "loss": 0.9833,
+      "step": 5860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1644168914845015e-08,
+      "loss": 1.0735,
+      "step": 5861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.115601384029666e-08,
+      "loss": 1.0039,
+      "step": 5862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0670730848385573e-08,
+      "loss": 1.0058,
+      "step": 5863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.018832007906537e-08,
+      "loss": 0.993,
+      "step": 5864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.970878167146253e-08,
+      "loss": 0.9451,
+      "step": 5865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.923211576387087e-08,
+      "loss": 0.9542,
+      "step": 5866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.875832249376044e-08,
+      "loss": 0.9868,
+      "step": 5867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8287401997770814e-08,
+      "loss": 0.9293,
+      "step": 5868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7819354411713364e-08,
+      "loss": 0.9814,
+      "step": 5869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7354179870568994e-08,
+      "loss": 0.9518,
+      "step": 5870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.689187850849374e-08,
+      "loss": 1.019,
+      "step": 5871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6432450458812055e-08,
+      "loss": 0.9902,
+      "step": 5872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.597589585402128e-08,
+      "loss": 0.9966,
+      "step": 5873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5522214825790546e-08,
+      "loss": 1.0145,
+      "step": 5874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 1.0128,
+      "step": 5875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.462347402153232e-08,
+      "loss": 1.0134,
+      "step": 5876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.417841450469861e-08,
+      "loss": 0.9417,
+      "step": 5877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.373622908280916e-08,
+      "loss": 1.0232,
+      "step": 5878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3296917883387516e-08,
+      "loss": 1.0016,
+      "step": 5879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2860481033128996e-08,
+      "loss": 1.0259,
+      "step": 5880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.242691865790071e-08,
+      "loss": 0.9669,
+      "step": 5881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1996230882738177e-08,
+      "loss": 0.9967,
+      "step": 5882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.156841783185205e-08,
+      "loss": 1.0079,
+      "step": 5883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.114347962861919e-08,
+      "loss": 0.9573,
+      "step": 5884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0721416395590454e-08,
+      "loss": 1.0485,
+      "step": 5885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.030222825448848e-08,
+      "loss": 0.9387,
+      "step": 5886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.988591532620322e-08,
+      "loss": 0.962,
+      "step": 5887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.947247773079753e-08,
+      "loss": 0.9361,
+      "step": 5888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.906191558750604e-08,
+      "loss": 1.013,
+      "step": 5889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8654229014730694e-08,
+      "loss": 0.9839,
+      "step": 5890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8249418130049665e-08,
+      "loss": 0.9588,
+      "step": 5891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7847483050205125e-08,
+      "loss": 1.0091,
+      "step": 5892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.744842389111546e-08,
+      "loss": 0.9994,
+      "step": 5893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7052240767865278e-08,
+      "loss": 0.9586,
+      "step": 5894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6658933794714292e-08,
+      "loss": 0.9866,
+      "step": 5895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6268503085089547e-08,
+      "loss": 1.0376,
+      "step": 5896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5880948751587642e-08,
+      "loss": 0.9588,
+      "step": 5897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5496270905980282e-08,
+      "loss": 0.9741,
+      "step": 5898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5114469659203166e-08,
+      "loss": 1.022,
+      "step": 5899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.473554512136933e-08,
+      "loss": 0.9891,
+      "step": 5900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 0.9173,
+      "step": 5901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3986326608818055e-08,
+      "loss": 0.9564,
+      "step": 5902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.361603285017111e-08,
+      "loss": 1.0091,
+      "step": 5903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3248616232608435e-08,
+      "loss": 0.9893,
+      "step": 5904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2884076862089712e-08,
+      "loss": 0.9767,
+      "step": 5905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2522414843748618e-08,
+      "loss": 0.9953,
+      "step": 5906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.216363028188506e-08,
+      "loss": 0.9897,
+      "step": 5907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1807723279971825e-08,
+      "loss": 0.9923,
+      "step": 5908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.145469394065014e-08,
+      "loss": 0.9886,
+      "step": 5909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1104542365730783e-08,
+      "loss": 0.9977,
+      "step": 5910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0757268656198536e-08,
+      "loss": 0.9684,
+      "step": 5911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0412872912203286e-08,
+      "loss": 1.0132,
+      "step": 5912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.00713552330678e-08,
+      "loss": 0.9697,
+      "step": 5913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.973271571728441e-08,
+      "loss": 0.9804,
+      "step": 5914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9396954462514996e-08,
+      "loss": 0.9588,
+      "step": 5915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.906407156559098e-08,
+      "loss": 0.9781,
+      "step": 5916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8734067122514464e-08,
+      "loss": 1.0092,
+      "step": 5917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.84069412284571e-08,
+      "loss": 0.9567,
+      "step": 5918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.808269397776119e-08,
+      "loss": 1.0092,
+      "step": 5919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7761325463937495e-08,
+      "loss": 0.99,
+      "step": 5920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7442835779666322e-08,
+      "loss": 0.9383,
+      "step": 5921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7127225016799754e-08,
+      "loss": 0.9926,
+      "step": 5922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6814493266357202e-08,
+      "loss": 0.9999,
+      "step": 5923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6504640618530966e-08,
+      "loss": 0.9548,
+      "step": 5924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6197667162679563e-08,
+      "loss": 0.9486,
+      "step": 5925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5893572987333293e-08,
+      "loss": 0.9733,
+      "step": 5926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5592358180189782e-08,
+      "loss": 1.0357,
+      "step": 5927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.529402282812065e-08,
+      "loss": 0.9626,
+      "step": 5928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4998567017162623e-08,
+      "loss": 1.0034,
+      "step": 5929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4705990832523109e-08,
+      "loss": 0.917,
+      "step": 5930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4416294358582383e-08,
+      "loss": 1.0325,
+      "step": 5931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4129477678884728e-08,
+      "loss": 0.9719,
+      "step": 5932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3845540876147312e-08,
+      "loss": 0.9068,
+      "step": 5933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3564484032257963e-08,
+      "loss": 1.0486,
+      "step": 5934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3286307228269623e-08,
+      "loss": 0.9018,
+      "step": 5935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3011010544409231e-08,
+      "loss": 0.9704,
+      "step": 5936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2738594060068832e-08,
+      "loss": 0.9852,
+      "step": 5937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.246905785381447e-08,
+      "loss": 1.0136,
+      "step": 5938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.22024020033773e-08,
+      "loss": 0.9305,
+      "step": 5939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1938626585660252e-08,
+      "loss": 1.0249,
+      "step": 5940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1677731676733584e-08,
+      "loss": 1.0442,
+      "step": 5941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1419717351841553e-08,
+      "loss": 0.9744,
+      "step": 5942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1164583685390196e-08,
+      "loss": 0.9815,
+      "step": 5943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0912330750961764e-08,
+      "loss": 0.9957,
+      "step": 5944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.066295862130362e-08,
+      "loss": 1.0152,
+      "step": 5945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0416467368332683e-08,
+      "loss": 1.0387,
+      "step": 5946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0172857063137643e-08,
+      "loss": 0.956,
+      "step": 5947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.932127775973411e-09,
+      "loss": 1.0213,
+      "step": 5948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.694279576265563e-09,
+      "loss": 0.9397,
+      "step": 5949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.459312532608122e-09,
+      "loss": 1.0071,
+      "step": 5950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.227226712763549e-09,
+      "loss": 1.0216,
+      "step": 5951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.998022183666077e-09,
+      "loss": 0.9798,
+      "step": 5952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 0.966,
+      "step": 5953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.548257261284942e-09,
+      "loss": 1.0182,
+      "step": 5954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.327696997710855e-09,
+      "loss": 1.0013,
+      "step": 5955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.110018284304132e-09,
+      "loss": 1.0129,
+      "step": 5956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.895221183840118e-09,
+      "loss": 0.9686,
+      "step": 5957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.683305758265924e-09,
+      "loss": 0.9395,
+      "step": 5958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.474272068698219e-09,
+      "loss": 1.0094,
+      "step": 5959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.268120175421001e-09,
+      "loss": 1.0157,
+      "step": 5960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.064850137885604e-09,
+      "loss": 0.9712,
+      "step": 5961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.864462014716244e-09,
+      "loss": 1.0355,
+      "step": 5962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.666955863703361e-09,
+      "loss": 0.9703,
+      "step": 5963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.472331741805837e-09,
+      "loss": 0.9642,
+      "step": 5964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.280589705153217e-09,
+      "loss": 0.9383,
+      "step": 5965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.091729809042379e-09,
+      "loss": 1.0234,
+      "step": 5966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.905752107940866e-09,
+      "loss": 0.9323,
+      "step": 5967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.722656655482439e-09,
+      "loss": 0.9985,
+      "step": 5968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.542443504471528e-09,
+      "loss": 1.0071,
+      "step": 5969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3651127068798935e-09,
+      "loss": 0.9608,
+      "step": 5970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.190664313851068e-09,
+      "loss": 0.9861,
+      "step": 5971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.019098375692588e-09,
+      "loss": 0.9518,
+      "step": 5972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.850414941883763e-09,
+      "loss": 1.0423,
+      "step": 5973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.684614061073456e-09,
+      "loss": 0.9803,
+      "step": 5974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.521695781076752e-09,
+      "loss": 1.0208,
+      "step": 5975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.361660148879398e-09,
+      "loss": 0.9919,
+      "step": 5976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.204507210633368e-09,
+      "loss": 1.0264,
+      "step": 5977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0502370116624055e-09,
+      "loss": 0.9092,
+      "step": 5978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.898849596456477e-09,
+      "loss": 0.9242,
+      "step": 5979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.750345008675105e-09,
+      "loss": 1.0121,
+      "step": 5980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6047232911462506e-09,
+      "loss": 0.9383,
+      "step": 5981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.461984485866321e-09,
+      "loss": 0.9481,
+      "step": 5982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.322128634000166e-09,
+      "loss": 1.0021,
+      "step": 5983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1851557758832975e-09,
+      "loss": 0.9831,
+      "step": 5984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.051065951016341e-09,
+      "loss": 0.9223,
+      "step": 5985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9198591980705847e-09,
+      "loss": 1.0352,
+      "step": 5986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7915355548857604e-09,
+      "loss": 1.0205,
+      "step": 5987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6660950584689317e-09,
+      "loss": 0.9757,
+      "step": 5988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.543537744997826e-09,
+      "loss": 0.9795,
+      "step": 5989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4238636498163935e-09,
+      "loss": 1.0149,
+      "step": 5990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3070728074381376e-09,
+      "loss": 0.9661,
+      "step": 5991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.193165251545004e-09,
+      "loss": 0.9708,
+      "step": 5992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0821410149884923e-09,
+      "loss": 1.0364,
+      "step": 5993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9740001297852142e-09,
+      "loss": 0.9746,
+      "step": 5994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8687426271246646e-09,
+      "loss": 0.9334,
+      "step": 5995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7663685373614515e-09,
+      "loss": 1.0261,
+      "step": 5996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.666877890020846e-09,
+      "loss": 0.9559,
+      "step": 5997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5702707137943418e-09,
+      "loss": 1.011,
+      "step": 5998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4765470365429857e-09,
+      "loss": 1.0418,
+      "step": 5999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3857068852962674e-09,
+      "loss": 0.968,
+      "step": 6000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2977502862532298e-09,
+      "loss": 0.9307,
+      "step": 6001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2126772647780283e-09,
+      "loss": 0.9955,
+      "step": 6002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1304878454077017e-09,
+      "loss": 0.9533,
+      "step": 6003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0511820518432915e-09,
+      "loss": 1.0167,
+      "step": 6004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.74759906957612e-10,
+      "loss": 0.9713,
+      "step": 6005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.012214327897006e-10,
+      "loss": 0.992,
+      "step": 6006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.30566650548148e-10,
+      "loss": 0.974,
+      "step": 6007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.627955806088772e-10,
+      "loss": 1.018,
+      "step": 6008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.979082425173645e-10,
+      "loss": 0.9543,
+      "step": 6009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.359046549864189e-10,
+      "loss": 0.966,
+      "step": 6010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.767848358972927e-10,
+      "loss": 0.9754,
+      "step": 6011
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.205488022996808e-10,
+      "loss": 0.9918,
+      "step": 6012
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.671965704128312e-10,
+      "loss": 0.9869,
+      "step": 6013
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.167281556233249e-10,
+      "loss": 1.0176,
+      "step": 6014
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.691435724850756e-10,
+      "loss": 0.9858,
+      "step": 6015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.244428347204398e-10,
+      "loss": 0.9813,
+      "step": 6016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8262595522354774e-10,
+      "loss": 0.9972,
+      "step": 6017
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.436929460525317e-10,
+      "loss": 1.0008,
+      "step": 6018
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0764381843507709e-10,
+      "loss": 0.9454,
+      "step": 6019
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7447858276842256e-10,
+      "loss": 0.9811,
+      "step": 6020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4419724861602923e-10,
+      "loss": 1.0323,
+      "step": 6021
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.167998247131319e-10,
+      "loss": 0.9797,
+      "step": 6022
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.228631895896733e-11,
+      "loss": 0.9441,
+      "step": 6023
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.065673842454601e-11,
+      "loss": 0.9523,
+      "step": 6024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1911089347100876e-11,
+      "loss": 0.996,
+      "step": 6025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6049377132307826e-11,
+      "loss": 0.9777,
+      "step": 6026
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3071606354285735e-11,
+      "loss": 1.0598,
+      "step": 6027
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.297778075781686e-11,
+      "loss": 0.9441,
+      "step": 6028
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.767903251685525e-12,
+      "loss": 0.9924,
+      "step": 6029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4419759164496783e-12,
+      "loss": 0.9131,
+      "step": 6030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.9783,
+      "step": 6031
+    },
+    {
+      "epoch": 1.0,
+      "step": 6031,
+      "total_flos": 0.0,
+      "train_loss": 0.021359840103137396,
+      "train_runtime": 3278.7012,
+      "train_samples_per_second": 1885.172,
+      "train_steps_per_second": 1.839
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6031,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}