diff --git "a/icl-llava-pad/trainer_state.json" "b/icl-llava-pad/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/icl-llava-pad/trainer_state.json"
@@ -0,0 +1,46770 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 7790,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.547008547008549e-08,
+      "loss": 1.3952,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7094017094017097e-07,
+      "loss": 0.2336,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 1.511,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4188034188034194e-07,
+      "loss": 1.4536,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.273504273504274e-07,
+      "loss": 1.4392,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.5123,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.982905982905984e-07,
+      "loss": 1.4538,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.837606837606839e-07,
+      "loss": 1.4792,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.3551,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.547008547008548e-07,
+      "loss": 1.4268,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.401709401709402e-07,
+      "loss": 1.3948,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.3316,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 1.3181,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1965811965811968e-06,
+      "loss": 1.2263,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 1.385,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3675213675213678e-06,
+      "loss": 1.1954,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4529914529914531e-06,
+      "loss": 1.1959,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.2255,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.623931623931624e-06,
+      "loss": 1.2403,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7094017094017097e-06,
+      "loss": 1.2431,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.159,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8803418803418804e-06,
+      "loss": 1.1631,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9658119658119658e-06,
+      "loss": 0.219,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.1698,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.136752136752137e-06,
+      "loss": 1.1627,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.0729,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.1226,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3931623931623937e-06,
+      "loss": 1.0669,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.478632478632479e-06,
+      "loss": 1.1172,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.0389,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.64957264957265e-06,
+      "loss": 1.0734,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7350427350427355e-06,
+      "loss": 1.045,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.0755,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9059829059829063e-06,
+      "loss": 1.0623,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9914529914529914e-06,
+      "loss": 1.1568,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.0662,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1623931623931626e-06,
+      "loss": 1.0306,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247863247863248e-06,
+      "loss": 0.9824,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0558,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4188034188034193e-06,
+      "loss": 1.0551,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5042735042735045e-06,
+      "loss": 0.9321,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.0474,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6752136752136756e-06,
+      "loss": 1.033,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.760683760683761e-06,
+      "loss": 0.9124,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.0906,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9316239316239315e-06,
+      "loss": 1.0252,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.017094017094018e-06,
+      "loss": 0.9947,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.9895,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.188034188034188e-06,
+      "loss": 0.9927,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.273504273504274e-06,
+      "loss": 0.9334,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 0.9612,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.9058,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5299145299145306e-06,
+      "loss": 0.9945,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.9543,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.700854700854701e-06,
+      "loss": 0.9904,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.786324786324787e-06,
+      "loss": 0.2149,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.0012,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957264957264958e-06,
+      "loss": 0.2569,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.042735042735043e-06,
+      "loss": 0.99,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.9787,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.213675213675214e-06,
+      "loss": 0.9926,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2991452991453e-06,
+      "loss": 1.0269,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 0.9744,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.470085470085471e-06,
+      "loss": 1.0017,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.9607,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.8694,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.726495726495727e-06,
+      "loss": 0.9049,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8119658119658126e-06,
+      "loss": 0.9216,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 0.9749,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.982905982905983e-06,
+      "loss": 0.9633,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0683760683760684e-06,
+      "loss": 0.9193,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.9663,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.23931623931624e-06,
+      "loss": 0.97,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.324786324786325e-06,
+      "loss": 0.9765,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 0.9547,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.495726495726496e-06,
+      "loss": 0.9553,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.581196581196582e-06,
+      "loss": 0.9333,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8872,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.752136752136753e-06,
+      "loss": 0.9629,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.837606837606839e-06,
+      "loss": 0.9099,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.8922,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.008547008547009e-06,
+      "loss": 0.884,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0940170940170945e-06,
+      "loss": 0.8633,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.9125,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.264957264957266e-06,
+      "loss": 0.8985,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.350427350427351e-06,
+      "loss": 0.9054,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.9288,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.521367521367522e-06,
+      "loss": 0.9365,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.606837606837607e-06,
+      "loss": 0.9258,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.9053,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.866,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.863247863247863e-06,
+      "loss": 0.9109,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 0.9418,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.034188034188036e-06,
+      "loss": 0.9337,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.11965811965812e-06,
+      "loss": 0.9609,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.9276,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.290598290598293e-06,
+      "loss": 0.8613,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.376068376068377e-06,
+      "loss": 0.94,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.8894,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.547008547008548e-06,
+      "loss": 0.9627,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.632478632478633e-06,
+      "loss": 0.9559,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.9068,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.803418803418804e-06,
+      "loss": 0.898,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.919,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.949,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.059829059829061e-06,
+      "loss": 0.9243,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.145299145299145e-06,
+      "loss": 0.8775,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9103,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.316239316239318e-06,
+      "loss": 0.9322,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.401709401709402e-06,
+      "loss": 0.943,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.9056,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.572649572649575e-06,
+      "loss": 0.8838,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.658119658119659e-06,
+      "loss": 0.8943,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.9079,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.82905982905983e-06,
+      "loss": 0.9568,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.914529914529915e-06,
+      "loss": 0.9642,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.8845,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0085470085470086e-05,
+      "loss": 0.9409,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0170940170940172e-05,
+      "loss": 0.9282,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.8973,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0341880341880343e-05,
+      "loss": 0.9182,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0427350427350429e-05,
+      "loss": 0.936,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.8937,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.05982905982906e-05,
+      "loss": 0.9104,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0683760683760684e-05,
+      "loss": 0.9669,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9099,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0854700854700855e-05,
+      "loss": 0.9235,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0940170940170942e-05,
+      "loss": 0.9005,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 0.911,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.9015,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1196581196581197e-05,
+      "loss": 0.7826,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.892,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1367521367521368e-05,
+      "loss": 0.9353,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1452991452991454e-05,
+      "loss": 0.9129,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.891,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1623931623931625e-05,
+      "loss": 0.8618,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.170940170940171e-05,
+      "loss": 0.9424,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.8621,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1880341880341882e-05,
+      "loss": 0.2361,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1965811965811966e-05,
+      "loss": 0.9687,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.9163,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2136752136752137e-05,
+      "loss": 0.9078,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 0.9193,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.8752,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2393162393162395e-05,
+      "loss": 0.8625,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.247863247863248e-05,
+      "loss": 0.9201,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.9183,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.264957264957265e-05,
+      "loss": 0.8514,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2735042735042736e-05,
+      "loss": 0.9559,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.9299,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2905982905982907e-05,
+      "loss": 0.9334,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2991452991452993e-05,
+      "loss": 0.8767,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.9194,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3162393162393164e-05,
+      "loss": 0.9292,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3247863247863248e-05,
+      "loss": 0.8717,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9279,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3418803418803419e-05,
+      "loss": 0.908,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3504273504273506e-05,
+      "loss": 0.8903,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 0.949,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3675213675213677e-05,
+      "loss": 0.8526,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3760683760683761e-05,
+      "loss": 0.958,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.8663,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3931623931623932e-05,
+      "loss": 0.9146,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4017094017094018e-05,
+      "loss": 0.9661,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 0.8471,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4188034188034189e-05,
+      "loss": 0.8714,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4273504273504275e-05,
+      "loss": 0.9221,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.8605,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.9238,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4529914529914531e-05,
+      "loss": 0.9225,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.8981,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4700854700854703e-05,
+      "loss": 0.8715,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4786324786324788e-05,
+      "loss": 0.8375,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.8576,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.495726495726496e-05,
+      "loss": 0.942,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5042735042735043e-05,
+      "loss": 0.9218,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.8908,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5213675213675214e-05,
+      "loss": 0.2441,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5299145299145298e-05,
+      "loss": 0.8776,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.8961,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5470085470085473e-05,
+      "loss": 0.9953,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.9198,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 0.9213,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5726495726495726e-05,
+      "loss": 0.9276,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.581196581196581e-05,
+      "loss": 0.876,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.9116,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5982905982905986e-05,
+      "loss": 0.8994,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6068376068376072e-05,
+      "loss": 0.874,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.9265,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.623931623931624e-05,
+      "loss": 0.9372,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6324786324786325e-05,
+      "loss": 0.925,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.8628,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6495726495726496e-05,
+      "loss": 0.8954,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6581196581196585e-05,
+      "loss": 0.8537,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.8549,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6752136752136753e-05,
+      "loss": 0.9142,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.683760683760684e-05,
+      "loss": 0.9162,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.2572,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.700854700854701e-05,
+      "loss": 0.9507,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7094017094017095e-05,
+      "loss": 0.9242,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 0.8474,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7264957264957267e-05,
+      "loss": 0.8531,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7350427350427352e-05,
+      "loss": 0.8665,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.8827,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7521367521367523e-05,
+      "loss": 0.876,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.760683760683761e-05,
+      "loss": 0.9416,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.2449,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.8564,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7863247863247866e-05,
+      "loss": 0.841,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.8921,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8034188034188037e-05,
+      "loss": 0.8901,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8119658119658122e-05,
+      "loss": 0.8796,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 0.9414,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.829059829059829e-05,
+      "loss": 0.8645,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8376068376068376e-05,
+      "loss": 0.9856,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.8488,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.854700854700855e-05,
+      "loss": 0.9567,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8632478632478636e-05,
+      "loss": 0.8306,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 0.8733,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8803418803418804e-05,
+      "loss": 0.8891,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.8884,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.9374,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.905982905982906e-05,
+      "loss": 0.8892,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.914529914529915e-05,
+      "loss": 0.844,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.8877,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9316239316239317e-05,
+      "loss": 0.8825,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9401709401709403e-05,
+      "loss": 0.8821,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.9534,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9572649572649574e-05,
+      "loss": 0.8881,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.965811965811966e-05,
+      "loss": 0.9245,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 0.9066,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982905982905983e-05,
+      "loss": 0.955,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9914529914529916e-05,
+      "loss": 0.967,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.22,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999135657546e-05,
+      "loss": 0.8674,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996542630333e-05,
+      "loss": 0.8895,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999222091881e-05,
+      "loss": 0.8643,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998617052372e-05,
+      "loss": 0.92,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999978391446114e-05,
+      "loss": 0.9308,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968883687337e-05,
+      "loss": 0.2456,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999957647249026e-05,
+      "loss": 0.9133,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999994468213313e-05,
+      "loss": 0.87,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999929988341886e-05,
+      "loss": 0.9514,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991356587784e-05,
+      "loss": 0.8816,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999895414743826e-05,
+      "loss": 0.9264,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987553494298e-05,
+      "loss": 0.7903,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999853926478745e-05,
+      "loss": 0.7522,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999830589354852e-05,
+      "loss": 0.8711,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999805523575335e-05,
+      "loss": 0.8664,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999977872914453e-05,
+      "loss": 0.8952,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999750206067064e-05,
+      "loss": 0.8498,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999971995434787e-05,
+      "loss": 0.9114,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999687973992183e-05,
+      "loss": 0.963,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999654265005528e-05,
+      "loss": 0.9521,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999618827393726e-05,
+      "loss": 0.8925,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999958166116291e-05,
+      "loss": 0.8774,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999542766319505e-05,
+      "loss": 0.9198,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999502142870232e-05,
+      "loss": 0.8679,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999945979082211e-05,
+      "loss": 0.8763,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999415710182467e-05,
+      "loss": 0.9077,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999369900958923e-05,
+      "loss": 0.8047,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999322363159393e-05,
+      "loss": 0.8351,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999273096792094e-05,
+      "loss": 0.9074,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999922210186555e-05,
+      "loss": 0.7997,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999916937838857e-05,
+      "loss": 0.8806,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999114926370266e-05,
+      "loss": 0.8522,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999058745820057e-05,
+      "loss": 0.8373,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999000836747653e-05,
+      "loss": 0.8637,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998941199163065e-05,
+      "loss": 0.9021,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99988798330766e-05,
+      "loss": 0.7955,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999881673849887e-05,
+      "loss": 0.887,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999875191544078e-05,
+      "loss": 0.8683,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999868536391353e-05,
+      "loss": 0.9225,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998617083928637e-05,
+      "loss": 0.8793,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998547075497897e-05,
+      "loss": 0.9041,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999847533863341e-05,
+      "loss": 0.9557,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998401873347585e-05,
+      "loss": 0.8196,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998326679653114e-05,
+      "loss": 0.9265,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998249757563004e-05,
+      "loss": 0.9156,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998171107090537e-05,
+      "loss": 0.819,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999809072824933e-05,
+      "loss": 0.8585,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999800862105326e-05,
+      "loss": 0.8933,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999792478551653e-05,
+      "loss": 0.9133,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999783922165363e-05,
+      "loss": 0.8764,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997751929479353e-05,
+      "loss": 0.9261,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999766290900879e-05,
+      "loss": 0.9515,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997572160257322e-05,
+      "loss": 0.9144,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997479683240647e-05,
+      "loss": 0.9141,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997385477974744e-05,
+      "loss": 0.9092,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997289544475905e-05,
+      "loss": 0.893,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997191882760706e-05,
+      "loss": 0.8734,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997092492846036e-05,
+      "loss": 0.9777,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996991374749074e-05,
+      "loss": 0.8539,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996888528487294e-05,
+      "loss": 0.944,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996783954078486e-05,
+      "loss": 0.8881,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996677651540723e-05,
+      "loss": 0.8707,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996569620892375e-05,
+      "loss": 0.8879,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996459862152128e-05,
+      "loss": 0.9187,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996348375338952e-05,
+      "loss": 0.9455,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996235160472116e-05,
+      "loss": 0.9163,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996120217571195e-05,
+      "loss": 0.818,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996003546656055e-05,
+      "loss": 0.8812,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995885147746868e-05,
+      "loss": 0.8825,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99957650208641e-05,
+      "loss": 0.8846,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995643166028517e-05,
+      "loss": 0.8806,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995519583261186e-05,
+      "loss": 0.9373,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999539427258347e-05,
+      "loss": 0.9642,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999526723401703e-05,
+      "loss": 0.8978,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999513846758383e-05,
+      "loss": 0.872,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995007973306122e-05,
+      "loss": 0.8443,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994875751206472e-05,
+      "loss": 0.8869,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994741801307736e-05,
+      "loss": 0.9048,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994606123633064e-05,
+      "loss": 0.8606,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999446871820592e-05,
+      "loss": 0.9178,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999432958505005e-05,
+      "loss": 0.8596,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994188724189503e-05,
+      "loss": 0.9061,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999404613564864e-05,
+      "loss": 0.8737,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993901819452102e-05,
+      "loss": 0.8307,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999375577562484e-05,
+      "loss": 0.9286,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993608004192094e-05,
+      "loss": 0.8342,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999345850517942e-05,
+      "loss": 0.9185,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993307278612653e-05,
+      "loss": 0.8766,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993154324517937e-05,
+      "loss": 0.8732,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992999642921714e-05,
+      "loss": 0.8281,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992843233850723e-05,
+      "loss": 0.2324,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992685097332005e-05,
+      "loss": 0.9028,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992525233392895e-05,
+      "loss": 0.9409,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992363642061027e-05,
+      "loss": 0.8729,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992200323364336e-05,
+      "loss": 0.8641,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992035277331055e-05,
+      "loss": 0.8566,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991868503989713e-05,
+      "loss": 0.9046,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991700003369145e-05,
+      "loss": 0.8952,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991529775498476e-05,
+      "loss": 0.9148,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991357820407133e-05,
+      "loss": 0.9488,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991184138124838e-05,
+      "loss": 0.8916,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991008728681622e-05,
+      "loss": 0.9252,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990831592107804e-05,
+      "loss": 0.8664,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990652728434006e-05,
+      "loss": 0.8533,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999047213769115e-05,
+      "loss": 0.9624,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999028981991045e-05,
+      "loss": 0.873,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999010577512343e-05,
+      "loss": 0.9401,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99899200033619e-05,
+      "loss": 0.8489,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998973250465797e-05,
+      "loss": 0.8939,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989543279044064e-05,
+      "loss": 0.2555,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989352326552886e-05,
+      "loss": 0.958,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998915964721744e-05,
+      "loss": 0.9378,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988965241071047e-05,
+      "loss": 0.9047,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988769108147306e-05,
+      "loss": 0.9265,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988571248480125e-05,
+      "loss": 0.8857,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988371662103704e-05,
+      "loss": 0.8699,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998817034905255e-05,
+      "loss": 0.887,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998796730936146e-05,
+      "loss": 0.9148,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987762543065536e-05,
+      "loss": 0.9358,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998755605020017e-05,
+      "loss": 0.8507,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987347830801068e-05,
+      "loss": 0.9049,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987137884904216e-05,
+      "loss": 0.8575,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986926212545907e-05,
+      "loss": 0.8957,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998671281376274e-05,
+      "loss": 0.9192,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986497688591597e-05,
+      "loss": 0.8985,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998628083706967e-05,
+      "loss": 0.8612,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986062259234448e-05,
+      "loss": 0.8194,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998584195512371e-05,
+      "loss": 0.9289,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985619924775545e-05,
+      "loss": 0.866,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985396168228335e-05,
+      "loss": 0.8482,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985170685520753e-05,
+      "loss": 0.8878,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998494347669179e-05,
+      "loss": 0.893,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984714541780715e-05,
+      "loss": 0.932,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984483880827104e-05,
+      "loss": 0.8925,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984251493870834e-05,
+      "loss": 0.8977,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984017380952076e-05,
+      "loss": 0.8327,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99837815421113e-05,
+      "loss": 0.81,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983543977389272e-05,
+      "loss": 0.8322,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983304686827067e-05,
+      "loss": 0.8325,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983063670466047e-05,
+      "loss": 0.8692,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982820928347876e-05,
+      "loss": 0.8942,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982576460514515e-05,
+      "loss": 0.8951,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998233026700823e-05,
+      "loss": 0.9461,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982082347871573e-05,
+      "loss": 0.8357,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981832703147406e-05,
+      "loss": 0.9291,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981581332878883e-05,
+      "loss": 0.8948,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998132823710946e-05,
+      "loss": 0.9331,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981073415882884e-05,
+      "loss": 0.8993,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980816869243216e-05,
+      "loss": 0.922,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998055859723479e-05,
+      "loss": 0.8908,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998029859990227e-05,
+      "loss": 0.8752,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980036877290587e-05,
+      "loss": 0.8879,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979773429444994e-05,
+      "loss": 0.7767,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979508256411028e-05,
+      "loss": 0.9281,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979241358234525e-05,
+      "loss": 0.8985,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997897273496163e-05,
+      "loss": 0.8465,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978702386638785e-05,
+      "loss": 0.8562,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997843031331271e-05,
+      "loss": 0.8558,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978156515030453e-05,
+      "loss": 0.8318,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997788099183933e-05,
+      "loss": 0.8131,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977603743786985e-05,
+      "loss": 0.9095,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977324770921333e-05,
+      "loss": 0.9187,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977044073290606e-05,
+      "loss": 0.9321,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976761650943327e-05,
+      "loss": 0.8872,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976477503928318e-05,
+      "loss": 0.8463,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99761916322947e-05,
+      "loss": 0.9496,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975904036091888e-05,
+      "loss": 0.944,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99756147153696e-05,
+      "loss": 0.8728,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975323670177853e-05,
+      "loss": 0.8451,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975030900566957e-05,
+      "loss": 0.9403,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974736406587523e-05,
+      "loss": 0.8491,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997444018829046e-05,
+      "loss": 0.2816,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997414224572697e-05,
+      "loss": 0.8744,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973842578948564e-05,
+      "loss": 0.8882,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997354118800705e-05,
+      "loss": 0.9145,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973238072954515e-05,
+      "loss": 0.8487,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972933233843367e-05,
+      "loss": 0.8837,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972626670726302e-05,
+      "loss": 0.2441,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972318383656314e-05,
+      "loss": 0.9494,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972008372686697e-05,
+      "loss": 0.8651,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971696637871042e-05,
+      "loss": 0.8745,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971383179263237e-05,
+      "loss": 0.8569,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971067996917473e-05,
+      "loss": 0.8808,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997075109088823e-05,
+      "loss": 0.8728,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970432461230292e-05,
+      "loss": 0.7646,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970112107998744e-05,
+      "loss": 0.8561,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969790031248955e-05,
+      "loss": 0.8586,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969466231036615e-05,
+      "loss": 0.8913,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969140707417693e-05,
+      "loss": 0.9115,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968813460448462e-05,
+      "loss": 0.9542,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996848449018549e-05,
+      "loss": 0.8543,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996815379668565e-05,
+      "loss": 0.9371,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996782138000611e-05,
+      "loss": 0.8675,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967487240204323e-05,
+      "loss": 0.864,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967151377338064e-05,
+      "loss": 0.8829,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966813791465386e-05,
+      "loss": 0.9439,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996647448264465e-05,
+      "loss": 0.8696,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966133450934512e-05,
+      "loss": 0.8934,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996579069639392e-05,
+      "loss": 0.8665,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965446219082137e-05,
+      "loss": 0.7797,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965100019058702e-05,
+      "loss": 0.8861,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964752096383467e-05,
+      "loss": 0.8236,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996440245111657e-05,
+      "loss": 0.8819,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964051083318464e-05,
+      "loss": 0.9013,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963697993049886e-05,
+      "loss": 0.8637,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963343180371868e-05,
+      "loss": 0.8871,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962986645345753e-05,
+      "loss": 0.9722,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996262838803317e-05,
+      "loss": 0.8231,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962268408496055e-05,
+      "loss": 0.9095,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961906706796633e-05,
+      "loss": 0.896,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961543282997434e-05,
+      "loss": 0.8714,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996117813716128e-05,
+      "loss": 0.8833,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960811269351294e-05,
+      "loss": 0.903,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960442679630896e-05,
+      "loss": 0.8539,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960072368063804e-05,
+      "loss": 0.8749,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959700334714032e-05,
+      "loss": 0.9005,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959326579645894e-05,
+      "loss": 0.8726,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958951102924003e-05,
+      "loss": 0.8334,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995857390461326e-05,
+      "loss": 0.8316,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958194984778877e-05,
+      "loss": 0.86,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957814343486353e-05,
+      "loss": 0.8226,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957431980801494e-05,
+      "loss": 0.7888,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957047896790394e-05,
+      "loss": 0.9701,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956662091519454e-05,
+      "loss": 0.8879,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995627456505536e-05,
+      "loss": 0.8092,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995588531746511e-05,
+      "loss": 0.9016,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995549434881599e-05,
+      "loss": 0.9005,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955101659175586e-05,
+      "loss": 0.9248,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995470724861178e-05,
+      "loss": 0.8689,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954311117192758e-05,
+      "loss": 0.8987,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995391326498699e-05,
+      "loss": 0.8641,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953513692063265e-05,
+      "loss": 0.8532,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953112398490645e-05,
+      "loss": 0.9151,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952709384338504e-05,
+      "loss": 0.9044,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952304649676517e-05,
+      "loss": 0.8983,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951898194574638e-05,
+      "loss": 0.8417,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995149001910314e-05,
+      "loss": 0.8321,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951080123332583e-05,
+      "loss": 0.8604,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950668507333817e-05,
+      "loss": 0.8303,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950255171178008e-05,
+      "loss": 0.8405,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99498401149366e-05,
+      "loss": 0.9408,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949423338681352e-05,
+      "loss": 0.8541,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99490048424843e-05,
+      "loss": 0.8794,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948584626417802e-05,
+      "loss": 0.2441,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994816269055449e-05,
+      "loss": 0.8865,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947739034967308e-05,
+      "loss": 0.8216,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994731365972949e-05,
+      "loss": 0.8726,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946886564914572e-05,
+      "loss": 0.8185,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946457750596385e-05,
+      "loss": 0.9221,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946027216849056e-05,
+      "loss": 0.8527,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945594963747014e-05,
+      "loss": 0.8615,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994516099136498e-05,
+      "loss": 0.8183,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944725299777975e-05,
+      "loss": 0.868,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944287889061312e-05,
+      "loss": 0.871,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943848759290612e-05,
+      "loss": 0.8967,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943407910541782e-05,
+      "loss": 0.9226,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942965342891032e-05,
+      "loss": 0.8266,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942521056414866e-05,
+      "loss": 0.8791,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942075051190096e-05,
+      "loss": 0.9032,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994162732729381e-05,
+      "loss": 0.2402,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941177884803415e-05,
+      "loss": 0.8692,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99407267237966e-05,
+      "loss": 0.8804,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994027384435136e-05,
+      "loss": 0.8799,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993981924654598e-05,
+      "loss": 0.8588,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993936293045905e-05,
+      "loss": 0.8315,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938904896169444e-05,
+      "loss": 0.8588,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938445143756353e-05,
+      "loss": 0.8754,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937983673299247e-05,
+      "loss": 0.8556,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99375204848779e-05,
+      "loss": 0.7996,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937055578572386e-05,
+      "loss": 0.9026,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993658895446307e-05,
+      "loss": 0.251,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936120612630615e-05,
+      "loss": 0.8471,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935650553155987e-05,
+      "loss": 0.8879,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993517877612044e-05,
+      "loss": 0.8806,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934705281605537e-05,
+      "loss": 0.859,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993423006969312e-05,
+      "loss": 0.849,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933753140465344e-05,
+      "loss": 0.8999,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993327449400465e-05,
+      "loss": 0.8598,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993279413039379e-05,
+      "loss": 0.8105,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99323120497158e-05,
+      "loss": 0.8367,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931828252054014e-05,
+      "loss": 0.29,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931342737492066e-05,
+      "loss": 0.915,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930855506113885e-05,
+      "loss": 0.8768,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99303665580037e-05,
+      "loss": 0.8391,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929875893246034e-05,
+      "loss": 0.8974,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992938351192571e-05,
+      "loss": 0.804,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992888941412784e-05,
+      "loss": 0.8245,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928393599937842e-05,
+      "loss": 0.2635,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927896069441426e-05,
+      "loss": 0.9372,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927396822724603e-05,
+      "loss": 0.2519,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926895859873672e-05,
+      "loss": 0.8549,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926393180975234e-05,
+      "loss": 0.8823,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925888786116183e-05,
+      "loss": 0.8535,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992538267538372e-05,
+      "loss": 0.8309,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924874848865332e-05,
+      "loss": 0.8986,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924365306648806e-05,
+      "loss": 0.869,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992385404882223e-05,
+      "loss": 0.8708,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923341075473978e-05,
+      "loss": 0.8547,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922826386692732e-05,
+      "loss": 0.8219,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992230998256746e-05,
+      "loss": 0.8922,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992179186318744e-05,
+      "loss": 0.8777,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992127202864223e-05,
+      "loss": 0.89,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920750479021703e-05,
+      "loss": 0.8407,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920227214416006e-05,
+      "loss": 0.8794,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919702234915603e-05,
+      "loss": 0.869,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919175540611246e-05,
+      "loss": 0.8252,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918647131593983e-05,
+      "loss": 0.2441,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991811700795516e-05,
+      "loss": 0.847,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917585169786412e-05,
+      "loss": 0.7858,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991705161717969e-05,
+      "loss": 0.7651,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916516350227217e-05,
+      "loss": 0.8866,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991597936902153e-05,
+      "loss": 0.8778,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915440673655452e-05,
+      "loss": 0.849,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914900264222107e-05,
+      "loss": 0.9663,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991435814081492e-05,
+      "loss": 0.8532,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99138143035276e-05,
+      "loss": 0.8529,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913268752454167e-05,
+      "loss": 0.8452,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991272148768892e-05,
+      "loss": 0.8948,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912172509326473e-05,
+      "loss": 0.8154,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991162181746172e-05,
+      "loss": 0.7642,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911069412189866e-05,
+      "loss": 0.8889,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910515293606396e-05,
+      "loss": 0.8913,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909959461807108e-05,
+      "loss": 0.9634,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990940191688808e-05,
+      "loss": 0.8456,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908842658945696e-05,
+      "loss": 0.8653,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908281688076635e-05,
+      "loss": 0.8657,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907719004377877e-05,
+      "loss": 0.8353,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907154607946682e-05,
+      "loss": 0.8801,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990658849888062e-05,
+      "loss": 0.8403,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990602067727756e-05,
+      "loss": 0.9283,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990545114323565e-05,
+      "loss": 0.8585,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990487989685335e-05,
+      "loss": 0.8791,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904306938229415e-05,
+      "loss": 0.8216,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903732267462882e-05,
+      "loss": 0.9313,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99031558846531e-05,
+      "loss": 0.812,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902577789899703e-05,
+      "loss": 0.8314,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990199798330263e-05,
+      "loss": 0.9042,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990141646496211e-05,
+      "loss": 0.8843,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900833234978667e-05,
+      "loss": 0.9317,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900248293453122e-05,
+      "loss": 0.9323,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98996616404866e-05,
+      "loss": 0.8006,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899073276180507e-05,
+      "loss": 0.8469,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898483200636555e-05,
+      "loss": 0.8433,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897891413956755e-05,
+      "loss": 0.8793,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897297916243398e-05,
+      "loss": 0.9296,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989670270759909e-05,
+      "loss": 0.892,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989610578812672e-05,
+      "loss": 0.7835,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895507157929478e-05,
+      "loss": 0.8531,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894906817110842e-05,
+      "loss": 0.8656,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894304765774605e-05,
+      "loss": 0.8933,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989370100402483e-05,
+      "loss": 0.8688,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893095531965894e-05,
+      "loss": 0.8059,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892488349702462e-05,
+      "loss": 0.9174,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891879457339503e-05,
+      "loss": 0.917,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891268854982266e-05,
+      "loss": 0.8407,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890656542736312e-05,
+      "loss": 0.8394,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890042520707483e-05,
+      "loss": 0.8482,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889426789001935e-05,
+      "loss": 0.8633,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98888093477261e-05,
+      "loss": 0.8528,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888190196986717e-05,
+      "loss": 0.2501,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988756933689082e-05,
+      "loss": 0.9125,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886946767545726e-05,
+      "loss": 0.9504,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988632248905907e-05,
+      "loss": 0.8495,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885696501538766e-05,
+      "loss": 0.7995,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885068805093025e-05,
+      "loss": 0.7995,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884439399830358e-05,
+      "loss": 0.8734,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988380828585957e-05,
+      "loss": 0.7904,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883175463289754e-05,
+      "loss": 0.8369,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882540932230317e-05,
+      "loss": 0.8658,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881904692790942e-05,
+      "loss": 0.9095,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881266745081615e-05,
+      "loss": 0.2616,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880627089212615e-05,
+      "loss": 0.888,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879985725294524e-05,
+      "loss": 0.856,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879342653438213e-05,
+      "loss": 0.248,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878697873754846e-05,
+      "loss": 0.8292,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878051386355886e-05,
+      "loss": 0.858,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987740319135309e-05,
+      "loss": 0.8376,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987675328885851e-05,
+      "loss": 0.8983,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987610167898449e-05,
+      "loss": 0.9003,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875448361843684e-05,
+      "loss": 0.8799,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874793337549022e-05,
+      "loss": 0.8624,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874136606213734e-05,
+      "loss": 0.8031,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873478167951358e-05,
+      "loss": 0.8146,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872818022875707e-05,
+      "loss": 0.8053,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987215617110091e-05,
+      "loss": 0.8906,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871492612741367e-05,
+      "loss": 0.7891,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870827347911793e-05,
+      "loss": 0.8936,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870160376727196e-05,
+      "loss": 0.9252,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869491699302868e-05,
+      "loss": 0.8615,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868821315754404e-05,
+      "loss": 0.7854,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868149226197693e-05,
+      "loss": 0.863,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867475430748923e-05,
+      "loss": 0.8363,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986679992952456e-05,
+      "loss": 0.9647,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866122722641385e-05,
+      "loss": 0.9571,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865443810216466e-05,
+      "loss": 0.925,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986476319236716e-05,
+      "loss": 0.7823,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864080869211137e-05,
+      "loss": 0.8999,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986339684086633e-05,
+      "loss": 0.8327,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862711107451004e-05,
+      "loss": 0.8829,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986202366908369e-05,
+      "loss": 0.8865,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861334525883223e-05,
+      "loss": 0.8595,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860643677968744e-05,
+      "loss": 0.81,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859951125459673e-05,
+      "loss": 0.848,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859256868475733e-05,
+      "loss": 0.81,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858560907136932e-05,
+      "loss": 0.8773,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985786324156359e-05,
+      "loss": 0.8952,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98571638718763e-05,
+      "loss": 0.8961,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856462798195973e-05,
+      "loss": 0.9164,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855760020643797e-05,
+      "loss": 0.8273,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855055539341256e-05,
+      "loss": 0.8742,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985434935441014e-05,
+      "loss": 0.7851,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985364146597252e-05,
+      "loss": 0.8723,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985293187415077e-05,
+      "loss": 0.8372,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985222057906756e-05,
+      "loss": 0.8913,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985150758084584e-05,
+      "loss": 0.8834,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850792879608877e-05,
+      "loss": 0.789,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850076475480216e-05,
+      "loss": 0.8982,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849358368583695e-05,
+      "loss": 0.8715,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848638559043462e-05,
+      "loss": 0.8642,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847917046983942e-05,
+      "loss": 0.9551,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847193832529866e-05,
+      "loss": 0.8655,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984646891580625e-05,
+      "loss": 0.8345,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984574229693841e-05,
+      "loss": 0.9316,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984501397605196e-05,
+      "loss": 0.9435,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844283953272806e-05,
+      "loss": 0.8544,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984355222872714e-05,
+      "loss": 0.8602,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984281880254145e-05,
+      "loss": 0.2645,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984208367484253e-05,
+      "loss": 0.9089,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984134684575746e-05,
+      "loss": 0.8775,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840608315413616e-05,
+      "loss": 0.8786,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983986808393866e-05,
+      "loss": 0.8524,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839126151460556e-05,
+      "loss": 0.8395,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838382518107564e-05,
+      "loss": 0.8544,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837637184008234e-05,
+      "loss": 0.2681,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983689014929141e-05,
+      "loss": 0.836,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983614141408623e-05,
+      "loss": 0.8752,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983539097852213e-05,
+      "loss": 0.9194,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834638842728836e-05,
+      "loss": 0.8376,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983388500683636e-05,
+      "loss": 0.8387,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983312947097503e-05,
+      "loss": 0.8955,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832372235275446e-05,
+      "loss": 0.9173,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831613299868512e-05,
+      "loss": 0.8197,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830852664885426e-05,
+      "loss": 0.8825,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830090330457674e-05,
+      "loss": 0.7959,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829326296717042e-05,
+      "loss": 0.8658,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828560563795608e-05,
+      "loss": 0.8355,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827793131825738e-05,
+      "loss": 0.8815,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827024000940106e-05,
+      "loss": 0.8725,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982625317127166e-05,
+      "loss": 0.8633,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982548064295366e-05,
+      "loss": 0.8847,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824706416119653e-05,
+      "loss": 0.8759,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823930490903468e-05,
+      "loss": 0.8252,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982315286743925e-05,
+      "loss": 0.874,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982237354586142e-05,
+      "loss": 0.9123,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821592526304692e-05,
+      "loss": 0.8553,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820809808904094e-05,
+      "loss": 0.9316,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982002539379492e-05,
+      "loss": 0.836,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819239281112778e-05,
+      "loss": 0.8759,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981845147099356e-05,
+      "loss": 0.8415,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817661963573454e-05,
+      "loss": 0.8928,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981687075898894e-05,
+      "loss": 0.8575,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816077857376794e-05,
+      "loss": 0.8656,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815283258874082e-05,
+      "loss": 0.9119,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814486963618166e-05,
+      "loss": 0.8649,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813688971746697e-05,
+      "loss": 0.8416,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981288928339763e-05,
+      "loss": 0.8046,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98120878987092e-05,
+      "loss": 0.9104,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981128481781994e-05,
+      "loss": 0.8949,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810480040868684e-05,
+      "loss": 0.7903,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980967356799455e-05,
+      "loss": 0.8389,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808865399336946e-05,
+      "loss": 0.8534,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808055535035582e-05,
+      "loss": 0.8007,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807243975230466e-05,
+      "loss": 0.8736,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980643072006188e-05,
+      "loss": 0.8124,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805615769670413e-05,
+      "loss": 0.8587,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980479912419695e-05,
+      "loss": 0.8682,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980398078378266e-05,
+      "loss": 0.9221,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803160748569e-05,
+      "loss": 0.9188,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802339018697743e-05,
+      "loss": 0.8254,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980151559431093e-05,
+      "loss": 0.8666,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980069047555091e-05,
+      "loss": 0.8308,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799863662560317e-05,
+      "loss": 0.8386,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979903515548208e-05,
+      "loss": 0.8366,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979820495445943e-05,
+      "loss": 0.8492,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979737305963587e-05,
+      "loss": 0.8606,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796539471155215e-05,
+      "loss": 0.8401,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795704189161568e-05,
+      "loss": 0.8795,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979486721379932e-05,
+      "loss": 0.8404,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979402854521316e-05,
+      "loss": 0.8438,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793188183548068e-05,
+      "loss": 0.8544,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979234612894931e-05,
+      "loss": 0.8736,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791502381562455e-05,
+      "loss": 0.8235,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979065694153336e-05,
+      "loss": 0.8633,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789809809008177e-05,
+      "loss": 0.9273,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978896098413334e-05,
+      "loss": 0.7818,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97881104670556e-05,
+      "loss": 0.8612,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787258257921974e-05,
+      "loss": 0.9224,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786404356879777e-05,
+      "loss": 0.9181,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978554876407663e-05,
+      "loss": 0.8227,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978469147966044e-05,
+      "loss": 0.9246,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783832503779397e-05,
+      "loss": 0.9192,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782971836581995e-05,
+      "loss": 0.8566,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782109478217014e-05,
+      "loss": 0.8323,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781245428833533e-05,
+      "loss": 0.9254,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780379688580913e-05,
+      "loss": 0.8434,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779512257608817e-05,
+      "loss": 0.8979,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778643136067195e-05,
+      "loss": 0.8737,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777772324106294e-05,
+      "loss": 0.8723,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776899821876646e-05,
+      "loss": 0.8344,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977602562952908e-05,
+      "loss": 0.8374,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775149747214716e-05,
+      "loss": 0.8722,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774272175084968e-05,
+      "loss": 0.8826,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977339291329154e-05,
+      "loss": 0.9044,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772511961986425e-05,
+      "loss": 0.9065,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977162932132192e-05,
+      "loss": 0.8964,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770744991450602e-05,
+      "loss": 0.8703,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769858972525338e-05,
+      "loss": 0.8992,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97689712646993e-05,
+      "loss": 0.9222,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768081868125944e-05,
+      "loss": 0.9121,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767190782959013e-05,
+      "loss": 0.7819,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766298009352554e-05,
+      "loss": 0.8146,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765403547460894e-05,
+      "loss": 0.8679,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764507397438663e-05,
+      "loss": 0.9005,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763609559440775e-05,
+      "loss": 0.8292,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762710033622438e-05,
+      "loss": 0.8949,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761808820139146e-05,
+      "loss": 0.809,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760905919146696e-05,
+      "loss": 0.7731,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760001330801172e-05,
+      "loss": 0.9114,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759095055258947e-05,
+      "loss": 0.8347,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758187092676685e-05,
+      "loss": 0.856,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975727744321135e-05,
+      "loss": 0.8259,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975636610702019e-05,
+      "loss": 0.2503,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975545308426074e-05,
+      "loss": 0.8216,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975453837509084e-05,
+      "loss": 0.8162,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753621979668614e-05,
+      "loss": 0.8067,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752703898152473e-05,
+      "loss": 0.8761,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975178413070113e-05,
+      "loss": 0.8659,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975086267747358e-05,
+      "loss": 0.8607,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749939538629114e-05,
+      "loss": 0.8537,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974901471432731e-05,
+      "loss": 0.9206,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748088204728052e-05,
+      "loss": 0.9354,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747160009991492e-05,
+      "loss": 0.8706,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746230130278095e-05,
+      "loss": 0.8464,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97452985657486e-05,
+      "loss": 0.8226,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974436531656405e-05,
+      "loss": 0.9347,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743430382885774e-05,
+      "loss": 0.8128,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742493764875393e-05,
+      "loss": 0.8285,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741555462694815e-05,
+      "loss": 0.8844,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740615476506248e-05,
+      "loss": 0.8557,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973967380647218e-05,
+      "loss": 0.8234,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738730452755404e-05,
+      "loss": 0.2578,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973778541551899e-05,
+      "loss": 0.8952,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736838694926306e-05,
+      "loss": 0.8313,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973589029114101e-05,
+      "loss": 0.8064,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734940204327057e-05,
+      "loss": 0.8973,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973398843464868e-05,
+      "loss": 0.8269,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733034982270406e-05,
+      "loss": 0.8706,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973207984735707e-05,
+      "loss": 0.8972,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731123030073777e-05,
+      "loss": 0.8644,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730164530585932e-05,
+      "loss": 0.869,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729204349059227e-05,
+      "loss": 0.8569,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972824248565965e-05,
+      "loss": 0.78,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727278940553475e-05,
+      "loss": 0.8503,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972631371390727e-05,
+      "loss": 0.9037,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972534680588789e-05,
+      "loss": 0.8213,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724378216662483e-05,
+      "loss": 0.8728,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972340794639849e-05,
+      "loss": 0.8145,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722435995263644e-05,
+      "loss": 0.9633,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721462363425958e-05,
+      "loss": 0.8808,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720487051053742e-05,
+      "loss": 0.8345,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97195100583156e-05,
+      "loss": 0.8363,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718531385380424e-05,
+      "loss": 0.8958,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717551032417393e-05,
+      "loss": 0.9184,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716568999595978e-05,
+      "loss": 0.9156,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715585287085944e-05,
+      "loss": 0.8892,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714599895057347e-05,
+      "loss": 0.8637,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713612823680523e-05,
+      "loss": 0.7832,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712624073126113e-05,
+      "loss": 0.8931,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711633643565035e-05,
+      "loss": 0.8808,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710641535168504e-05,
+      "loss": 0.8381,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709647748108027e-05,
+      "loss": 0.752,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708652282555397e-05,
+      "loss": 0.841,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97076551386827e-05,
+      "loss": 0.8215,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706656316662307e-05,
+      "loss": 0.7986,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705655816666888e-05,
+      "loss": 0.8793,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704653638869392e-05,
+      "loss": 0.913,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970364978344307e-05,
+      "loss": 0.7803,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970264425056145e-05,
+      "loss": 0.786,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970163704039836e-05,
+      "loss": 0.28,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700628153127923e-05,
+      "loss": 0.87,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699617588924532e-05,
+      "loss": 0.264,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698605347962888e-05,
+      "loss": 0.889,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697591430417974e-05,
+      "loss": 0.9005,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696575836465065e-05,
+      "loss": 0.8613,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695558566279723e-05,
+      "loss": 0.8868,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694539620037803e-05,
+      "loss": 0.8453,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693518997915454e-05,
+      "loss": 0.8536,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96924967000891e-05,
+      "loss": 0.8425,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969147272673547e-05,
+      "loss": 0.8063,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690447078031574e-05,
+      "loss": 0.8412,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968941975415472e-05,
+      "loss": 0.8406,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688390755282493e-05,
+      "loss": 0.9021,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687360081592777e-05,
+      "loss": 0.8416,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686327733263743e-05,
+      "loss": 0.8419,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685293710473852e-05,
+      "loss": 0.8138,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684258013401856e-05,
+      "loss": 0.7438,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968322064222679e-05,
+      "loss": 0.8599,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682181597127986e-05,
+      "loss": 0.9224,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681140878285058e-05,
+      "loss": 0.8887,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680098485877922e-05,
+      "loss": 0.8719,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679054420086766e-05,
+      "loss": 0.8145,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678008681092077e-05,
+      "loss": 0.8556,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676961269074634e-05,
+      "loss": 0.2756,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675912184215506e-05,
+      "loss": 0.8337,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674861426696036e-05,
+      "loss": 0.8629,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673808996697872e-05,
+      "loss": 0.8391,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672754894402946e-05,
+      "loss": 0.2595,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967169911999348e-05,
+      "loss": 0.9051,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967064167365198e-05,
+      "loss": 0.8029,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966958255556125e-05,
+      "loss": 0.8215,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966852176590438e-05,
+      "loss": 0.8658,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667459304864738e-05,
+      "loss": 0.7893,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666395172625997e-05,
+      "loss": 0.8426,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966532936937211e-05,
+      "loss": 0.8293,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664261895287324e-05,
+      "loss": 0.837,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663192750556165e-05,
+      "loss": 0.824,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662121935363464e-05,
+      "loss": 0.2482,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966104944989432e-05,
+      "loss": 0.7434,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965997529433414e-05,
+      "loss": 0.8208,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965889946886861e-05,
+      "loss": 0.8213,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657821973683703e-05,
+      "loss": 0.8869,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656742808965687e-05,
+      "loss": 0.8266,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655661974901114e-05,
+      "loss": 0.879,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965457947167683e-05,
+      "loss": 0.7614,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965349529947996e-05,
+      "loss": 0.7769,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965240945849793e-05,
+      "loss": 0.8467,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965132194891844e-05,
+      "loss": 0.7949,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965023277092949e-05,
+      "loss": 0.8212,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649141924719367e-05,
+      "loss": 0.7611,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648049410476638e-05,
+      "loss": 0.793,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964695522839017e-05,
+      "loss": 0.9282,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645859378649114e-05,
+      "loss": 0.8609,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644761861442897e-05,
+      "loss": 0.8008,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643662676961256e-05,
+      "loss": 0.7703,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96425618253942e-05,
+      "loss": 0.7478,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641459306932034e-05,
+      "loss": 0.8563,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640355121765346e-05,
+      "loss": 0.82,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963924927008502e-05,
+      "loss": 0.843,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638141752082217e-05,
+      "loss": 0.8214,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637032567948398e-05,
+      "loss": 0.8341,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96359217178753e-05,
+      "loss": 0.8903,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634809202054958e-05,
+      "loss": 0.8883,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633695020679687e-05,
+      "loss": 0.8329,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632579173942097e-05,
+      "loss": 0.8726,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631461662035087e-05,
+      "loss": 0.8225,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630342485151828e-05,
+      "loss": 0.8449,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96292216434858e-05,
+      "loss": 0.7607,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628099137230755e-05,
+      "loss": 0.7754,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626974966580747e-05,
+      "loss": 0.7527,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625849131730105e-05,
+      "loss": 0.8668,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624721632873446e-05,
+      "loss": 0.8818,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623592470205686e-05,
+      "loss": 0.8873,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622461643922023e-05,
+      "loss": 0.8231,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962132915421793e-05,
+      "loss": 0.8632,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962019500128919e-05,
+      "loss": 0.8606,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619059185331857e-05,
+      "loss": 0.8703,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961792170654228e-05,
+      "loss": 0.8116,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961678256511709e-05,
+      "loss": 0.8528,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615641761253214e-05,
+      "loss": 0.8442,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961449929514786e-05,
+      "loss": 0.8682,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613355166998517e-05,
+      "loss": 0.8916,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612209377002977e-05,
+      "loss": 0.7959,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961106192535931e-05,
+      "loss": 0.9135,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960991281226587e-05,
+      "loss": 0.8638,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608762037921303e-05,
+      "loss": 0.8477,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607609602524547e-05,
+      "loss": 0.8155,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606455506274814e-05,
+      "loss": 0.913,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960529974937162e-05,
+      "loss": 0.8809,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604142332014753e-05,
+      "loss": 0.3049,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602983254404295e-05,
+      "loss": 0.8536,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960182251674061e-05,
+      "loss": 0.763,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960066011922436e-05,
+      "loss": 0.8778,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599496062056483e-05,
+      "loss": 0.8139,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959833034543821e-05,
+      "loss": 0.8883,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597162969571054e-05,
+      "loss": 0.8383,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595993934656823e-05,
+      "loss": 0.8119,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594823240897597e-05,
+      "loss": 0.881,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593650888495757e-05,
+      "loss": 0.9111,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959247687765397e-05,
+      "loss": 0.8414,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959130120857518e-05,
+      "loss": 0.9026,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590123881462624e-05,
+      "loss": 0.8097,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958894489651983e-05,
+      "loss": 0.8428,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587764253950595e-05,
+      "loss": 0.8581,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586581953959027e-05,
+      "loss": 0.8042,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585397996749506e-05,
+      "loss": 0.8538,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584212382526693e-05,
+      "loss": 0.8243,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583025111495554e-05,
+      "loss": 0.8677,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581836183861327e-05,
+      "loss": 0.8099,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580645599829536e-05,
+      "loss": 0.7055,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579453359606e-05,
+      "loss": 0.8354,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578259463396817e-05,
+      "loss": 0.892,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577063911408378e-05,
+      "loss": 0.7912,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957586670384735e-05,
+      "loss": 0.843,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95746678409207e-05,
+      "loss": 0.8323,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957346732283567e-05,
+      "loss": 0.8634,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957226514979979e-05,
+      "loss": 0.9385,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957106132202088e-05,
+      "loss": 0.8502,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956985583970704e-05,
+      "loss": 0.7824,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956864870306667e-05,
+      "loss": 0.8691,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567439912308432e-05,
+      "loss": 0.8201,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566229467641298e-05,
+      "loss": 0.8363,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565017369274516e-05,
+      "loss": 0.8763,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563803617417608e-05,
+      "loss": 0.7925,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562588212280407e-05,
+      "loss": 0.8191,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561371154073016e-05,
+      "loss": 0.8279,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956015244300582e-05,
+      "loss": 0.7885,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558932079289498e-05,
+      "loss": 0.8503,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557710063135015e-05,
+      "loss": 0.8633,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556486394753613e-05,
+      "loss": 0.836,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555261074356834e-05,
+      "loss": 0.8747,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554034102156492e-05,
+      "loss": 0.8014,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552805478364693e-05,
+      "loss": 0.8226,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551575203193832e-05,
+      "loss": 0.8691,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550343276856574e-05,
+      "loss": 0.8313,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954910969956589e-05,
+      "loss": 0.2773,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954787447153502e-05,
+      "loss": 0.8176,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95466375929775e-05,
+      "loss": 0.9314,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545399064107148e-05,
+      "loss": 0.8933,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544158885138066e-05,
+      "loss": 0.8791,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542917056284637e-05,
+      "loss": 0.8342,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541673577761542e-05,
+      "loss": 0.8482,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540428449783734e-05,
+      "loss": 0.8104,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953918167256646e-05,
+      "loss": 0.8608,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537933246325243e-05,
+      "loss": 0.8658,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536683171275902e-05,
+      "loss": 0.8614,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953543144763453e-05,
+      "loss": 0.9157,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534178075617518e-05,
+      "loss": 0.8578,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532923055441525e-05,
+      "loss": 0.8345,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531666387323516e-05,
+      "loss": 0.7977,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953040807148072e-05,
+      "loss": 0.8717,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529148108130665e-05,
+      "loss": 0.8141,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527886497491156e-05,
+      "loss": 0.8399,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526623239780286e-05,
+      "loss": 0.8619,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525358335216436e-05,
+      "loss": 0.8595,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524091784018264e-05,
+      "loss": 0.9009,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952282358640472e-05,
+      "loss": 0.8047,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521553742595033e-05,
+      "loss": 0.2897,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520282252808717e-05,
+      "loss": 0.9205,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951900911726558e-05,
+      "loss": 0.8743,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517734336185697e-05,
+      "loss": 0.8902,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516457909789446e-05,
+      "loss": 0.8157,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951517983829748e-05,
+      "loss": 0.9001,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513900121930735e-05,
+      "loss": 0.8198,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951261876091043e-05,
+      "loss": 0.8659,
+      "step": 988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951133575545808e-05,
+      "loss": 0.8324,
+      "step": 989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951005110579547e-05,
+      "loss": 0.872,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508764812144677e-05,
+      "loss": 0.8544,
+      "step": 991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950747687472806e-05,
+      "loss": 0.7669,
+      "step": 992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506187293768262e-05,
+      "loss": 0.8454,
+      "step": 993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504896069488217e-05,
+      "loss": 0.8617,
+      "step": 994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503603202111134e-05,
+      "loss": 0.879,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502308691860502e-05,
+      "loss": 0.8633,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950101253896011e-05,
+      "loss": 0.8524,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499714743634017e-05,
+      "loss": 0.8075,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498415306106574e-05,
+      "loss": 0.7779,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949711422660241e-05,
+      "loss": 0.9108,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495811505346444e-05,
+      "loss": 0.9164,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949450714256387e-05,
+      "loss": 0.8759,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949320113848018e-05,
+      "loss": 0.9299,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491893493321133e-05,
+      "loss": 0.8544,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490584207312783e-05,
+      "loss": 0.8629,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489273280681463e-05,
+      "loss": 0.8471,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487960713653794e-05,
+      "loss": 0.8536,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486646506456673e-05,
+      "loss": 0.7755,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485330659317287e-05,
+      "loss": 0.7873,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484013172463102e-05,
+      "loss": 0.8665,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948269404612188e-05,
+      "loss": 0.8156,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948137328052164e-05,
+      "loss": 0.822,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480050875890716e-05,
+      "loss": 0.8427,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94787268324577e-05,
+      "loss": 0.7916,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477401150451483e-05,
+      "loss": 0.8134,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947607383010123e-05,
+      "loss": 0.8918,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474744871636397e-05,
+      "loss": 0.2839,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473414275286715e-05,
+      "loss": 0.9256,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472082041282204e-05,
+      "loss": 0.8617,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947074816985317e-05,
+      "loss": 0.8909,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946941266123019e-05,
+      "loss": 0.748,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946807551564413e-05,
+      "loss": 0.8696,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946673673332615e-05,
+      "loss": 0.8371,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946539631450767e-05,
+      "loss": 0.3005,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464054259420418e-05,
+      "loss": 0.8356,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946271056829639e-05,
+      "loss": 0.8168,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461365241367867e-05,
+      "loss": 0.8729,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946001827886741e-05,
+      "loss": 0.8613,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458669681027872e-05,
+      "loss": 0.7793,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945731944808238e-05,
+      "loss": 0.8183,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455967580264347e-05,
+      "loss": 0.8341,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945461407780747e-05,
+      "loss": 0.8942,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453258940945724e-05,
+      "loss": 0.8988,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451902169913374e-05,
+      "loss": 0.8214,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450543764944955e-05,
+      "loss": 0.8603,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449183726275302e-05,
+      "loss": 0.849,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447822054139517e-05,
+      "loss": 0.8853,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944645874877299e-05,
+      "loss": 0.2885,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94450938104114e-05,
+      "loss": 0.8716,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944372723929069e-05,
+      "loss": 0.7495,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442359035647107e-05,
+      "loss": 0.8046,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944098919971717e-05,
+      "loss": 0.9072,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439617731737678e-05,
+      "loss": 0.825,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438244631945712e-05,
+      "loss": 0.8537,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943686990057864e-05,
+      "loss": 0.8204,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435493537874105e-05,
+      "loss": 0.8979,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943411554407005e-05,
+      "loss": 0.8026,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432735919404676e-05,
+      "loss": 0.8313,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431354664116476e-05,
+      "loss": 0.8547,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942997177844423e-05,
+      "loss": 0.8227,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428587262626994e-05,
+      "loss": 0.8124,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427201116904108e-05,
+      "loss": 0.8219,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942581334151519e-05,
+      "loss": 0.8743,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424423936700146e-05,
+      "loss": 0.8912,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423032902699157e-05,
+      "loss": 0.8155,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421640239752693e-05,
+      "loss": 0.8674,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94202459481015e-05,
+      "loss": 0.7234,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418850027986603e-05,
+      "loss": 0.2584,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417452479649317e-05,
+      "loss": 0.8546,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416053303331233e-05,
+      "loss": 0.8886,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414652499274224e-05,
+      "loss": 0.8188,
+      "step": 1061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941325006772045e-05,
+      "loss": 0.2368,
+      "step": 1062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411846008912337e-05,
+      "loss": 0.8387,
+      "step": 1063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941044032309261e-05,
+      "loss": 0.8691,
+      "step": 1064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940903301050427e-05,
+      "loss": 0.7939,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407624071390586e-05,
+      "loss": 0.8604,
+      "step": 1066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406213505995134e-05,
+      "loss": 0.837,
+      "step": 1067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940480131456174e-05,
+      "loss": 0.8005,
+      "step": 1068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403387497334544e-05,
+      "loss": 0.8787,
+      "step": 1069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401972054557936e-05,
+      "loss": 0.8415,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400554986476613e-05,
+      "loss": 0.8648,
+      "step": 1071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399136293335533e-05,
+      "loss": 0.2573,
+      "step": 1072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397715975379943e-05,
+      "loss": 0.7873,
+      "step": 1073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939629403285538e-05,
+      "loss": 0.8572,
+      "step": 1074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939487046600765e-05,
+      "loss": 0.8256,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393445275082836e-05,
+      "loss": 0.8182,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392018460327312e-05,
+      "loss": 0.7997,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390590021987735e-05,
+      "loss": 0.7835,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389159960311032e-05,
+      "loss": 0.853,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387728275544413e-05,
+      "loss": 0.7845,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938629496793538e-05,
+      "loss": 0.2689,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384860037731698e-05,
+      "loss": 0.8723,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383423485181422e-05,
+      "loss": 0.8593,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938198531053289e-05,
+      "loss": 0.8139,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380545514034716e-05,
+      "loss": 0.8539,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93791040959358e-05,
+      "loss": 0.8455,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937766105648531e-05,
+      "loss": 0.8936,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376216395932703e-05,
+      "loss": 0.8234,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374770114527722e-05,
+      "loss": 0.8948,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373322212520373e-05,
+      "loss": 0.8683,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371872690160965e-05,
+      "loss": 0.8284,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370421547700066e-05,
+      "loss": 0.8337,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368968785388535e-05,
+      "loss": 0.8792,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936751440347751e-05,
+      "loss": 0.8931,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366058402218405e-05,
+      "loss": 0.8045,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936460078186292e-05,
+      "loss": 0.8421,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363141542663035e-05,
+      "loss": 0.8973,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361680684870997e-05,
+      "loss": 0.2856,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360218208739344e-05,
+      "loss": 0.8784,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93587541145209e-05,
+      "loss": 0.7568,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357288402468754e-05,
+      "loss": 0.8331,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355821072836285e-05,
+      "loss": 0.8542,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354352125877144e-05,
+      "loss": 0.9047,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935288156184527e-05,
+      "loss": 0.7469,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935140938099487e-05,
+      "loss": 0.8325,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934993558358045e-05,
+      "loss": 0.8129,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348460169856774e-05,
+      "loss": 0.8376,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346983140078897e-05,
+      "loss": 0.8664,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934550449450215e-05,
+      "loss": 0.8933,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344024233382143e-05,
+      "loss": 0.8859,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342542356974766e-05,
+      "loss": 0.8186,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341058865536194e-05,
+      "loss": 0.8498,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339573759322875e-05,
+      "loss": 0.937,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338087038591534e-05,
+      "loss": 0.7979,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336598703599178e-05,
+      "loss": 0.8303,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335108754603095e-05,
+      "loss": 0.8372,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333617191860847e-05,
+      "loss": 0.8589,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332124015630282e-05,
+      "loss": 0.8575,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933062922616952e-05,
+      "loss": 0.8326,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329132823736968e-05,
+      "loss": 0.8236,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327634808591303e-05,
+      "loss": 0.8175,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326135180991486e-05,
+      "loss": 0.8017,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324633941196755e-05,
+      "loss": 0.7818,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323131089466625e-05,
+      "loss": 0.8235,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321626626060895e-05,
+      "loss": 0.8225,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320120551239636e-05,
+      "loss": 0.8791,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93186128652632e-05,
+      "loss": 0.8355,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317103568392227e-05,
+      "loss": 0.9224,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315592660887618e-05,
+      "loss": 0.8589,
+      "step": 1129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314080143010567e-05,
+      "loss": 0.7811,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312566015022535e-05,
+      "loss": 0.8368,
+      "step": 1131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311050277185273e-05,
+      "loss": 0.8342,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309532929760798e-05,
+      "loss": 0.873,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308013973011413e-05,
+      "loss": 0.8762,
+      "step": 1134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306493407199705e-05,
+      "loss": 0.8797,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930497123258852e-05,
+      "loss": 0.2786,
+      "step": 1136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303447449441005e-05,
+      "loss": 0.9428,
+      "step": 1137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301922058020574e-05,
+      "loss": 0.892,
+      "step": 1138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300395058590908e-05,
+      "loss": 0.8192,
+      "step": 1139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298866451415983e-05,
+      "loss": 0.8332,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929733623676005e-05,
+      "loss": 0.8584,
+      "step": 1141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295804414887634e-05,
+      "loss": 0.9076,
+      "step": 1142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294270986063533e-05,
+      "loss": 0.827,
+      "step": 1143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929273595055284e-05,
+      "loss": 0.8515,
+      "step": 1144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291199308620903e-05,
+      "loss": 0.8452,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289661060533365e-05,
+      "loss": 0.8746,
+      "step": 1146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288121206556137e-05,
+      "loss": 0.8123,
+      "step": 1147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928657974695541e-05,
+      "loss": 0.8358,
+      "step": 1148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285036681997665e-05,
+      "loss": 0.8342,
+      "step": 1149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283492011949633e-05,
+      "loss": 0.8395,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281945737078352e-05,
+      "loss": 0.2662,
+      "step": 1151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928039785765112e-05,
+      "loss": 0.8484,
+      "step": 1152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278848373935512e-05,
+      "loss": 0.8583,
+      "step": 1153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927729728619939e-05,
+      "loss": 0.9027,
+      "step": 1154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275744594710888e-05,
+      "loss": 0.8377,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274190299738416e-05,
+      "loss": 0.8788,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272634401550662e-05,
+      "loss": 0.8374,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271076900416596e-05,
+      "loss": 0.9079,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926951779660545e-05,
+      "loss": 0.8444,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926795709038676e-05,
+      "loss": 0.8866,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926639478203031e-05,
+      "loss": 0.7663,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926483087180618e-05,
+      "loss": 0.2639,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263265359984717e-05,
+      "loss": 0.7737,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261698246836548e-05,
+      "loss": 0.853,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260129532632584e-05,
+      "loss": 0.717,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258559217644005e-05,
+      "loss": 0.8241,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256987302142263e-05,
+      "loss": 0.8074,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255413786399098e-05,
+      "loss": 0.811,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253838670686518e-05,
+      "loss": 0.2559,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925226195527681e-05,
+      "loss": 0.7835,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250683640442545e-05,
+      "loss": 0.8051,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924910372645656e-05,
+      "loss": 0.8605,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247522213591967e-05,
+      "loss": 0.901,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924593910212217e-05,
+      "loss": 0.8139,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244354392320832e-05,
+      "loss": 0.8605,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92427680844619e-05,
+      "loss": 0.7814,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92411801788196e-05,
+      "loss": 0.8455,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239590675668425e-05,
+      "loss": 0.836,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237999575283156e-05,
+      "loss": 0.8412,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236406877938846e-05,
+      "loss": 0.8557,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234812583910812e-05,
+      "loss": 0.7773,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233216693474664e-05,
+      "loss": 0.8747,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231619206906282e-05,
+      "loss": 0.8891,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230020124481817e-05,
+      "loss": 0.9016,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228419446477706e-05,
+      "loss": 0.8122,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922681717317065e-05,
+      "loss": 0.8082,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225213304837636e-05,
+      "loss": 0.8876,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223607841755918e-05,
+      "loss": 0.8298,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222000784203035e-05,
+      "loss": 0.8001,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220392132456788e-05,
+      "loss": 0.8126,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218781886795274e-05,
+      "loss": 0.8081,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217170047496845e-05,
+      "loss": 0.8446,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215556614840138e-05,
+      "loss": 0.8143,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921394158910407e-05,
+      "loss": 0.8068,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212324970567825e-05,
+      "loss": 0.8462,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921070675951086e-05,
+      "loss": 0.7802,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920908695621292e-05,
+      "loss": 0.819,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207465560954017e-05,
+      "loss": 0.8345,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205842574014438e-05,
+      "loss": 0.8672,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204217995674746e-05,
+      "loss": 0.788,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202591826215778e-05,
+      "loss": 0.9039,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920096406591865e-05,
+      "loss": 0.833,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199334715064747e-05,
+      "loss": 0.7737,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919770377393574e-05,
+      "loss": 0.8278,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919607124281356e-05,
+      "loss": 0.9267,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919443712198042e-05,
+      "loss": 0.2881,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192801411718812e-05,
+      "loss": 0.8337,
+      "step": 1207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91911641123115e-05,
+      "loss": 0.7142,
+      "step": 1208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189525224041514e-05,
+      "loss": 0.8585,
+      "step": 1209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918788474719217e-05,
+      "loss": 0.8314,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918624268204706e-05,
+      "loss": 0.8986,
+      "step": 1211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918459902889004e-05,
+      "loss": 0.2497,
+      "step": 1212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182953788005244e-05,
+      "loss": 0.8017,
+      "step": 1213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918130695967709e-05,
+      "loss": 0.8555,
+      "step": 1214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179658544190253e-05,
+      "loss": 0.9,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91780085418297e-05,
+      "loss": 0.9321,
+      "step": 1216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176356952880656e-05,
+      "loss": 0.9122,
+      "step": 1217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917470377762864e-05,
+      "loss": 0.8255,
+      "step": 1218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173049016359427e-05,
+      "loss": 0.8188,
+      "step": 1219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917139266935907e-05,
+      "loss": 0.7975,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169734736913907e-05,
+      "loss": 0.8553,
+      "step": 1221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168075219310537e-05,
+      "loss": 0.8638,
+      "step": 1222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916641411683584e-05,
+      "loss": 0.8643,
+      "step": 1223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164751429776968e-05,
+      "loss": 0.8127,
+      "step": 1224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163087158421345e-05,
+      "loss": 0.8137,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161421303056676e-05,
+      "loss": 0.8644,
+      "step": 1226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915975386397093e-05,
+      "loss": 0.909,
+      "step": 1227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915808484145236e-05,
+      "loss": 0.8506,
+      "step": 1228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156414235789483e-05,
+      "loss": 0.9053,
+      "step": 1229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154742047271092e-05,
+      "loss": 0.8174,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153068276186264e-05,
+      "loss": 0.873,
+      "step": 1231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151392922824333e-05,
+      "loss": 0.8591,
+      "step": 1232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914971598747492e-05,
+      "loss": 0.6957,
+      "step": 1233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148037470427914e-05,
+      "loss": 0.8398,
+      "step": 1234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146357371973474e-05,
+      "loss": 0.8613,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144675692402036e-05,
+      "loss": 0.8178,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142992432004314e-05,
+      "loss": 0.8523,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141307591071288e-05,
+      "loss": 0.8745,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139621169894216e-05,
+      "loss": 0.8025,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137933168764622e-05,
+      "loss": 0.8009,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136243587974316e-05,
+      "loss": 0.8406,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134552427815362e-05,
+      "loss": 0.7973,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132859688580122e-05,
+      "loss": 0.8524,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131165370561205e-05,
+      "loss": 0.8674,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129469474051515e-05,
+      "loss": 0.7712,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127771999344214e-05,
+      "loss": 0.7884,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912607294673274e-05,
+      "loss": 0.8152,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124372316510808e-05,
+      "loss": 0.827,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122670108972405e-05,
+      "loss": 0.8065,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120966324411788e-05,
+      "loss": 0.7983,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119260963123487e-05,
+      "loss": 0.8299,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117554025402304e-05,
+      "loss": 0.7643,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911584551154332e-05,
+      "loss": 0.8201,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911413542184188e-05,
+      "loss": 0.7935,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91124237565936e-05,
+      "loss": 0.8454,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110710516094377e-05,
+      "loss": 0.8674,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108995700640374e-05,
+      "loss": 0.8465,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910727931052804e-05,
+      "loss": 0.9098,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910556134605407e-05,
+      "loss": 0.7664,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103841807515452e-05,
+      "loss": 0.8361,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910212069520944e-05,
+      "loss": 0.8098,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100398009433557e-05,
+      "loss": 0.829,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098673750485608e-05,
+      "loss": 0.7906,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096947918663656e-05,
+      "loss": 0.8525,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909522051426605e-05,
+      "loss": 0.7678,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093491537591396e-05,
+      "loss": 0.8078,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091760988938582e-05,
+      "loss": 0.9142,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909002886860677e-05,
+      "loss": 0.8154,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088295176895385e-05,
+      "loss": 0.8505,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086559914104125e-05,
+      "loss": 0.8141,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908482308053297e-05,
+      "loss": 0.8528,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908308467648216e-05,
+      "loss": 0.2278,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081344702252204e-05,
+      "loss": 0.2603,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079603158143898e-05,
+      "loss": 0.789,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077860044458296e-05,
+      "loss": 0.8828,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076115361496728e-05,
+      "loss": 0.8202,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074369109560795e-05,
+      "loss": 0.9139,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072621288952366e-05,
+      "loss": 0.8635,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907087189997359e-05,
+      "loss": 0.8579,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069120942926873e-05,
+      "loss": 0.2969,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906736841811491e-05,
+      "loss": 0.8351,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906561432584065e-05,
+      "loss": 0.9381,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063858666407326e-05,
+      "loss": 0.8405,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906210144011843e-05,
+      "loss": 0.8811,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060342647277732e-05,
+      "loss": 0.8684,
+      "step": 1285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058582288189277e-05,
+      "loss": 0.8551,
+      "step": 1286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056820363157375e-05,
+      "loss": 0.8467,
+      "step": 1287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90550568724866e-05,
+      "loss": 0.8543,
+      "step": 1288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905329181648181e-05,
+      "loss": 0.8295,
+      "step": 1289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051525195448127e-05,
+      "loss": 0.7809,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049757009690942e-05,
+      "loss": 0.8637,
+      "step": 1291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047987259515922e-05,
+      "loss": 0.818,
+      "step": 1292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046215945228997e-05,
+      "loss": 0.8605,
+      "step": 1293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044443067136375e-05,
+      "loss": 0.8151,
+      "step": 1294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904266862554453e-05,
+      "loss": 0.844,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040892620760205e-05,
+      "loss": 0.8113,
+      "step": 1296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903911505309042e-05,
+      "loss": 0.8618,
+      "step": 1297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037335922842454e-05,
+      "loss": 0.7536,
+      "step": 1298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035555230323862e-05,
+      "loss": 0.2413,
+      "step": 1299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033772975842478e-05,
+      "loss": 0.8575,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031989159706392e-05,
+      "loss": 0.7989,
+      "step": 1301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903020378222397e-05,
+      "loss": 0.8172,
+      "step": 1302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902841684370385e-05,
+      "loss": 0.8416,
+      "step": 1303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902662834445493e-05,
+      "loss": 0.8659,
+      "step": 1304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024838284786397e-05,
+      "loss": 0.8013,
+      "step": 1305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023046665007684e-05,
+      "loss": 0.8334,
+      "step": 1306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021253485428516e-05,
+      "loss": 0.793,
+      "step": 1307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019458746358868e-05,
+      "loss": 0.868,
+      "step": 1308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017662448109003e-05,
+      "loss": 0.7636,
+      "step": 1309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015864590989436e-05,
+      "loss": 0.7511,
+      "step": 1310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014065175310964e-05,
+      "loss": 0.8566,
+      "step": 1311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901226420138465e-05,
+      "loss": 0.8448,
+      "step": 1312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901046166952182e-05,
+      "loss": 0.857,
+      "step": 1313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008657580034083e-05,
+      "loss": 0.7337,
+      "step": 1314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006851933233307e-05,
+      "loss": 0.8682,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005044729431627e-05,
+      "loss": 0.8011,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003235968941453e-05,
+      "loss": 0.7728,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001425652075466e-05,
+      "loss": 0.8226,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899961377914661e-05,
+      "loss": 0.8436,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997800350468105e-05,
+      "loss": 0.8143,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995985366353427e-05,
+      "loss": 0.8187,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899416882711634e-05,
+      "loss": 0.808,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899235073307086e-05,
+      "loss": 0.8699,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990531084531278e-05,
+      "loss": 0.8274,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988709881812155e-05,
+      "loss": 0.7914,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898688712522832e-05,
+      "loss": 0.2586,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985062815094874e-05,
+      "loss": 0.7656,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898323695172717e-05,
+      "loss": 0.7846,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898140953544086e-05,
+      "loss": 0.8639,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979580566551833e-05,
+      "loss": 0.8249,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897775004537627e-05,
+      "loss": 0.8903,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89759179722306e-05,
+      "loss": 0.7949,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974084347431537e-05,
+      "loss": 0.2724,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972249171296057e-05,
+      "loss": 0.8669,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970412444141404e-05,
+      "loss": 0.8244,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896857416628509e-05,
+      "loss": 0.8814,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966734338044892e-05,
+      "loss": 0.8557,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964892959738863e-05,
+      "loss": 0.7995,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896305003168532e-05,
+      "loss": 0.8752,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961205554202837e-05,
+      "loss": 0.75,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959359527610278e-05,
+      "loss": 0.8341,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957511952226758e-05,
+      "loss": 0.793,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955662828371664e-05,
+      "loss": 0.8161,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895381215636465e-05,
+      "loss": 0.8326,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951959936525646e-05,
+      "loss": 0.8598,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950106169174834e-05,
+      "loss": 0.7581,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948250854632682e-05,
+      "loss": 0.8629,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946393993219903e-05,
+      "loss": 0.7976,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944535585257497e-05,
+      "loss": 0.7242,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894267563106672e-05,
+      "loss": 0.8246,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940814130969107e-05,
+      "loss": 0.9079,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938951085286445e-05,
+      "loss": 0.7939,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937086494340802e-05,
+      "loss": 0.8,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89352203584545e-05,
+      "loss": 0.8526,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933352677950144e-05,
+      "loss": 0.7658,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893148345315059e-05,
+      "loss": 0.8116,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929612684378967e-05,
+      "loss": 0.8378,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927740371958677e-05,
+      "loss": 0.2616,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925866516213386e-05,
+      "loss": 0.834,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923991117467016e-05,
+      "loss": 0.9119,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892211417604377e-05,
+      "loss": 0.7986,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920235692268113e-05,
+      "loss": 0.8477,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918355666464773e-05,
+      "loss": 0.7213,
+      "step": 1363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916474098958747e-05,
+      "loss": 0.7829,
+      "step": 1364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89145909900753e-05,
+      "loss": 0.806,
+      "step": 1365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891270634013996e-05,
+      "loss": 0.8805,
+      "step": 1366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891082014947853e-05,
+      "loss": 0.7828,
+      "step": 1367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908932418417066e-05,
+      "loss": 0.8625,
+      "step": 1368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89070431472819e-05,
+      "loss": 0.786,
+      "step": 1369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905152336399628e-05,
+      "loss": 0.8269,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890325998609711e-05,
+      "loss": 0.8124,
+      "step": 1371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901366096701474e-05,
+      "loss": 0.7525,
+      "step": 1372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899470668540112e-05,
+      "loss": 0.8497,
+      "step": 1373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889757370194069e-05,
+      "loss": 0.7916,
+      "step": 1374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895675197231128e-05,
+      "loss": 0.8353,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889377515473962e-05,
+      "loss": 0.8232,
+      "step": 1376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891873574794624e-05,
+      "loss": 0.7606,
+      "step": 1377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889970457724856e-05,
+      "loss": 0.7631,
+      "step": 1378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888065803859316e-05,
+      "loss": 0.8181,
+      "step": 1379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886159613527258e-05,
+      "loss": 0.7859,
+      "step": 1380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884251887058192e-05,
+      "loss": 0.8072,
+      "step": 1381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882342624781912e-05,
+      "loss": 0.8972,
+      "step": 1382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880431827028466e-05,
+      "loss": 0.8411,
+      "step": 1383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878519494128173e-05,
+      "loss": 0.8656,
+      "step": 1384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876605626411614e-05,
+      "loss": 0.8117,
+      "step": 1385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874690224209635e-05,
+      "loss": 0.868,
+      "step": 1386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872773287853352e-05,
+      "loss": 0.86,
+      "step": 1387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887085481767414e-05,
+      "loss": 0.7999,
+      "step": 1388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868934814003644e-05,
+      "loss": 0.8332,
+      "step": 1389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867013277173768e-05,
+      "loss": 0.7822,
+      "step": 1390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886509020751669e-05,
+      "loss": 0.8146,
+      "step": 1391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886316560536485e-05,
+      "loss": 0.8741,
+      "step": 1392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886123947105094e-05,
+      "loss": 0.2607,
+      "step": 1393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859311804907943e-05,
+      "loss": 0.9012,
+      "step": 1394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885738260726908e-05,
+      "loss": 0.7978,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855451878467853e-05,
+      "loss": 0.7833,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853519618838026e-05,
+      "loss": 0.7489,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885158582871362e-05,
+      "loss": 0.8291,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849650508428935e-05,
+      "loss": 0.8261,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847713658318518e-05,
+      "loss": 0.8681,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845775278717194e-05,
+      "loss": 0.7998,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843835369960047e-05,
+      "loss": 0.7949,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884189393238243e-05,
+      "loss": 0.8221,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883995096631995e-05,
+      "loss": 0.8456,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838006472108485e-05,
+      "loss": 0.7797,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883606045008418e-05,
+      "loss": 0.2559,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883411290058344e-05,
+      "loss": 0.8003,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883216382394294e-05,
+      "loss": 0.7946,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88302132204996e-05,
+      "loss": 0.8242,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882826109059063e-05,
+      "loss": 0.8435,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826307434553488e-05,
+      "loss": 0.7917,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88243522527259e-05,
+      "loss": 0.849,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822395545445858e-05,
+      "loss": 0.7845,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882043731305161e-05,
+      "loss": 0.8461,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818477555881672e-05,
+      "loss": 0.2797,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881651627427483e-05,
+      "loss": 0.8033,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814553468570128e-05,
+      "loss": 0.7528,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812589139106866e-05,
+      "loss": 0.8591,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810623286224622e-05,
+      "loss": 0.7784,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808655910263223e-05,
+      "loss": 0.804,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806687011562777e-05,
+      "loss": 0.8194,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804716590463632e-05,
+      "loss": 0.7686,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880274464730642e-05,
+      "loss": 0.8579,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800771182432024e-05,
+      "loss": 0.8607,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798796196181595e-05,
+      "loss": 0.9113,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796819688896546e-05,
+      "loss": 0.7973,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794841660918556e-05,
+      "loss": 0.8995,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792862112589558e-05,
+      "loss": 0.8156,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790881044251757e-05,
+      "loss": 0.8363,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788898456247616e-05,
+      "loss": 0.2804,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786914348919864e-05,
+      "loss": 0.8048,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784928722611485e-05,
+      "loss": 0.7842,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878294157766574e-05,
+      "loss": 0.8343,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878095291442614e-05,
+      "loss": 0.843,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778962733236456e-05,
+      "loss": 0.8291,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776971034440736e-05,
+      "loss": 0.8377,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877497781838328e-05,
+      "loss": 0.242,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772983085408648e-05,
+      "loss": 0.8925,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770986835861672e-05,
+      "loss": 0.872,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876898907008744e-05,
+      "loss": 0.7997,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766989788431296e-05,
+      "loss": 0.8642,
+      "step": 1441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764988991238864e-05,
+      "loss": 0.889,
+      "step": 1442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762986678856007e-05,
+      "loss": 0.2667,
+      "step": 1443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876098285162887e-05,
+      "loss": 0.8499,
+      "step": 1444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875897750990385e-05,
+      "loss": 0.7281,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756970654027605e-05,
+      "loss": 0.8099,
+      "step": 1446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875496228434706e-05,
+      "loss": 0.2586,
+      "step": 1447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752952401209397e-05,
+      "loss": 0.2654,
+      "step": 1448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875094100496206e-05,
+      "loss": 0.9257,
+      "step": 1449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874892809595276e-05,
+      "loss": 0.2605,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874691367452946e-05,
+      "loss": 0.8143,
+      "step": 1451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744897741040397e-05,
+      "loss": 0.8098,
+      "step": 1452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742880295834055e-05,
+      "loss": 0.8597,
+      "step": 1453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740861339259195e-05,
+      "loss": 0.728,
+      "step": 1454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873884087166482e-05,
+      "loss": 0.8509,
+      "step": 1455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736818893400218e-05,
+      "loss": 0.8168,
+      "step": 1456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873479540481492e-05,
+      "loss": 0.9032,
+      "step": 1457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873277040625872e-05,
+      "loss": 0.8694,
+      "step": 1458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730743898081676e-05,
+      "loss": 0.8307,
+      "step": 1459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872871588063411e-05,
+      "loss": 0.8592,
+      "step": 1460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726686354266607e-05,
+      "loss": 0.864,
+      "step": 1461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724655319330002e-05,
+      "loss": 0.8717,
+      "step": 1462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722622776175398e-05,
+      "loss": 0.784,
+      "step": 1463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720588725154158e-05,
+      "loss": 0.8665,
+      "step": 1464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871855316661791e-05,
+      "loss": 0.7588,
+      "step": 1465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716516100918527e-05,
+      "loss": 0.7592,
+      "step": 1466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714477528408164e-05,
+      "loss": 0.7502,
+      "step": 1467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871243744943922e-05,
+      "loss": 0.8691,
+      "step": 1468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710395864364362e-05,
+      "loss": 0.8109,
+      "step": 1469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870835277353652e-05,
+      "loss": 0.8116,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870630817730887e-05,
+      "loss": 0.8787,
+      "step": 1471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870426207603487e-05,
+      "loss": 0.7323,
+      "step": 1472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870221447006822e-05,
+      "loss": 0.8035,
+      "step": 1473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700165359762884e-05,
+      "loss": 0.8437,
+      "step": 1474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698114745473095e-05,
+      "loss": 0.7649,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696062627553336e-05,
+      "loss": 0.8756,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869400900635835e-05,
+      "loss": 0.8148,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691953882243156e-05,
+      "loss": 0.8247,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689897255563007e-05,
+      "loss": 0.9037,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687839126673433e-05,
+      "loss": 0.8059,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868577949593022e-05,
+      "loss": 0.8553,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683718363689416e-05,
+      "loss": 0.7832,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681655730307323e-05,
+      "loss": 0.7771,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679591596140503e-05,
+      "loss": 0.8677,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677525961545785e-05,
+      "loss": 0.8537,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675458826880247e-05,
+      "loss": 0.831,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673390192501237e-05,
+      "loss": 0.7803,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671320058766353e-05,
+      "loss": 0.8743,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669248426033454e-05,
+      "loss": 0.8545,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667175294660666e-05,
+      "loss": 0.8221,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665100665006365e-05,
+      "loss": 0.893,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663024537429188e-05,
+      "loss": 0.8737,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660946912288032e-05,
+      "loss": 0.8639,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658867789942054e-05,
+      "loss": 0.7584,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865678717075067e-05,
+      "loss": 0.8086,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865470505507355e-05,
+      "loss": 0.7311,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865262144327063e-05,
+      "loss": 0.8351,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650536335702097e-05,
+      "loss": 0.8255,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648449732728406e-05,
+      "loss": 0.7441,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646361634710263e-05,
+      "loss": 0.7655,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644272042008627e-05,
+      "loss": 0.7395,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864218095498473e-05,
+      "loss": 0.8328,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640088374000054e-05,
+      "loss": 0.8332,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863799429941634e-05,
+      "loss": 0.8226,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635898731595587e-05,
+      "loss": 0.8044,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633801670900053e-05,
+      "loss": 0.8342,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631703117692255e-05,
+      "loss": 0.8209,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629603072334964e-05,
+      "loss": 0.8769,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627501535191215e-05,
+      "loss": 0.8923,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625398506624295e-05,
+      "loss": 0.7847,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862329398699775e-05,
+      "loss": 0.8039,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862118797667539e-05,
+      "loss": 0.8158,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619080476021272e-05,
+      "loss": 0.8258,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616971485399724e-05,
+      "loss": 0.8503,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614861005175314e-05,
+      "loss": 0.8676,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861274903571289e-05,
+      "loss": 0.9072,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610635577377532e-05,
+      "loss": 0.8851,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86085206305346e-05,
+      "loss": 0.8518,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606404195549695e-05,
+      "loss": 0.827,
+      "step": 1519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604286272788688e-05,
+      "loss": 0.8333,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602166862617693e-05,
+      "loss": 0.8365,
+      "step": 1521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86000459654031e-05,
+      "loss": 0.8,
+      "step": 1522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597923581511537e-05,
+      "loss": 0.7925,
+      "step": 1523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85957997113099e-05,
+      "loss": 0.8127,
+      "step": 1524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859367435516534e-05,
+      "loss": 0.7701,
+      "step": 1525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591547513445262e-05,
+      "loss": 0.8616,
+      "step": 1526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589419186517333e-05,
+      "loss": 0.8436,
+      "step": 1527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587289374749467e-05,
+      "loss": 0.8551,
+      "step": 1528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585158078509852e-05,
+      "loss": 0.8855,
+      "step": 1529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583025298166914e-05,
+      "loss": 0.8402,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580891034089345e-05,
+      "loss": 0.8557,
+      "step": 1531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578755286646096e-05,
+      "loss": 0.8373,
+      "step": 1532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576618056206362e-05,
+      "loss": 0.7961,
+      "step": 1533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574479343139612e-05,
+      "loss": 0.8377,
+      "step": 1534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572339147815555e-05,
+      "loss": 0.7852,
+      "step": 1535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857019747060417e-05,
+      "loss": 0.856,
+      "step": 1536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568054311875678e-05,
+      "loss": 0.7823,
+      "step": 1537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856590967200057e-05,
+      "loss": 0.7435,
+      "step": 1538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563763551349584e-05,
+      "loss": 0.8688,
+      "step": 1539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561615950293714e-05,
+      "loss": 0.772,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855946686920422e-05,
+      "loss": 0.8358,
+      "step": 1541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85573163084526e-05,
+      "loss": 0.8659,
+      "step": 1542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855516426841063e-05,
+      "loss": 0.8174,
+      "step": 1543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855301074945032e-05,
+      "loss": 0.782,
+      "step": 1544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855085575194395e-05,
+      "loss": 0.8112,
+      "step": 1545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854869927626405e-05,
+      "loss": 0.794,
+      "step": 1546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854654132278341e-05,
+      "loss": 0.7828,
+      "step": 1547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544381891875068e-05,
+      "loss": 0.8581,
+      "step": 1548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854222098391232e-05,
+      "loss": 0.7871,
+      "step": 1549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540058599268726e-05,
+      "loss": 0.7886,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853789473831809e-05,
+      "loss": 0.7935,
+      "step": 1551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535729401434476e-05,
+      "loss": 0.8328,
+      "step": 1552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533562588992198e-05,
+      "loss": 0.8598,
+      "step": 1553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531394301365836e-05,
+      "loss": 0.8971,
+      "step": 1554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529224538930216e-05,
+      "loss": 0.8401,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852705330206042e-05,
+      "loss": 0.8427,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852488059113179e-05,
+      "loss": 0.8701,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852270640651992e-05,
+      "loss": 0.788,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852053074860065e-05,
+      "loss": 0.7435,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851835361775009e-05,
+      "loss": 0.8643,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516175014344593e-05,
+      "loss": 0.8291,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851399493876078e-05,
+      "loss": 0.7837,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511813391375507e-05,
+      "loss": 0.8893,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509630372565895e-05,
+      "loss": 0.7587,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850744588270933e-05,
+      "loss": 0.8549,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850525992218343e-05,
+      "loss": 0.8542,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503072491366082e-05,
+      "loss": 0.8511,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850088359063543e-05,
+      "loss": 0.8803,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498693220369855e-05,
+      "loss": 0.766,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849650138094801e-05,
+      "loss": 0.7892,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494308072748796e-05,
+      "loss": 0.7838,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492113296151363e-05,
+      "loss": 0.817,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848991705153512e-05,
+      "loss": 0.8594,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848771933927973e-05,
+      "loss": 0.8658,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485520159765104e-05,
+      "loss": 0.768,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483319513371418e-05,
+      "loss": 0.8578,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481117400479086e-05,
+      "loss": 0.789,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847891382146879e-05,
+      "loss": 0.8329,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476708776721458e-05,
+      "loss": 0.9183,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847450226661827e-05,
+      "loss": 0.8228,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472294291540666e-05,
+      "loss": 0.734,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470084851870333e-05,
+      "loss": 0.7561,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467873947989218e-05,
+      "loss": 0.8286,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846566158027951e-05,
+      "loss": 0.7632,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846344774912366e-05,
+      "loss": 0.8141,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846123245490437e-05,
+      "loss": 0.7721,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459015698004593e-05,
+      "loss": 0.7422,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845679747880754e-05,
+      "loss": 0.8167,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454577797696668e-05,
+      "loss": 0.7737,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452356655055697e-05,
+      "loss": 0.7629,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845013405126858e-05,
+      "loss": 0.8034,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447909986719544e-05,
+      "loss": 0.8021,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445684461793057e-05,
+      "loss": 0.7853,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443457476873847e-05,
+      "loss": 0.906,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441229032346882e-05,
+      "loss": 0.8786,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438999128597394e-05,
+      "loss": 0.849,
+      "step": 1596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843676776601086e-05,
+      "loss": 0.7874,
+      "step": 1597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434534944973016e-05,
+      "loss": 0.7402,
+      "step": 1598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432300665869845e-05,
+      "loss": 0.8405,
+      "step": 1599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430064929087583e-05,
+      "loss": 0.8646,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427827735012722e-05,
+      "loss": 0.7305,
+      "step": 1601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425589084031992e-05,
+      "loss": 0.9019,
+      "step": 1602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84233489765324e-05,
+      "loss": 0.8806,
+      "step": 1603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842110741290118e-05,
+      "loss": 0.8255,
+      "step": 1604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841886439352583e-05,
+      "loss": 0.7874,
+      "step": 1605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416619918794095e-05,
+      "loss": 0.8417,
+      "step": 1606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841437398909398e-05,
+      "loss": 0.8757,
+      "step": 1607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841212660481373e-05,
+      "loss": 0.8091,
+      "step": 1608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840987776634185e-05,
+      "loss": 0.8039,
+      "step": 1609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407627474067092e-05,
+      "loss": 0.9185,
+      "step": 1610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840537572837846e-05,
+      "loss": 0.8559,
+      "step": 1611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403122529665213e-05,
+      "loss": 0.801,
+      "step": 1612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400867878316854e-05,
+      "loss": 0.8197,
+      "step": 1613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839861177472314e-05,
+      "loss": 0.8265,
+      "step": 1614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396354219274085e-05,
+      "loss": 0.772,
+      "step": 1615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839409521235995e-05,
+      "loss": 0.8639,
+      "step": 1616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839183475437124e-05,
+      "loss": 0.8679,
+      "step": 1617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389572845698726e-05,
+      "loss": 0.7924,
+      "step": 1618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838730948673341e-05,
+      "loss": 0.7968,
+      "step": 1619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838504467786656e-05,
+      "loss": 0.7726,
+      "step": 1620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382778419489695e-05,
+      "loss": 0.7793,
+      "step": 1621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380510711994573e-05,
+      "loss": 0.8243,
+      "step": 1622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378241555773213e-05,
+      "loss": 0.7441,
+      "step": 1623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375970951217877e-05,
+      "loss": 0.8439,
+      "step": 1624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373698898721084e-05,
+      "loss": 0.8527,
+      "step": 1625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371425398675603e-05,
+      "loss": 0.829,
+      "step": 1626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836915045147444e-05,
+      "loss": 0.772,
+      "step": 1627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836687405751087e-05,
+      "loss": 0.7889,
+      "step": 1628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364596217178414e-05,
+      "loss": 0.9161,
+      "step": 1629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836231693087083e-05,
+      "loss": 0.8889,
+      "step": 1630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360036198982136e-05,
+      "loss": 0.799,
+      "step": 1631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357754021906603e-05,
+      "loss": 0.8476,
+      "step": 1632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355470400038745e-05,
+      "loss": 0.8057,
+      "step": 1633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353185333773327e-05,
+      "loss": 0.8364,
+      "step": 1634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835089882350537e-05,
+      "loss": 0.8818,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348610869630136e-05,
+      "loss": 0.8083,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346321472543135e-05,
+      "loss": 0.8389,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834403063264014e-05,
+      "loss": 0.8344,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341738350317162e-05,
+      "loss": 0.8761,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339444625970463e-05,
+      "loss": 0.8169,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337149459996558e-05,
+      "loss": 0.8859,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334852852792207e-05,
+      "loss": 0.7582,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332554804754425e-05,
+      "loss": 0.2739,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330255316280466e-05,
+      "loss": 0.8491,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327954387767843e-05,
+      "loss": 0.7542,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832565201961431e-05,
+      "loss": 0.7908,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832334821221788e-05,
+      "loss": 0.7964,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321042965976805e-05,
+      "loss": 0.8589,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318736281289592e-05,
+      "loss": 0.7737,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831642815855499e-05,
+      "loss": 0.7684,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314118598171998e-05,
+      "loss": 0.8265,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311807600539877e-05,
+      "loss": 0.8122,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830949516605812e-05,
+      "loss": 0.7852,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307181295126473e-05,
+      "loss": 0.7498,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304865988144932e-05,
+      "loss": 0.7911,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830254924551374e-05,
+      "loss": 0.8742,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300231067633388e-05,
+      "loss": 0.8499,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829791145490462e-05,
+      "loss": 0.9065,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295590407728424e-05,
+      "loss": 0.2797,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829326792650603e-05,
+      "loss": 0.8635,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290944011638926e-05,
+      "loss": 0.788,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288618663528842e-05,
+      "loss": 0.2951,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828629188257776e-05,
+      "loss": 0.8258,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283963669187902e-05,
+      "loss": 0.8209,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281634023761752e-05,
+      "loss": 0.7986,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279302946702018e-05,
+      "loss": 0.7833,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276970438411686e-05,
+      "loss": 0.8004,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274636499293963e-05,
+      "loss": 0.2544,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272301129752316e-05,
+      "loss": 0.8117,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269964330190456e-05,
+      "loss": 0.8129,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267626101012344e-05,
+      "loss": 0.7845,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265286442622184e-05,
+      "loss": 0.7881,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262945355424427e-05,
+      "loss": 0.7921,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260602839823784e-05,
+      "loss": 0.7976,
+      "step": 1674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258258896225188e-05,
+      "loss": 0.8234,
+      "step": 1675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825591352503384e-05,
+      "loss": 0.8092,
+      "step": 1676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253566726655184e-05,
+      "loss": 0.845,
+      "step": 1677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82512185014949e-05,
+      "loss": 0.9107,
+      "step": 1678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248868849958928e-05,
+      "loss": 0.8403,
+      "step": 1679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246517772453443e-05,
+      "loss": 0.821,
+      "step": 1680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244165269384878e-05,
+      "loss": 0.287,
+      "step": 1681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241811341159906e-05,
+      "loss": 0.8772,
+      "step": 1682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239455988185446e-05,
+      "loss": 0.833,
+      "step": 1683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823709921086866e-05,
+      "loss": 0.7569,
+      "step": 1684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823474100961697e-05,
+      "loss": 0.762,
+      "step": 1685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232381384838025e-05,
+      "loss": 0.8157,
+      "step": 1686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230020336939733e-05,
+      "loss": 0.835,
+      "step": 1687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822765786633025e-05,
+      "loss": 0.7993,
+      "step": 1688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225293973417965e-05,
+      "loss": 0.8597,
+      "step": 1689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222928658611523e-05,
+      "loss": 0.2856,
+      "step": 1690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220561922319814e-05,
+      "loss": 0.778,
+      "step": 1691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218193764951974e-05,
+      "loss": 0.7654,
+      "step": 1692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215824186917378e-05,
+      "loss": 0.8308,
+      "step": 1693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213453188625654e-05,
+      "loss": 0.7997,
+      "step": 1694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211080770486672e-05,
+      "loss": 0.8088,
+      "step": 1695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820870693291055e-05,
+      "loss": 0.7611,
+      "step": 1696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206331676307646e-05,
+      "loss": 0.8547,
+      "step": 1697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820395500108857e-05,
+      "loss": 0.891,
+      "step": 1698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201576907664175e-05,
+      "loss": 0.8054,
+      "step": 1699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819919739644556e-05,
+      "loss": 0.7316,
+      "step": 1700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819681646784406e-05,
+      "loss": 0.7923,
+      "step": 1701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194434122271272e-05,
+      "loss": 0.7551,
+      "step": 1702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819205036013902e-05,
+      "loss": 0.8333,
+      "step": 1703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189665181859387e-05,
+      "loss": 0.8816,
+      "step": 1704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187278587844695e-05,
+      "loss": 0.8052,
+      "step": 1705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184890578507506e-05,
+      "loss": 0.841,
+      "step": 1706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182501154260638e-05,
+      "loss": 0.8405,
+      "step": 1707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818011031551714e-05,
+      "loss": 0.8458,
+      "step": 1708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817771806269032e-05,
+      "loss": 0.8282,
+      "step": 1709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817532439619372e-05,
+      "loss": 0.8433,
+      "step": 1710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172929316441127e-05,
+      "loss": 0.8298,
+      "step": 1711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170532823846575e-05,
+      "loss": 0.7931,
+      "step": 1712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816813491882435e-05,
+      "loss": 0.8418,
+      "step": 1713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165735601788967e-05,
+      "loss": 0.8643,
+      "step": 1714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163334873155188e-05,
+      "loss": 0.8673,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160932733338037e-05,
+      "loss": 0.8166,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158529182752754e-05,
+      "loss": 0.7598,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815612422181485e-05,
+      "loss": 0.8222,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153717850940058e-05,
+      "loss": 0.8495,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151310070544364e-05,
+      "loss": 0.8806,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148900881044002e-05,
+      "loss": 0.7961,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814649028285544e-05,
+      "loss": 0.7963,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144078276395397e-05,
+      "loss": 0.8601,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141664862080834e-05,
+      "loss": 0.8818,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139250040328954e-05,
+      "loss": 0.7147,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81368338115572e-05,
+      "loss": 0.8035,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813441617618327e-05,
+      "loss": 0.8657,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131997134625086e-05,
+      "loss": 0.8233,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812957668730083e-05,
+      "loss": 0.7872,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127154834628925e-05,
+      "loss": 0.7957,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124731577028027e-05,
+      "loss": 0.8203,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812230691491704e-05,
+      "loss": 0.8404,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119880848715115e-05,
+      "loss": 0.8515,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117453378841644e-05,
+      "loss": 0.7712,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115024505716255e-05,
+      "loss": 0.7833,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112594229758832e-05,
+      "loss": 0.8908,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110162551389484e-05,
+      "loss": 0.8317,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107729471028573e-05,
+      "loss": 0.788,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105294989096705e-05,
+      "loss": 0.8067,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102859106014725e-05,
+      "loss": 0.8305,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810042182220372e-05,
+      "loss": 0.8235,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097983138085017e-05,
+      "loss": 0.7605,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095543054080195e-05,
+      "loss": 0.776,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093101570611057e-05,
+      "loss": 0.7981,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090658688099667e-05,
+      "loss": 0.8415,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808821440696832e-05,
+      "loss": 0.8641,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808576872763955e-05,
+      "loss": 0.7372,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808332165053615e-05,
+      "loss": 0.8481,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808087317608113e-05,
+      "loss": 0.7818,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807842330469776e-05,
+      "loss": 0.7245,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8075972036809542e-05,
+      "loss": 0.8478,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807351937284023e-05,
+      "loss": 0.8112,
+      "step": 1752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807106531321381e-05,
+      "loss": 0.8207,
+      "step": 1753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068609858354508e-05,
+      "loss": 0.7945,
+      "step": 1754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066153008686792e-05,
+      "loss": 0.8703,
+      "step": 1755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063694764635383e-05,
+      "loss": 0.8094,
+      "step": 1756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061235126625227e-05,
+      "loss": 0.8142,
+      "step": 1757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805877409508152e-05,
+      "loss": 0.789,
+      "step": 1758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056311670429697e-05,
+      "loss": 0.8376,
+      "step": 1759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053847853095438e-05,
+      "loss": 0.7278,
+      "step": 1760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805138264350465e-05,
+      "loss": 0.8271,
+      "step": 1761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048916042083497e-05,
+      "loss": 0.8241,
+      "step": 1762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046448049258373e-05,
+      "loss": 0.8792,
+      "step": 1763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043978665455923e-05,
+      "loss": 0.8463,
+      "step": 1764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041507891103014e-05,
+      "loss": 0.2832,
+      "step": 1765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039035726626774e-05,
+      "loss": 0.8682,
+      "step": 1766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036562172454562e-05,
+      "loss": 0.8953,
+      "step": 1767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034087229013973e-05,
+      "loss": 0.772,
+      "step": 1768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803161089673285e-05,
+      "loss": 0.7788,
+      "step": 1769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802913317603927e-05,
+      "loss": 0.7992,
+      "step": 1770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802665406736156e-05,
+      "loss": 0.7919,
+      "step": 1771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802417357112827e-05,
+      "loss": 0.7314,
+      "step": 1772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021691687768208e-05,
+      "loss": 0.7937,
+      "step": 1773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801920841771041e-05,
+      "loss": 0.8516,
+      "step": 1774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016723761384153e-05,
+      "loss": 0.754,
+      "step": 1775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014237719218954e-05,
+      "loss": 0.7538,
+      "step": 1776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801175029164458e-05,
+      "loss": 0.7896,
+      "step": 1777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009261479091024e-05,
+      "loss": 0.9065,
+      "step": 1778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800677128198852e-05,
+      "loss": 0.8037,
+      "step": 1779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800427970076755e-05,
+      "loss": 0.8258,
+      "step": 1780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001786735858825e-05,
+      "loss": 0.7919,
+      "step": 1781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799929238769331e-05,
+      "loss": 0.8164,
+      "step": 1782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996796656702183e-05,
+      "loss": 0.7842,
+      "step": 1783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994299543316892e-05,
+      "loss": 0.802,
+      "step": 1784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79918010479691e-05,
+      "loss": 0.7967,
+      "step": 1785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798930117109072e-05,
+      "loss": 0.7782,
+      "step": 1786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798679991311391e-05,
+      "loss": 0.7784,
+      "step": 1787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798429727447105e-05,
+      "loss": 0.8614,
+      "step": 1788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798179325559477e-05,
+      "loss": 0.7773,
+      "step": 1789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979287856917934e-05,
+      "loss": 0.785,
+      "step": 1790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976781078873646e-05,
+      "loss": 0.8031,
+      "step": 1791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974272921895252e-05,
+      "loss": 0.8093,
+      "step": 1792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971763386416334e-05,
+      "loss": 0.7514,
+      "step": 1793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969252472870708e-05,
+      "loss": 0.8437,
+      "step": 1794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966740181692438e-05,
+      "loss": 0.8559,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796422651331581e-05,
+      "loss": 0.31,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961711468175365e-05,
+      "loss": 0.7942,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959195046705877e-05,
+      "loss": 0.7328,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956677249342353e-05,
+      "loss": 0.8742,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795415807652004e-05,
+      "loss": 0.8003,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795163752867442e-05,
+      "loss": 0.8457,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949115606241225e-05,
+      "loss": 0.7784,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946592309656408e-05,
+      "loss": 0.696,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944067639356176e-05,
+      "loss": 0.8057,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941541595776954e-05,
+      "loss": 0.7856,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939014179355422e-05,
+      "loss": 0.8136,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936485390528493e-05,
+      "loss": 0.776,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793395522973331e-05,
+      "loss": 0.8033,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931423697407263e-05,
+      "loss": 0.8206,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928890793987966e-05,
+      "loss": 0.7764,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792635651991329e-05,
+      "loss": 0.8307,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792382087562132e-05,
+      "loss": 0.8765,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792128386155039e-05,
+      "loss": 0.858,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918745478139075e-05,
+      "loss": 0.7059,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916205725826185e-05,
+      "loss": 0.8746,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791366460505075e-05,
+      "loss": 0.7539,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911122116252066e-05,
+      "loss": 0.818,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908578259869633e-05,
+      "loss": 0.7914,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906033036343217e-05,
+      "loss": 0.8265,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79034864461128e-05,
+      "loss": 0.8269,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900938489618607e-05,
+      "loss": 0.8891,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78983891673011e-05,
+      "loss": 0.8775,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895838479600978e-05,
+      "loss": 0.8267,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893286426959177e-05,
+      "loss": 0.9169,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789073300981686e-05,
+      "loss": 0.287,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788817822861544e-05,
+      "loss": 0.8256,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885622083796546e-05,
+      "loss": 0.7697,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788306457580207e-05,
+      "loss": 0.8394,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7880505705074115e-05,
+      "loss": 0.8666,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877945472055033e-05,
+      "loss": 0.8113,
+      "step": 1830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7875383877187403e-05,
+      "loss": 0.7634,
+      "step": 1831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787282092091405e-05,
+      "loss": 0.8129,
+      "step": 1832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870256603678024e-05,
+      "loss": 0.8711,
+      "step": 1833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867690925922615e-05,
+      "loss": 0.8684,
+      "step": 1834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865123888091354e-05,
+      "loss": 0.7314,
+      "step": 1835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786255549062799e-05,
+      "loss": 0.826,
+      "step": 1836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785998573397653e-05,
+      "loss": 0.8834,
+      "step": 1837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857414618581193e-05,
+      "loss": 0.8223,
+      "step": 1838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785484214488645e-05,
+      "loss": 0.7881,
+      "step": 1839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852268313337004e-05,
+      "loss": 0.8103,
+      "step": 1840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784969312437778e-05,
+      "loss": 0.7758,
+      "step": 1841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847116578453952e-05,
+      "loss": 0.8415,
+      "step": 1842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844538676010926e-05,
+      "loss": 0.8686,
+      "step": 1843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784195941749433e-05,
+      "loss": 0.8377,
+      "step": 1844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839378803350053e-05,
+      "loss": 0.7584,
+      "step": 1845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783679683402419e-05,
+      "loss": 0.7755,
+      "step": 1846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783421350996309e-05,
+      "loss": 0.7764,
+      "step": 1847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831628831613316e-05,
+      "loss": 0.7738,
+      "step": 1848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829042799421686e-05,
+      "loss": 0.843,
+      "step": 1849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826455413835243e-05,
+      "loss": 0.8075,
+      "step": 1850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823866675301267e-05,
+      "loss": 0.7739,
+      "step": 1851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821276584267263e-05,
+      "loss": 0.8189,
+      "step": 1852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818685141180977e-05,
+      "loss": 0.8679,
+      "step": 1853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816092346490396e-05,
+      "loss": 0.7195,
+      "step": 1854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781349820064372e-05,
+      "loss": 0.8498,
+      "step": 1855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781090270408941e-05,
+      "loss": 0.8159,
+      "step": 1856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780830585727613e-05,
+      "loss": 0.7194,
+      "step": 1857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805707660652804e-05,
+      "loss": 0.8184,
+      "step": 1858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803108114668574e-05,
+      "loss": 0.8563,
+      "step": 1859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800507219772818e-05,
+      "loss": 0.8539,
+      "step": 1860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797904976415155e-05,
+      "loss": 0.7934,
+      "step": 1861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795301385045425e-05,
+      "loss": 0.8209,
+      "step": 1862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779269644611371e-05,
+      "loss": 0.8251,
+      "step": 1863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790090160070318e-05,
+      "loss": 0.7906,
+      "step": 1864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787482527365796e-05,
+      "loss": 0.307,
+      "step": 1865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784873548450922e-05,
+      "loss": 0.7525,
+      "step": 1866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778226322377671e-05,
+      "loss": 0.7671,
+      "step": 1867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779651553794394e-05,
+      "loss": 0.839,
+      "step": 1868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777038538955457e-05,
+      "loss": 0.7932,
+      "step": 1869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774424179711605e-05,
+      "loss": 0.8112,
+      "step": 1870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7771808476514776e-05,
+      "loss": 0.806,
+      "step": 1871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769191429817146e-05,
+      "loss": 0.7688,
+      "step": 1872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776657304007112e-05,
+      "loss": 0.8652,
+      "step": 1873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763953307729326e-05,
+      "loss": 0.8016,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761332233244647e-05,
+      "loss": 0.8323,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758709817070178e-05,
+      "loss": 0.7423,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756086059659247e-05,
+      "loss": 0.7107,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753460961465427e-05,
+      "loss": 0.7955,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750834522942515e-05,
+      "loss": 0.8171,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748206744544533e-05,
+      "loss": 0.8026,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745577626725745e-05,
+      "loss": 0.3033,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742947169940638e-05,
+      "loss": 0.7701,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774031537464394e-05,
+      "loss": 0.2985,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773768224129061e-05,
+      "loss": 0.863,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773504777033582e-05,
+      "loss": 0.2515,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732411962235005e-05,
+      "loss": 0.8274,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729774817443796e-05,
+      "loss": 0.88,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772713633641808e-05,
+      "loss": 0.8462,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772449651961397e-05,
+      "loss": 0.7522,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721855367487803e-05,
+      "loss": 0.8244,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719212880496147e-05,
+      "loss": 0.9163,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716569059095816e-05,
+      "loss": 0.819,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713923903743835e-05,
+      "loss": 0.7978,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771127741489747e-05,
+      "loss": 0.8818,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708629593014215e-05,
+      "loss": 0.7868,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705980438551798e-05,
+      "loss": 0.8147,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770332995196817e-05,
+      "loss": 0.7972,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770067813372152e-05,
+      "loss": 0.8479,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769802498427026e-05,
+      "loss": 0.7884,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769537050407304e-05,
+      "loss": 0.8037,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692714693588733e-05,
+      "loss": 0.8399,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690057553276447e-05,
+      "loss": 0.7927,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687399083595518e-05,
+      "loss": 0.3284,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768473928500551e-05,
+      "loss": 0.2765,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768207815796622e-05,
+      "loss": 0.7551,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767941570293767e-05,
+      "loss": 0.8147,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767675192038012e-05,
+      "loss": 0.8409,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767408681075405e-05,
+      "loss": 0.8581,
+      "step": 1908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671420374520174e-05,
+      "loss": 0.6907,
+      "step": 1909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668752612139433e-05,
+      "loss": 0.8053,
+      "step": 1910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666083524073003e-05,
+      "loss": 0.8466,
+      "step": 1911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663413110782284e-05,
+      "loss": 0.8311,
+      "step": 1912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660741372728904e-05,
+      "loss": 0.8722,
+      "step": 1913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658068310374724e-05,
+      "loss": 0.777,
+      "step": 1914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655393924181834e-05,
+      "loss": 0.8302,
+      "step": 1915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765271821461255e-05,
+      "loss": 0.8334,
+      "step": 1916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650041182129415e-05,
+      "loss": 0.7631,
+      "step": 1917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647362827195204e-05,
+      "loss": 0.8168,
+      "step": 1918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644683150272922e-05,
+      "loss": 0.7856,
+      "step": 1919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642002151825804e-05,
+      "loss": 0.7661,
+      "step": 1920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639319832317306e-05,
+      "loss": 0.7732,
+      "step": 1921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763663619221112e-05,
+      "loss": 0.7885,
+      "step": 1922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633951231971157e-05,
+      "loss": 0.8363,
+      "step": 1923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631264952061563e-05,
+      "loss": 0.7541,
+      "step": 1924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762857735294672e-05,
+      "loss": 0.8366,
+      "step": 1925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625888435091217e-05,
+      "loss": 0.8335,
+      "step": 1926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762319819895989e-05,
+      "loss": 0.8191,
+      "step": 1927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620506645017793e-05,
+      "loss": 0.8183,
+      "step": 1928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617813773730216e-05,
+      "loss": 0.8531,
+      "step": 1929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615119585562667e-05,
+      "loss": 0.7956,
+      "step": 1930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761242408098089e-05,
+      "loss": 0.778,
+      "step": 1931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609727260450846e-05,
+      "loss": 0.7909,
+      "step": 1932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607029124438737e-05,
+      "loss": 0.738,
+      "step": 1933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604329673410983e-05,
+      "loss": 0.7682,
+      "step": 1934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601628907834234e-05,
+      "loss": 0.7428,
+      "step": 1935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598926828175368e-05,
+      "loss": 0.3316,
+      "step": 1936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596223434901486e-05,
+      "loss": 0.8148,
+      "step": 1937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759351872847993e-05,
+      "loss": 0.7861,
+      "step": 1938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590812709378246e-05,
+      "loss": 0.7613,
+      "step": 1939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758810537806422e-05,
+      "loss": 0.796,
+      "step": 1940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585396735005876e-05,
+      "loss": 0.7991,
+      "step": 1941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582686780671445e-05,
+      "loss": 0.8082,
+      "step": 1942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757997551552939e-05,
+      "loss": 0.7463,
+      "step": 1943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577262940048408e-05,
+      "loss": 0.7943,
+      "step": 1944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757454905469742e-05,
+      "loss": 0.7575,
+      "step": 1945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571833859945565e-05,
+      "loss": 0.942,
+      "step": 1946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756911735626222e-05,
+      "loss": 0.8523,
+      "step": 1947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566399544116976e-05,
+      "loss": 0.8192,
+      "step": 1948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563680423979666e-05,
+      "loss": 0.8069,
+      "step": 1949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560959996320334e-05,
+      "loss": 0.8354,
+      "step": 1950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755823826160926e-05,
+      "loss": 0.8672,
+      "step": 1951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755551522031694e-05,
+      "loss": 0.8105,
+      "step": 1952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755279087291411e-05,
+      "loss": 0.7806,
+      "step": 1953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550065219871716e-05,
+      "loss": 0.7749,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547338261660944e-05,
+      "loss": 0.8031,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544609998753197e-05,
+      "loss": 0.8162,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541880431620106e-05,
+      "loss": 0.8247,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539149560733524e-05,
+      "loss": 0.8114,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536417386565536e-05,
+      "loss": 0.7847,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753368390958845e-05,
+      "loss": 0.7287,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530949130274795e-05,
+      "loss": 0.8128,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528213049097328e-05,
+      "loss": 0.7305,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525475666529032e-05,
+      "loss": 0.8484,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522736983043114e-05,
+      "loss": 0.8444,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519996999113008e-05,
+      "loss": 0.7207,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751725571521237e-05,
+      "loss": 0.8217,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751451313181508e-05,
+      "loss": 0.8474,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511769249395244e-05,
+      "loss": 0.8765,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509024068427198e-05,
+      "loss": 0.7781,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750627758938549e-05,
+      "loss": 0.7676,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503529812744903e-05,
+      "loss": 0.7939,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500780738980443e-05,
+      "loss": 0.7792,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498030368567335e-05,
+      "loss": 0.8367,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495278701981037e-05,
+      "loss": 0.8683,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492525739697214e-05,
+      "loss": 0.8496,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748977148219178e-05,
+      "loss": 0.7745,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748701592994085e-05,
+      "loss": 0.7524,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484259083420774e-05,
+      "loss": 0.7843,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748150094310813e-05,
+      "loss": 0.7981,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74787415094797e-05,
+      "loss": 0.8268,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747598078301252e-05,
+      "loss": 0.2947,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473218764183818e-05,
+      "loss": 0.8466,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747045545347107e-05,
+      "loss": 0.8338,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467690851351963e-05,
+      "loss": 0.7523,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464924958304408e-05,
+      "loss": 0.881,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746215777480654e-05,
+      "loss": 0.82,
+      "step": 1986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459389301336715e-05,
+      "loss": 0.7894,
+      "step": 1987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456619538373525e-05,
+      "loss": 0.8012,
+      "step": 1988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453848486395765e-05,
+      "loss": 0.8418,
+      "step": 1989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451076145882466e-05,
+      "loss": 0.7599,
+      "step": 1990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744830251731288e-05,
+      "loss": 0.8739,
+      "step": 1991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445527601166477e-05,
+      "loss": 0.8453,
+      "step": 1992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744275139792295e-05,
+      "loss": 0.7955,
+      "step": 1993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743997390806223e-05,
+      "loss": 0.7795,
+      "step": 1994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437195132064443e-05,
+      "loss": 0.7786,
+      "step": 1995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434415070409957e-05,
+      "loss": 0.8306,
+      "step": 1996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431633723579357e-05,
+      "loss": 0.8621,
+      "step": 1997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428851092053455e-05,
+      "loss": 0.8564,
+      "step": 1998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742606717631327e-05,
+      "loss": 0.7219,
+      "step": 1999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423281976840063e-05,
+      "loss": 0.8136,
+      "step": 2000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420495494115304e-05,
+      "loss": 0.8414,
+      "step": 2001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417707728620686e-05,
+      "loss": 0.8149,
+      "step": 2002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414918680838128e-05,
+      "loss": 0.8326,
+      "step": 2003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412128351249763e-05,
+      "loss": 0.9178,
+      "step": 2004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740933674033796e-05,
+      "loss": 0.8436,
+      "step": 2005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74065438485853e-05,
+      "loss": 0.8225,
+      "step": 2006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403749676474577e-05,
+      "loss": 0.8471,
+      "step": 2007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7400954224488823e-05,
+      "loss": 0.8133,
+      "step": 2008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739815749311128e-05,
+      "loss": 0.8358,
+      "step": 2009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395359482825416e-05,
+      "loss": 0.8515,
+      "step": 2010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392560194114917e-05,
+      "loss": 0.8339,
+      "step": 2011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73897596274637e-05,
+      "loss": 0.7215,
+      "step": 2012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386957783355884e-05,
+      "loss": 0.7755,
+      "step": 2013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384154662275823e-05,
+      "loss": 0.7839,
+      "step": 2014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381350264708092e-05,
+      "loss": 0.7921,
+      "step": 2015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378544591137477e-05,
+      "loss": 0.7779,
+      "step": 2016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375737642048997e-05,
+      "loss": 0.8133,
+      "step": 2017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737292941792788e-05,
+      "loss": 0.8472,
+      "step": 2018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370119919259582e-05,
+      "loss": 0.8071,
+      "step": 2019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367309146529775e-05,
+      "loss": 0.8224,
+      "step": 2020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364497100224355e-05,
+      "loss": 0.8175,
+      "step": 2021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361683780829435e-05,
+      "loss": 0.8094,
+      "step": 2022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358869188831348e-05,
+      "loss": 0.7677,
+      "step": 2023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356053324716653e-05,
+      "loss": 0.7098,
+      "step": 2024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735323618897212e-05,
+      "loss": 0.7967,
+      "step": 2025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350417782084747e-05,
+      "loss": 0.7922,
+      "step": 2026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734759810454174e-05,
+      "loss": 0.8072,
+      "step": 2027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344777156830536e-05,
+      "loss": 0.8086,
+      "step": 2028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341954939438796e-05,
+      "loss": 0.7694,
+      "step": 2029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339131452854377e-05,
+      "loss": 0.8264,
+      "step": 2030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336306697565386e-05,
+      "loss": 0.8186,
+      "step": 2031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333480674060127e-05,
+      "loss": 0.8279,
+      "step": 2032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733065338282713e-05,
+      "loss": 0.3118,
+      "step": 2033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327824824355144e-05,
+      "loss": 0.7894,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324994999133144e-05,
+      "loss": 0.8571,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322163907650308e-05,
+      "loss": 0.7938,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731933155039605e-05,
+      "loss": 0.326,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316497927859993e-05,
+      "loss": 0.8377,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313663040531975e-05,
+      "loss": 0.8352,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310826888902072e-05,
+      "loss": 0.8472,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307989473460556e-05,
+      "loss": 0.8091,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305150794697927e-05,
+      "loss": 0.8862,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302310853104906e-05,
+      "loss": 0.7954,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729946964917243e-05,
+      "loss": 0.837,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296627183391646e-05,
+      "loss": 0.7102,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293783456253935e-05,
+      "loss": 0.7948,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290938468250884e-05,
+      "loss": 0.7358,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288092219874303e-05,
+      "loss": 0.8826,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285244711616223e-05,
+      "loss": 0.7856,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728239594396888e-05,
+      "loss": 0.8035,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279545917424742e-05,
+      "loss": 0.8286,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276694632476487e-05,
+      "loss": 0.8123,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273842089617013e-05,
+      "loss": 0.8287,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270988289339433e-05,
+      "loss": 0.8372,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268133232137082e-05,
+      "loss": 0.843,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265276918503504e-05,
+      "loss": 0.727,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262419348932473e-05,
+      "loss": 0.8117,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725956052391797e-05,
+      "loss": 0.8175,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725670044395419e-05,
+      "loss": 0.2859,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725383910953556e-05,
+      "loss": 0.7511,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250976521156708e-05,
+      "loss": 0.8959,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248112679312487e-05,
+      "loss": 0.8558,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724524758449797e-05,
+      "loss": 0.8604,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242381237208434e-05,
+      "loss": 0.2996,
+      "step": 2064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239513637939383e-05,
+      "loss": 0.7583,
+      "step": 2065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236644787186533e-05,
+      "loss": 0.8332,
+      "step": 2066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233774685445824e-05,
+      "loss": 0.7574,
+      "step": 2067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72309033332134e-05,
+      "loss": 0.6847,
+      "step": 2068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722803073098563e-05,
+      "loss": 0.8199,
+      "step": 2069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225156879259094e-05,
+      "loss": 0.2354,
+      "step": 2070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222281778530588e-05,
+      "loss": 0.813,
+      "step": 2071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219405429297138e-05,
+      "loss": 0.7865,
+      "step": 2072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216527832055962e-05,
+      "loss": 0.8045,
+      "step": 2073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213648987304512e-05,
+      "loss": 0.8803,
+      "step": 2074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210768895540448e-05,
+      "loss": 0.7937,
+      "step": 2075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207887557261647e-05,
+      "loss": 0.2919,
+      "step": 2076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72050049729662e-05,
+      "loss": 0.8442,
+      "step": 2077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720212114315242e-05,
+      "loss": 0.7586,
+      "step": 2078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199236068318826e-05,
+      "loss": 0.7907,
+      "step": 2079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196349748964154e-05,
+      "loss": 0.8195,
+      "step": 2080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193462185587365e-05,
+      "loss": 0.7851,
+      "step": 2081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719057337868762e-05,
+      "loss": 0.7541,
+      "step": 2082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187683328764314e-05,
+      "loss": 0.8565,
+      "step": 2083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184792036317032e-05,
+      "loss": 0.791,
+      "step": 2084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7181899501845594e-05,
+      "loss": 0.8677,
+      "step": 2085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717900572585003e-05,
+      "loss": 0.8428,
+      "step": 2086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717611070883058e-05,
+      "loss": 0.758,
+      "step": 2087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173214451287697e-05,
+      "loss": 0.874,
+      "step": 2088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170316953722064e-05,
+      "loss": 0.7534,
+      "step": 2089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167418216634555e-05,
+      "loss": 0.7165,
+      "step": 2090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716451824052628e-05,
+      "loss": 0.8664,
+      "step": 2091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161617025898546e-05,
+      "loss": 0.8689,
+      "step": 2092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158714573252885e-05,
+      "loss": 0.8531,
+      "step": 2093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155810883091038e-05,
+      "loss": 0.8744,
+      "step": 2094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152905955914964e-05,
+      "loss": 0.7624,
+      "step": 2095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149999792226833e-05,
+      "loss": 0.8454,
+      "step": 2096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147092392529025e-05,
+      "loss": 0.7983,
+      "step": 2097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144183757324143e-05,
+      "loss": 0.8312,
+      "step": 2098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141273887114998e-05,
+      "loss": 0.7832,
+      "step": 2099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138362782404612e-05,
+      "loss": 0.7271,
+      "step": 2100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135450443696226e-05,
+      "loss": 0.7656,
+      "step": 2101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132536871493284e-05,
+      "loss": 0.7108,
+      "step": 2102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129622066299464e-05,
+      "loss": 0.7761,
+      "step": 2103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126706028618637e-05,
+      "loss": 0.8404,
+      "step": 2104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123788758954888e-05,
+      "loss": 0.7897,
+      "step": 2105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120870257812535e-05,
+      "loss": 0.8101,
+      "step": 2106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117950525696082e-05,
+      "loss": 0.8625,
+      "step": 2107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115029563110265e-05,
+      "loss": 0.7866,
+      "step": 2108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112107370560022e-05,
+      "loss": 0.791,
+      "step": 2109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109183948550512e-05,
+      "loss": 0.7875,
+      "step": 2110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106259297587103e-05,
+      "loss": 0.7675,
+      "step": 2111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103333418175375e-05,
+      "loss": 0.7828,
+      "step": 2112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100406310821118e-05,
+      "loss": 0.839,
+      "step": 2113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097477976030336e-05,
+      "loss": 0.7808,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094548414309248e-05,
+      "loss": 0.842,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091617626164282e-05,
+      "loss": 0.7542,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708868561210208e-05,
+      "loss": 0.8213,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085752372629496e-05,
+      "loss": 0.7711,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082817908253587e-05,
+      "loss": 0.8551,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079882219481642e-05,
+      "loss": 0.8407,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076945306821136e-05,
+      "loss": 0.7736,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074007170779778e-05,
+      "loss": 0.7962,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.7933,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068127230586353e-05,
+      "loss": 0.8125,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065185427450745e-05,
+      "loss": 0.8066,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706224240296719e-05,
+      "loss": 0.8342,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705929815764445e-05,
+      "loss": 0.8358,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056352691991488e-05,
+      "loss": 0.8135,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053406006517486e-05,
+      "loss": 0.8784,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050458101731836e-05,
+      "loss": 0.7779,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047508978144135e-05,
+      "loss": 0.7341,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704455863626419e-05,
+      "loss": 0.8436,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041607076602026e-05,
+      "loss": 0.8186,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703865429966787e-05,
+      "loss": 0.7355,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703570030597217e-05,
+      "loss": 0.8058,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032745096025574e-05,
+      "loss": 0.8573,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702978867033895e-05,
+      "loss": 0.7871,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026831029423366e-05,
+      "loss": 0.7211,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023872173790105e-05,
+      "loss": 0.756,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020912103950663e-05,
+      "loss": 0.3059,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017950820416738e-05,
+      "loss": 0.8133,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701498832370025e-05,
+      "loss": 0.755,
+      "step": 2142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012024614313313e-05,
+      "loss": 0.8273,
+      "step": 2143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009059692768265e-05,
+      "loss": 0.8622,
+      "step": 2144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006093559577647e-05,
+      "loss": 0.7912,
+      "step": 2145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003126215254207e-05,
+      "loss": 0.7781,
+      "step": 2146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000157660310906e-05,
+      "loss": 0.8062,
+      "step": 2147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997187895260918e-05,
+      "loss": 0.8788,
+      "step": 2148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994216920617618e-05,
+      "loss": 0.7773,
+      "step": 2149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991244736894593e-05,
+      "loss": 0.8503,
+      "step": 2150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698827134460564e-05,
+      "loss": 0.877,
+      "step": 2151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985296744264765e-05,
+      "loss": 0.7777,
+      "step": 2152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698232093638618e-05,
+      "loss": 0.7453,
+      "step": 2153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697934392148432e-05,
+      "loss": 0.7891,
+      "step": 2154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976365700073808e-05,
+      "loss": 0.8332,
+      "step": 2155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697338627266948e-05,
+      "loss": 0.781,
+      "step": 2156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970405639786397e-05,
+      "loss": 0.8644,
+      "step": 2157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696742380193981e-05,
+      "loss": 0.8393,
+      "step": 2158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696444075964518e-05,
+      "loss": 0.8236,
+      "step": 2159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696145651341819e-05,
+      "loss": 0.8244,
+      "step": 2160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958471063774718e-05,
+      "loss": 0.7876,
+      "step": 2161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955484411230854e-05,
+      "loss": 0.8577,
+      "step": 2162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952496556302894e-05,
+      "loss": 0.8039,
+      "step": 2163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694950749950735e-05,
+      "loss": 0.8816,
+      "step": 2164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694651724136093e-05,
+      "loss": 0.7187,
+      "step": 2165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943525782380562e-05,
+      "loss": 0.8242,
+      "step": 2166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940533123083366e-05,
+      "loss": 0.7428,
+      "step": 2167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937539263986687e-05,
+      "loss": 0.7589,
+      "step": 2168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934544205608065e-05,
+      "loss": 0.7753,
+      "step": 2169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931547948465245e-05,
+      "loss": 0.7863,
+      "step": 2170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928550493076197e-05,
+      "loss": 0.8036,
+      "step": 2171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925551839959082e-05,
+      "loss": 0.8426,
+      "step": 2172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692255198963227e-05,
+      "loss": 0.81,
+      "step": 2173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919550942614344e-05,
+      "loss": 0.7811,
+      "step": 2174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691654869942409e-05,
+      "loss": 0.8015,
+      "step": 2175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913545260580497e-05,
+      "loss": 0.8178,
+      "step": 2176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910540626602773e-05,
+      "loss": 0.7589,
+      "step": 2177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907534798010316e-05,
+      "loss": 0.7954,
+      "step": 2178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904527775322743e-05,
+      "loss": 0.8003,
+      "step": 2179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901519559059874e-05,
+      "loss": 0.7992,
+      "step": 2180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898510149741735e-05,
+      "loss": 0.7998,
+      "step": 2181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895499547888556e-05,
+      "loss": 0.8209,
+      "step": 2182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689248775402078e-05,
+      "loss": 0.8527,
+      "step": 2183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889474768659046e-05,
+      "loss": 0.7497,
+      "step": 2184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886460592324203e-05,
+      "loss": 0.7378,
+      "step": 2185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883445225537313e-05,
+      "loss": 0.8067,
+      "step": 2186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688042866881963e-05,
+      "loss": 0.7737,
+      "step": 2187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877410922692633e-05,
+      "loss": 0.8302,
+      "step": 2188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874391987677985e-05,
+      "loss": 0.7479,
+      "step": 2189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687137186429757e-05,
+      "loss": 0.7557,
+      "step": 2190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868350553073472e-05,
+      "loss": 0.7273,
+      "step": 2191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865328054527976e-05,
+      "loss": 0.6955,
+      "step": 2192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862304369183583e-05,
+      "loss": 0.7884,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859279497562985e-05,
+      "loss": 0.805,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856253440189095e-05,
+      "loss": 0.7341,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853226197585022e-05,
+      "loss": 0.765,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850197770274075e-05,
+      "loss": 0.7942,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684716815877978e-05,
+      "loss": 0.758,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844137363625855e-05,
+      "loss": 0.8383,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841105385336237e-05,
+      "loss": 0.7495,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838072224435055e-05,
+      "loss": 0.7744,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835037881446644e-05,
+      "loss": 0.8008,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832002356895553e-05,
+      "loss": 0.9408,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828965651306522e-05,
+      "loss": 0.8764,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825927765204504e-05,
+      "loss": 0.8354,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682288869911466e-05,
+      "loss": 0.8424,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819848453562337e-05,
+      "loss": 0.7956,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681680702907311e-05,
+      "loss": 0.8643,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813764426172735e-05,
+      "loss": 0.8816,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810720645387186e-05,
+      "loss": 0.7278,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807675687242638e-05,
+      "loss": 0.779,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680462955226547e-05,
+      "loss": 0.8408,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680158224098226e-05,
+      "loss": 0.867,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798533753919786e-05,
+      "loss": 0.8471,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679548409160505e-05,
+      "loss": 0.813,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679243325456523e-05,
+      "loss": 0.8322,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789381243327728e-05,
+      "loss": 0.8356,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786328058420132e-05,
+      "loss": 0.6986,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678327370037025e-05,
+      "loss": 0.7405,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6780218169706078e-05,
+      "loss": 0.7751,
+      "step": 2220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777161466955826e-05,
+      "loss": 0.8042,
+      "step": 2221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774103592647893e-05,
+      "loss": 0.7249,
+      "step": 2222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67710445473109e-05,
+      "loss": 0.7658,
+      "step": 2223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767984331473654e-05,
+      "loss": 0.8374,
+      "step": 2224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676492294566517e-05,
+      "loss": 0.8203,
+      "step": 2225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761860390414667e-05,
+      "loss": 0.8061,
+      "step": 2226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758796666251558e-05,
+      "loss": 0.8658,
+      "step": 2227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755731773705472e-05,
+      "loss": 0.8679,
+      "step": 2228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752665713306232e-05,
+      "loss": 0.7294,
+      "step": 2229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749598485583857e-05,
+      "loss": 0.8191,
+      "step": 2230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746530091068583e-05,
+      "loss": 0.8248,
+      "step": 2231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674346053029083e-05,
+      "loss": 0.7818,
+      "step": 2232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740389803781232e-05,
+      "loss": 0.839,
+      "step": 2233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673731791207062e-05,
+      "loss": 0.8264,
+      "step": 2234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673424485569003e-05,
+      "loss": 0.7637,
+      "step": 2235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731170635170693e-05,
+      "loss": 0.7986,
+      "step": 2236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728095251044048e-05,
+      "loss": 0.8646,
+      "step": 2237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672501870384173e-05,
+      "loss": 0.8129,
+      "step": 2238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672194099409558e-05,
+      "loss": 0.8407,
+      "step": 2239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718862122337632e-05,
+      "loss": 0.7472,
+      "step": 2240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671578208910013e-05,
+      "loss": 0.7973,
+      "step": 2241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671270089491551e-05,
+      "loss": 0.8398,
+      "step": 2242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709618540316418e-05,
+      "loss": 0.7414,
+      "step": 2243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706535025835697e-05,
+      "loss": 0.7508,
+      "step": 2244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703450352006385e-05,
+      "loss": 0.7448,
+      "step": 2245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700364519361726e-05,
+      "loss": 0.6872,
+      "step": 2246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697277528435166e-05,
+      "loss": 0.8093,
+      "step": 2247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669418937976035e-05,
+      "loss": 0.8053,
+      "step": 2248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691100073871112e-05,
+      "loss": 0.8162,
+      "step": 2249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688009611301508e-05,
+      "loss": 0.8235,
+      "step": 2250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668491799258577e-05,
+      "loss": 0.7891,
+      "step": 2251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681825218258353e-05,
+      "loss": 0.7824,
+      "step": 2252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678731288853893e-05,
+      "loss": 0.7339,
+      "step": 2253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675636204907233e-05,
+      "loss": 0.8079,
+      "step": 2254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672539966953416e-05,
+      "loss": 0.7843,
+      "step": 2255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669442575527685e-05,
+      "loss": 0.8232,
+      "step": 2256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666634403116548e-05,
+      "loss": 0.8027,
+      "step": 2257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666324433440244e-05,
+      "loss": 0.8111,
+      "step": 2258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660143485774413e-05,
+      "loss": 0.8175,
+      "step": 2259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657041485817433e-05,
+      "loss": 0.7974,
+      "step": 2260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653938335067735e-05,
+      "loss": 0.703,
+      "step": 2261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650834034061763e-05,
+      "loss": 0.7711,
+      "step": 2262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647728583336144e-05,
+      "loss": 0.7406,
+      "step": 2263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664462198342772e-05,
+      "loss": 0.7976,
+      "step": 2264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664151423487352e-05,
+      "loss": 0.8294,
+      "step": 2265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638405338210776e-05,
+      "loss": 0.7621,
+      "step": 2266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635295293976922e-05,
+      "loss": 0.8098,
+      "step": 2267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6632184102709578e-05,
+      "loss": 0.8036,
+      "step": 2268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662907176494658e-05,
+      "loss": 0.7461,
+      "step": 2269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625958281225955e-05,
+      "loss": 0.7755,
+      "step": 2270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622843652085916e-05,
+      "loss": 0.832,
+      "step": 2271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619727878064894e-05,
+      "loss": 0.8394,
+      "step": 2272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616610959701502e-05,
+      "loss": 0.8045,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613492897534557e-05,
+      "loss": 0.7999,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610373692103078e-05,
+      "loss": 0.2801,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607253343946274e-05,
+      "loss": 0.795,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604131853603552e-05,
+      "loss": 0.7488,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660100922161453e-05,
+      "loss": 0.7783,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597885448519004e-05,
+      "loss": 0.8167,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594760534856972e-05,
+      "loss": 0.8445,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659163448116864e-05,
+      "loss": 0.8178,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588507287994405e-05,
+      "loss": 0.7968,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658537895587486e-05,
+      "loss": 0.8205,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582249485350793e-05,
+      "loss": 0.7673,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657911887696319e-05,
+      "loss": 0.808,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575987131253233e-05,
+      "loss": 0.7752,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657285424876231e-05,
+      "loss": 0.7545,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569720230031986e-05,
+      "loss": 0.8401,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566585075604047e-05,
+      "loss": 0.7871,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563448786020452e-05,
+      "loss": 0.8249,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560311361823372e-05,
+      "loss": 0.8223,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557172803555164e-05,
+      "loss": 0.7526,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554033111758392e-05,
+      "loss": 0.8219,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550892286975805e-05,
+      "loss": 0.7855,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547750329750358e-05,
+      "loss": 0.8061,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544607240625187e-05,
+      "loss": 0.8661,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541463020143646e-05,
+      "loss": 0.7844,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.653831766884926e-05,
+      "loss": 0.8132,
+      "step": 2298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653517118728577e-05,
+      "loss": 0.772,
+      "step": 2299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532023575997095e-05,
+      "loss": 0.762,
+      "step": 2300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528874835527366e-05,
+      "loss": 0.9097,
+      "step": 2301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652572496642089e-05,
+      "loss": 0.7319,
+      "step": 2302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522573969222193e-05,
+      "loss": 0.7153,
+      "step": 2303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651942184447598e-05,
+      "loss": 0.8142,
+      "step": 2304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516268592727147e-05,
+      "loss": 0.8102,
+      "step": 2305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65131142145208e-05,
+      "loss": 0.8447,
+      "step": 2306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509958710402224e-05,
+      "loss": 0.8564,
+      "step": 2307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650680208091691e-05,
+      "loss": 0.2951,
+      "step": 2308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650364432661054e-05,
+      "loss": 0.7823,
+      "step": 2309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500485448028992e-05,
+      "loss": 0.8321,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497325445718333e-05,
+      "loss": 0.7048,
+      "step": 2311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649416432022483e-05,
+      "loss": 0.7625,
+      "step": 2312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491002072094942e-05,
+      "loss": 0.6925,
+      "step": 2313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487838701875324e-05,
+      "loss": 0.8302,
+      "step": 2314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484674210112818e-05,
+      "loss": 0.7729,
+      "step": 2315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481508597354468e-05,
+      "loss": 0.7578,
+      "step": 2316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647834186414751e-05,
+      "loss": 0.7837,
+      "step": 2317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647517401103937e-05,
+      "loss": 0.7821,
+      "step": 2318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647200503857767e-05,
+      "loss": 0.7697,
+      "step": 2319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468834947310224e-05,
+      "loss": 0.8042,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465663737785046e-05,
+      "loss": 0.7696,
+      "step": 2321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462491410550336e-05,
+      "loss": 0.8339,
+      "step": 2322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459317966154484e-05,
+      "loss": 0.7598,
+      "step": 2323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456143405146088e-05,
+      "loss": 0.7851,
+      "step": 2324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452967728073922e-05,
+      "loss": 0.791,
+      "step": 2325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449790935486964e-05,
+      "loss": 0.8902,
+      "step": 2326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446613027934383e-05,
+      "loss": 0.8484,
+      "step": 2327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644343400596553e-05,
+      "loss": 0.8674,
+      "step": 2328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644025387012997e-05,
+      "loss": 0.8199,
+      "step": 2329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643707262097744e-05,
+      "loss": 0.847,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433890259057883e-05,
+      "loss": 0.8336,
+      "step": 2331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430706784921425e-05,
+      "loss": 0.8374,
+      "step": 2332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427522199118393e-05,
+      "loss": 0.8095,
+      "step": 2333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424336502199295e-05,
+      "loss": 0.7491,
+      "step": 2334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421149694714843e-05,
+      "loss": 0.8227,
+      "step": 2335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417961777215932e-05,
+      "loss": 0.807,
+      "step": 2336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414772750253654e-05,
+      "loss": 0.8135,
+      "step": 2337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641158261437929e-05,
+      "loss": 0.8389,
+      "step": 2338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640839137014432e-05,
+      "loss": 0.8731,
+      "step": 2339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64051990181004e-05,
+      "loss": 0.8037,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402005558799396e-05,
+      "loss": 0.7936,
+      "step": 2341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398810992793352e-05,
+      "loss": 0.8603,
+      "step": 2342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395615320634507e-05,
+      "loss": 0.7774,
+      "step": 2343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392418542875293e-05,
+      "loss": 0.823,
+      "step": 2344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389220660068334e-05,
+      "loss": 0.2849,
+      "step": 2345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638602167276644e-05,
+      "loss": 0.7429,
+      "step": 2346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638282158152262e-05,
+      "loss": 0.8114,
+      "step": 2347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379620386890064e-05,
+      "loss": 0.7485,
+      "step": 2348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637641808942216e-05,
+      "loss": 0.8717,
+      "step": 2349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637321468967248e-05,
+      "loss": 0.7643,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370010188194798e-05,
+      "loss": 0.8388,
+      "step": 2351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366804585543066e-05,
+      "loss": 0.7317,
+      "step": 2352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363597882271438e-05,
+      "loss": 0.7497,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360390078934244e-05,
+      "loss": 0.7731,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357181176086018e-05,
+      "loss": 0.2922,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353971174281474e-05,
+      "loss": 0.8695,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635076007407552e-05,
+      "loss": 0.7579,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634754787602326e-05,
+      "loss": 0.7996,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634433458067998e-05,
+      "loss": 0.7986,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341120188601152e-05,
+      "loss": 0.821,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337904700342446e-05,
+      "loss": 0.7961,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633468811645972e-05,
+      "loss": 0.842,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633147043750902e-05,
+      "loss": 0.797,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632825166404658e-05,
+      "loss": 0.7956,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325031796628824e-05,
+      "loss": 0.8372,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321810835812367e-05,
+      "loss": 0.7882,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318588782154012e-05,
+      "loss": 0.7798,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631536563621075e-05,
+      "loss": 0.8405,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312141398539764e-05,
+      "loss": 0.8472,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6308916069698417e-05,
+      "loss": 0.8261,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305689650244268e-05,
+      "loss": 0.7423,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630246214073507e-05,
+      "loss": 0.7185,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629923354172875e-05,
+      "loss": 0.7701,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296003853783437e-05,
+      "loss": 0.8044,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292773077457433e-05,
+      "loss": 0.8166,
+      "step": 2375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628954121330925e-05,
+      "loss": 0.7817,
+      "step": 2376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286308261897568e-05,
+      "loss": 0.7661,
+      "step": 2377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283074223781262e-05,
+      "loss": 0.777,
+      "step": 2378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279839099519398e-05,
+      "loss": 0.7618,
+      "step": 2379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276602889671225e-05,
+      "loss": 0.7577,
+      "step": 2380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273365594796185e-05,
+      "loss": 0.7835,
+      "step": 2381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270127215453898e-05,
+      "loss": 0.8195,
+      "step": 2382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266887752204186e-05,
+      "loss": 0.8306,
+      "step": 2383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263647205607043e-05,
+      "loss": 0.7513,
+      "step": 2384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260405576222665e-05,
+      "loss": 0.7988,
+      "step": 2385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625716286461142e-05,
+      "loss": 0.765,
+      "step": 2386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625391907133387e-05,
+      "loss": 0.7973,
+      "step": 2387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250674196950772e-05,
+      "loss": 0.8037,
+      "step": 2388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247428242023056e-05,
+      "loss": 0.7902,
+      "step": 2389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244181207111852e-05,
+      "loss": 0.7962,
+      "step": 2390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240933092778462e-05,
+      "loss": 0.7746,
+      "step": 2391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237683899584388e-05,
+      "loss": 0.7209,
+      "step": 2392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234433628091314e-05,
+      "loss": 0.8174,
+      "step": 2393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231182278861104e-05,
+      "loss": 0.8325,
+      "step": 2394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227929852455822e-05,
+      "loss": 0.8048,
+      "step": 2395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62246763494377e-05,
+      "loss": 0.8283,
+      "step": 2396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221421770369172e-05,
+      "loss": 0.7637,
+      "step": 2397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218166115812853e-05,
+      "loss": 0.9022,
+      "step": 2398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6214909386331542e-05,
+      "loss": 0.8053,
+      "step": 2399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211651582488222e-05,
+      "loss": 0.7598,
+      "step": 2400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620839270484607e-05,
+      "loss": 0.7627,
+      "step": 2401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620513275396844e-05,
+      "loss": 0.8231,
+      "step": 2402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201871730418878e-05,
+      "loss": 0.8006,
+      "step": 2403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198609634761106e-05,
+      "loss": 0.7978,
+      "step": 2404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619534646755904e-05,
+      "loss": 0.8054,
+      "step": 2405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619208222937678e-05,
+      "loss": 0.7766,
+      "step": 2406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618881692077861e-05,
+      "loss": 0.8212,
+      "step": 2407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185550542329002e-05,
+      "loss": 0.823,
+      "step": 2408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182283094592604e-05,
+      "loss": 0.787,
+      "step": 2409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179014578134258e-05,
+      "loss": 0.839,
+      "step": 2410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175744993518986e-05,
+      "loss": 0.7484,
+      "step": 2411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172474341311994e-05,
+      "loss": 0.8096,
+      "step": 2412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169202622078682e-05,
+      "loss": 0.7408,
+      "step": 2413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616592983638462e-05,
+      "loss": 0.7989,
+      "step": 2414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616265598479558e-05,
+      "loss": 0.8172,
+      "step": 2415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159381067877495e-05,
+      "loss": 0.7374,
+      "step": 2416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61561050861965e-05,
+      "loss": 0.8055,
+      "step": 2417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152828040318906e-05,
+      "loss": 0.8243,
+      "step": 2418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149549930811213e-05,
+      "loss": 0.7578,
+      "step": 2419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146270758240113e-05,
+      "loss": 0.7832,
+      "step": 2420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614299052317246e-05,
+      "loss": 0.7725,
+      "step": 2421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61397092261753e-05,
+      "loss": 0.7051,
+      "step": 2422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6136426867815874e-05,
+      "loss": 0.7843,
+      "step": 2423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61331434486616e-05,
+      "loss": 0.7659,
+      "step": 2424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612985896928007e-05,
+      "loss": 0.8139,
+      "step": 2425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612657343023907e-05,
+      "loss": 0.8309,
+      "step": 2426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612328683210657e-05,
+      "loss": 0.8095,
+      "step": 2427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119999175450717e-05,
+      "loss": 0.8324,
+      "step": 2428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116710460839837e-05,
+      "loss": 0.7803,
+      "step": 2429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113420688842455e-05,
+      "loss": 0.7278,
+      "step": 2430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110129860027262e-05,
+      "loss": 0.8119,
+      "step": 2431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106837974963143e-05,
+      "loss": 0.7629,
+      "step": 2432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610354503421916e-05,
+      "loss": 0.7971,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610025103836456e-05,
+      "loss": 0.8025,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096955987968766e-05,
+      "loss": 0.8025,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093659883601392e-05,
+      "loss": 0.8243,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609036272583223e-05,
+      "loss": 0.7862,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087064515231255e-05,
+      "loss": 0.7764,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608376525236862e-05,
+      "loss": 0.8068,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608046493781467e-05,
+      "loss": 0.8028,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077163572139925e-05,
+      "loss": 0.8234,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607386115591508e-05,
+      "loss": 0.8488,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070557689711025e-05,
+      "loss": 0.2857,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067253174098826e-05,
+      "loss": 0.8713,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063947609649723e-05,
+      "loss": 0.7742,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606064099693515e-05,
+      "loss": 0.7964,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057333336526715e-05,
+      "loss": 0.7179,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605402462899621e-05,
+      "loss": 0.8382,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60507148749156e-05,
+      "loss": 0.8511,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604740407485704e-05,
+      "loss": 0.7527,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044092229392866e-05,
+      "loss": 0.8242,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604077933909559e-05,
+      "loss": 0.7648,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037465404537905e-05,
+      "loss": 0.7346,
+      "step": 2453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603415042629268e-05,
+      "loss": 0.7621,
+      "step": 2454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6030834404932986e-05,
+      "loss": 0.8159,
+      "step": 2455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602751734103205e-05,
+      "loss": 0.8362,
+      "step": 2456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024199235163287e-05,
+      "loss": 0.7712,
+      "step": 2457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602088008790029e-05,
+      "loss": 0.8529,
+      "step": 2458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601755989981684e-05,
+      "loss": 0.7496,
+      "step": 2459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014238671486894e-05,
+      "loss": 0.8112,
+      "step": 2460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601091640348458e-05,
+      "loss": 0.787,
+      "step": 2461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007593096384228e-05,
+      "loss": 0.8459,
+      "step": 2462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004268750760315e-05,
+      "loss": 0.8248,
+      "step": 2463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000943367187524e-05,
+      "loss": 0.7998,
+      "step": 2464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997616946240713e-05,
+      "loss": 0.8232,
+      "step": 2465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994289488494912e-05,
+      "loss": 0.8203,
+      "step": 2466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990960994525335e-05,
+      "loss": 0.8238,
+      "step": 2467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987631464907366e-05,
+      "loss": 0.782,
+      "step": 2468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984300900216586e-05,
+      "loss": 0.8015,
+      "step": 2469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598096930102874e-05,
+      "loss": 0.2996,
+      "step": 2470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977636667919756e-05,
+      "loss": 0.7591,
+      "step": 2471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974303001465742e-05,
+      "loss": 0.773,
+      "step": 2472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970968302242988e-05,
+      "loss": 0.8293,
+      "step": 2473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596763257082795e-05,
+      "loss": 0.7591,
+      "step": 2474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596429580779728e-05,
+      "loss": 0.7594,
+      "step": 2475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596095801372779e-05,
+      "loss": 0.7662,
+      "step": 2476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957619189196484e-05,
+      "loss": 0.7699,
+      "step": 2477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954279334780543e-05,
+      "loss": 0.7777,
+      "step": 2478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950938451057317e-05,
+      "loss": 0.7402,
+      "step": 2479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947596538604343e-05,
+      "loss": 0.7814,
+      "step": 2480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944253597999332e-05,
+      "loss": 0.81,
+      "step": 2481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594090962982017e-05,
+      "loss": 0.7756,
+      "step": 2482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937564634644927e-05,
+      "loss": 0.7979,
+      "step": 2483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593421861305185e-05,
+      "loss": 0.7658,
+      "step": 2484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593087156561935e-05,
+      "loss": 0.8184,
+      "step": 2485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927523492926035e-05,
+      "loss": 0.7749,
+      "step": 2486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924174395550683e-05,
+      "loss": 0.3099,
+      "step": 2487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592082427407224e-05,
+      "loss": 0.8532,
+      "step": 2488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917473129069838e-05,
+      "loss": 0.2781,
+      "step": 2489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914120961122792e-05,
+      "loss": 0.7428,
+      "step": 2490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910767770810577e-05,
+      "loss": 0.7396,
+      "step": 2491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907413558712858e-05,
+      "loss": 0.7987,
+      "step": 2492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904058325409476e-05,
+      "loss": 0.7125,
+      "step": 2493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900702071480437e-05,
+      "loss": 0.8197,
+      "step": 2494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897344797505936e-05,
+      "loss": 0.8517,
+      "step": 2495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589398650406634e-05,
+      "loss": 0.8005,
+      "step": 2496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890627191742195e-05,
+      "loss": 0.8202,
+      "step": 2497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887266861114216e-05,
+      "loss": 0.7764,
+      "step": 2498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883905512763296e-05,
+      "loss": 0.7754,
+      "step": 2499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880543147270514e-05,
+      "loss": 0.8419,
+      "step": 2500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877179765217107e-05,
+      "loss": 0.8434,
+      "step": 2501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873815367184508e-05,
+      "loss": 0.8126,
+      "step": 2502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587044995375431e-05,
+      "loss": 0.8548,
+      "step": 2503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586708352550829e-05,
+      "loss": 0.8173,
+      "step": 2504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586371608302839e-05,
+      "loss": 0.777,
+      "step": 2505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860347626896745e-05,
+      "loss": 0.7406,
+      "step": 2506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5856978157695645e-05,
+      "loss": 0.6979,
+      "step": 2507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853607676007575e-05,
+      "loss": 0.763,
+      "step": 2508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850236182415177e-05,
+      "loss": 0.8401,
+      "step": 2509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584686367750128e-05,
+      "loss": 0.7679,
+      "step": 2510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584349016184888e-05,
+      "loss": 0.841,
+      "step": 2511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584011563604116e-05,
+      "loss": 0.7716,
+      "step": 2512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836740100661456e-05,
+      "loss": 0.8225,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833363556293305e-05,
+      "loss": 0.8065,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829986003520395e-05,
+      "loss": 0.8712,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5826607442926607e-05,
+      "loss": 0.8308,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823227875095984e-05,
+      "loss": 0.8104,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819847300612744e-05,
+      "loss": 0.7765,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816465720061283e-05,
+      "loss": 0.7482,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813083134026172e-05,
+      "loss": 0.7824,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809699543092155e-05,
+      "loss": 0.8069,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806314947844142e-05,
+      "loss": 0.7928,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802929348867226e-05,
+      "loss": 0.7828,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799542746746674e-05,
+      "loss": 0.7819,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579615514206792e-05,
+      "loss": 0.8252,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579276653541657e-05,
+      "loss": 0.8288,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789376927378413e-05,
+      "loss": 0.2856,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785986318539404e-05,
+      "loss": 0.7788,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578259470948567e-05,
+      "loss": 0.8185,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577920210080352e-05,
+      "loss": 0.8176,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577580849307942e-05,
+      "loss": 0.8584,
+      "step": 2531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772413886900024e-05,
+      "loss": 0.8073,
+      "step": 2532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769018282852152e-05,
+      "loss": 0.8183,
+      "step": 2533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765621681522795e-05,
+      "loss": 0.7183,
+      "step": 2534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762224083499114e-05,
+      "loss": 0.7576,
+      "step": 2535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758825489368458e-05,
+      "loss": 0.8373,
+      "step": 2536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575542589971833e-05,
+      "loss": 0.8119,
+      "step": 2537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575202531513641e-05,
+      "loss": 0.7712,
+      "step": 2538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5748623736210555e-05,
+      "loss": 0.7363,
+      "step": 2539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574522116352879e-05,
+      "loss": 0.8207,
+      "step": 2540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574181759767931e-05,
+      "loss": 0.7532,
+      "step": 2541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738413039250494e-05,
+      "loss": 0.8348,
+      "step": 2542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735007488830873e-05,
+      "loss": 0.8121,
+      "step": 2543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731600947009158e-05,
+      "loss": 0.8161,
+      "step": 2544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572819341437424e-05,
+      "loss": 0.8412,
+      "step": 2545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724784891515172e-05,
+      "loss": 0.7848,
+      "step": 2546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572137537902118e-05,
+      "loss": 0.2645,
+      "step": 2547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571796487748166e-05,
+      "loss": 0.8025,
+      "step": 2548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571455338748618e-05,
+      "loss": 0.8429,
+      "step": 2549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571114090962448e-05,
+      "loss": 0.8123,
+      "step": 2550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570772744448647e-05,
+      "loss": 0.8037,
+      "step": 2551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704312992662232e-05,
+      "loss": 0.7796,
+      "step": 2552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700897554742015e-05,
+      "loss": 0.7282,
+      "step": 2553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697481131316238e-05,
+      "loss": 0.8337,
+      "step": 2554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56940637229755e-05,
+      "loss": 0.7717,
+      "step": 2555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569064533031056e-05,
+      "loss": 0.7546,
+      "step": 2556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687225953912343e-05,
+      "loss": 0.8471,
+      "step": 2557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568380559437196e-05,
+      "loss": 0.7459,
+      "step": 2558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5680384252280685e-05,
+      "loss": 0.8582,
+      "step": 2559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567696192822996e-05,
+      "loss": 0.8248,
+      "step": 2560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567353862281139e-05,
+      "loss": 0.9123,
+      "step": 2561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567011433661676e-05,
+      "loss": 0.8054,
+      "step": 2562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566668907023802e-05,
+      "loss": 0.7697,
+      "step": 2563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663262824267293e-05,
+      "loss": 0.8041,
+      "step": 2564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659835599296872e-05,
+      "loss": 0.8335,
+      "step": 2565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565640739591921e-05,
+      "loss": 0.8194,
+      "step": 2566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565297821472694e-05,
+      "loss": 0.8715,
+      "step": 2567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649548056312852e-05,
+      "loss": 0.8086,
+      "step": 2568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564611692126992e-05,
+      "loss": 0.8171,
+      "step": 2569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642684810191274e-05,
+      "loss": 0.8804,
+      "step": 2570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639251723670222e-05,
+      "loss": 0.7681,
+      "step": 2571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635817662300235e-05,
+      "loss": 0.7778,
+      "step": 2572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632382626674955e-05,
+      "loss": 0.818,
+      "step": 2573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628946617388195e-05,
+      "loss": 0.854,
+      "step": 2574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625509635033922e-05,
+      "loss": 0.7735,
+      "step": 2575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562207168020629e-05,
+      "loss": 0.7687,
+      "step": 2576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561863275349961e-05,
+      "loss": 0.7844,
+      "step": 2577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615192855508366e-05,
+      "loss": 0.798,
+      "step": 2578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611751986827205e-05,
+      "loss": 0.748,
+      "step": 2579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608310148050953e-05,
+      "loss": 0.8156,
+      "step": 2580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560486733977458e-05,
+      "loss": 0.7456,
+      "step": 2581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5601423562593255e-05,
+      "loss": 0.7887,
+      "step": 2582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559797881710229e-05,
+      "loss": 0.799,
+      "step": 2583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594533103897175e-05,
+      "loss": 0.7689,
+      "step": 2584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559108642357357e-05,
+      "loss": 0.7379,
+      "step": 2585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558763877672729e-05,
+      "loss": 0.8296,
+      "step": 2586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584190163954325e-05,
+      "loss": 0.76,
+      "step": 2587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580740585850833e-05,
+      "loss": 0.812,
+      "step": 2588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557729004301314e-05,
+      "loss": 0.7652,
+      "step": 2589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573838536037734e-05,
+      "loss": 0.2933,
+      "step": 2590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5570386065521272e-05,
+      "loss": 0.8568,
+      "step": 2591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566932632060575e-05,
+      "loss": 0.7353,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563478236252638e-05,
+      "loss": 0.86,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560022878694616e-05,
+      "loss": 0.8121,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556566559983825e-05,
+      "loss": 0.7208,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555310928071776e-05,
+      "loss": 0.8083,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549651041494074e-05,
+      "loss": 0.7455,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546191842910593e-05,
+      "loss": 0.7226,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554273168556529e-05,
+      "loss": 0.8152,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553927057005633e-05,
+      "loss": 0.2728,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535808496982026e-05,
+      "loss": 0.776,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553234546694086e-05,
+      "loss": 0.7426,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528881480531482e-05,
+      "loss": 0.8218,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525416538352707e-05,
+      "loss": 0.7953,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521950641003514e-05,
+      "loss": 0.8541,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518483789083047e-05,
+      "loss": 0.8165,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515015983190614e-05,
+      "loss": 0.8001,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511547223925693e-05,
+      "loss": 0.7723,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550807751188792e-05,
+      "loss": 0.8539,
+      "step": 2609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55046068476771e-05,
+      "loss": 0.7774,
+      "step": 2610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55011352318932e-05,
+      "loss": 0.7839,
+      "step": 2611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497662665136357e-05,
+      "loss": 0.7776,
+      "step": 2612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5494189148006866e-05,
+      "loss": 0.7469,
+      "step": 2613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490714681105185e-05,
+      "loss": 0.7835,
+      "step": 2614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487239265031947e-05,
+      "loss": 0.2846,
+      "step": 2615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483762900387937e-05,
+      "loss": 0.7614,
+      "step": 2616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548028558777411e-05,
+      "loss": 0.7991,
+      "step": 2617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5476807327791586e-05,
+      "loss": 0.8382,
+      "step": 2618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547332812104164e-05,
+      "loss": 0.7003,
+      "step": 2619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469847968125723e-05,
+      "loss": 0.721,
+      "step": 2620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466366869645442e-05,
+      "loss": 0.742,
+      "step": 2621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462884826202574e-05,
+      "loss": 0.7174,
+      "step": 2622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459401838399048e-05,
+      "loss": 0.2943,
+      "step": 2623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455917906836965e-05,
+      "loss": 0.7682,
+      "step": 2624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5452433032118586e-05,
+      "loss": 0.8533,
+      "step": 2625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448947214846342e-05,
+      "loss": 0.8005,
+      "step": 2626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544546045562281e-05,
+      "loss": 0.7325,
+      "step": 2627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544197275505075e-05,
+      "loss": 0.747,
+      "step": 2628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438484113733073e-05,
+      "loss": 0.756,
+      "step": 2629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434994532272852e-05,
+      "loss": 0.8588,
+      "step": 2630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543150401127333e-05,
+      "loss": 0.8017,
+      "step": 2631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428012551337908e-05,
+      "loss": 0.8043,
+      "step": 2632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424520153070145e-05,
+      "loss": 0.8058,
+      "step": 2633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421026817073773e-05,
+      "loss": 0.8109,
+      "step": 2634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541753254395267e-05,
+      "loss": 0.8642,
+      "step": 2635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414037334310897e-05,
+      "loss": 0.7789,
+      "step": 2636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410541188752662e-05,
+      "loss": 0.6911,
+      "step": 2637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407044107882335e-05,
+      "loss": 0.8151,
+      "step": 2638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540354609230445e-05,
+      "loss": 0.7816,
+      "step": 2639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540004714262371e-05,
+      "loss": 0.7759,
+      "step": 2640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396547259444968e-05,
+      "loss": 0.7825,
+      "step": 2641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393046443373246e-05,
+      "loss": 0.7998,
+      "step": 2642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389544695013728e-05,
+      "loss": 0.7961,
+      "step": 2643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538604201497175e-05,
+      "loss": 0.7623,
+      "step": 2644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382538403852815e-05,
+      "loss": 0.7918,
+      "step": 2645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537903386226259e-05,
+      "loss": 0.8533,
+      "step": 2646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53755283908069e-05,
+      "loss": 0.8667,
+      "step": 2647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372021990091725e-05,
+      "loss": 0.8077,
+      "step": 2648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536851466072322e-05,
+      "loss": 0.7435,
+      "step": 2649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365006403307686e-05,
+      "loss": 0.8306,
+      "step": 2650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361497218451593e-05,
+      "loss": 0.8004,
+      "step": 2651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357987106761566e-05,
+      "loss": 0.8226,
+      "step": 2652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354476068844396e-05,
+      "loss": 0.8199,
+      "step": 2653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350964105307023e-05,
+      "loss": 0.7626,
+      "step": 2654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534745121675657e-05,
+      "loss": 0.7418,
+      "step": 2655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343937403800286e-05,
+      "loss": 0.815,
+      "step": 2656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340422667045612e-05,
+      "loss": 0.7757,
+      "step": 2657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533690700710013e-05,
+      "loss": 0.8424,
+      "step": 2658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533339042457159e-05,
+      "loss": 0.7959,
+      "step": 2659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5329872920067893e-05,
+      "loss": 0.7624,
+      "step": 2660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326354494197108e-05,
+      "loss": 0.8166,
+      "step": 2661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532283514756746e-05,
+      "loss": 0.7848,
+      "step": 2662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319314880787335e-05,
+      "loss": 0.8676,
+      "step": 2663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315793694465272e-05,
+      "loss": 0.7748,
+      "step": 2664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531227158920998e-05,
+      "loss": 0.7965,
+      "step": 2665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308748565630306e-05,
+      "loss": 0.7924,
+      "step": 2666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305224624335283e-05,
+      "loss": 0.7486,
+      "step": 2667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530169976593409e-05,
+      "loss": 0.7508,
+      "step": 2668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298173991036058e-05,
+      "loss": 0.8012,
+      "step": 2669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529464730025068e-05,
+      "loss": 0.8009,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291119694187616e-05,
+      "loss": 0.8299,
+      "step": 2671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528759117345668e-05,
+      "loss": 0.8437,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528406173866783e-05,
+      "loss": 0.7478,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280531390431203e-05,
+      "loss": 0.8055,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277000129357086e-05,
+      "loss": 0.7514,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527346795605592e-05,
+      "loss": 0.8283,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269934871138303e-05,
+      "loss": 0.8408,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266400875215004e-05,
+      "loss": 0.8288,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526286596889693e-05,
+      "loss": 0.7864,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525933015279516e-05,
+      "loss": 0.7651,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255793427520922e-05,
+      "loss": 0.793,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252255793685606e-05,
+      "loss": 0.7137,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248717251900758e-05,
+      "loss": 0.796,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245177802778081e-05,
+      "loss": 0.7908,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524163744692943e-05,
+      "loss": 0.8097,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238096184966823e-05,
+      "loss": 0.7178,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234554017502437e-05,
+      "loss": 0.7942,
+      "step": 2687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231010945148596e-05,
+      "loss": 0.8041,
+      "step": 2688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227466968517789e-05,
+      "loss": 0.7531,
+      "step": 2689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223922088222654e-05,
+      "loss": 0.7211,
+      "step": 2690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220376304875993e-05,
+      "loss": 0.7664,
+      "step": 2691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216829619090756e-05,
+      "loss": 0.7311,
+      "step": 2692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213282031480057e-05,
+      "loss": 0.6797,
+      "step": 2693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209733542657161e-05,
+      "loss": 0.7652,
+      "step": 2694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520618415323549e-05,
+      "loss": 0.765,
+      "step": 2695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202633863828622e-05,
+      "loss": 0.7895,
+      "step": 2696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199082675050287e-05,
+      "loss": 0.7484,
+      "step": 2697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195530587514377e-05,
+      "loss": 0.7497,
+      "step": 2698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191977601834936e-05,
+      "loss": 0.7785,
+      "step": 2699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188423718626162e-05,
+      "loss": 0.7262,
+      "step": 2700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184868938502409e-05,
+      "loss": 0.7957,
+      "step": 2701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518131326207819e-05,
+      "loss": 0.7555,
+      "step": 2702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177756689968163e-05,
+      "loss": 0.8091,
+      "step": 2703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174199222787151e-05,
+      "loss": 0.7896,
+      "step": 2704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170640861150131e-05,
+      "loss": 0.7122,
+      "step": 2705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167081605672227e-05,
+      "loss": 0.8813,
+      "step": 2706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163521456968724e-05,
+      "loss": 0.7514,
+      "step": 2707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515996041565506e-05,
+      "loss": 0.8292,
+      "step": 2708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156398482346824e-05,
+      "loss": 0.7313,
+      "step": 2709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152835657659766e-05,
+      "loss": 0.7615,
+      "step": 2710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149271942209782e-05,
+      "loss": 0.8339,
+      "step": 2711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145707336612932e-05,
+      "loss": 0.7811,
+      "step": 2712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142141841485418e-05,
+      "loss": 0.792,
+      "step": 2713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138575457443603e-05,
+      "loss": 0.709,
+      "step": 2714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135008185104006e-05,
+      "loss": 0.8059,
+      "step": 2715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513144002508329e-05,
+      "loss": 0.8315,
+      "step": 2716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127870977998282e-05,
+      "loss": 0.8804,
+      "step": 2717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124301044465958e-05,
+      "loss": 0.8059,
+      "step": 2718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120730225103449e-05,
+      "loss": 0.7714,
+      "step": 2719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117158520528024e-05,
+      "loss": 0.7322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113585931357135e-05,
+      "loss": 0.782,
+      "step": 2721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110012458208365e-05,
+      "loss": 0.7183,
+      "step": 2722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106438101699448e-05,
+      "loss": 0.263,
+      "step": 2723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102862862448286e-05,
+      "loss": 0.7188,
+      "step": 2724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099286741072923e-05,
+      "loss": 0.7798,
+      "step": 2725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095709738191553e-05,
+      "loss": 0.7919,
+      "step": 2726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092131854422531e-05,
+      "loss": 0.7477,
+      "step": 2727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088553090384363e-05,
+      "loss": 0.8568,
+      "step": 2728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50849734466957e-05,
+      "loss": 0.7437,
+      "step": 2729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508139292397535e-05,
+      "loss": 0.7715,
+      "step": 2730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077811522842276e-05,
+      "loss": 0.2833,
+      "step": 2731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074229243915585e-05,
+      "loss": 0.8422,
+      "step": 2732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070646087814542e-05,
+      "loss": 0.735,
+      "step": 2733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067062055158563e-05,
+      "loss": 0.8168,
+      "step": 2734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063477146567216e-05,
+      "loss": 0.7661,
+      "step": 2735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059891362660212e-05,
+      "loss": 0.7938,
+      "step": 2736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056304704057429e-05,
+      "loss": 0.7339,
+      "step": 2737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052717171378879e-05,
+      "loss": 0.7688,
+      "step": 2738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049128765244736e-05,
+      "loss": 0.7364,
+      "step": 2739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045539486275327e-05,
+      "loss": 0.8037,
+      "step": 2740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504194933509112e-05,
+      "loss": 0.2898,
+      "step": 2741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038358312312737e-05,
+      "loss": 0.8255,
+      "step": 2742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503476641856096e-05,
+      "loss": 0.75,
+      "step": 2743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031173654456707e-05,
+      "loss": 0.7283,
+      "step": 2744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027580020621061e-05,
+      "loss": 0.8026,
+      "step": 2745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502398551767524e-05,
+      "loss": 0.7792,
+      "step": 2746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020390146240628e-05,
+      "loss": 0.3261,
+      "step": 2747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016793906938745e-05,
+      "loss": 0.76,
+      "step": 2748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013196800391274e-05,
+      "loss": 0.7365,
+      "step": 2749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009598827220036e-05,
+      "loss": 0.7967,
+      "step": 2750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005999988047006e-05,
+      "loss": 0.8435,
+      "step": 2751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002400283494318e-05,
+      "loss": 0.7921,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499879971418424e-05,
+      "loss": 0.7617,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995198280739199e-05,
+      "loss": 0.8131,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991595983781768e-05,
+      "loss": 0.772,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987992823934673e-05,
+      "loss": 0.7519,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498438880182079e-05,
+      "loss": 0.7754,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980783918063134e-05,
+      "loss": 0.8046,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977178173284878e-05,
+      "loss": 0.7627,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973571568109344e-05,
+      "loss": 0.7786,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969964103159997e-05,
+      "loss": 0.7351,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966355779060457e-05,
+      "loss": 0.7968,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496274659643449e-05,
+      "loss": 0.7724,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4959136555906004e-05,
+      "loss": 0.7996,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495552565809907e-05,
+      "loss": 0.7335,
+      "step": 2765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951913903637893e-05,
+      "loss": 0.8107,
+      "step": 2766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948301293146829e-05,
+      "loss": 0.7747,
+      "step": 2767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494468782725039e-05,
+      "loss": 0.8117,
+      "step": 2768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941073506573232e-05,
+      "loss": 0.7583,
+      "step": 2769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937458331740151e-05,
+      "loss": 0.7303,
+      "step": 2770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933842303376098e-05,
+      "loss": 0.766,
+      "step": 2771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930225422106175e-05,
+      "loss": 0.8122,
+      "step": 2772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926607688555623e-05,
+      "loss": 0.7723,
+      "step": 2773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922989103349835e-05,
+      "loss": 0.6996,
+      "step": 2774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491936966711435e-05,
+      "loss": 0.801,
+      "step": 2775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915749380474853e-05,
+      "loss": 0.8697,
+      "step": 2776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912128244057178e-05,
+      "loss": 0.73,
+      "step": 2777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908506258487307e-05,
+      "loss": 0.7688,
+      "step": 2778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490488342439137e-05,
+      "loss": 0.8473,
+      "step": 2779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901259742395635e-05,
+      "loss": 0.8151,
+      "step": 2780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897635213126527e-05,
+      "loss": 0.7956,
+      "step": 2781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489400983721061e-05,
+      "loss": 0.7792,
+      "step": 2782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890383615274595e-05,
+      "loss": 0.7812,
+      "step": 2783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886756547945348e-05,
+      "loss": 0.7406,
+      "step": 2784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883128635849873e-05,
+      "loss": 0.8088,
+      "step": 2785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879499879615317e-05,
+      "loss": 0.777,
+      "step": 2786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875870279868982e-05,
+      "loss": 0.734,
+      "step": 2787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872239837238312e-05,
+      "loss": 0.7592,
+      "step": 2788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486860855235089e-05,
+      "loss": 0.7589,
+      "step": 2789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864976425834456e-05,
+      "loss": 0.8207,
+      "step": 2790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486134345831689e-05,
+      "loss": 0.8015,
+      "step": 2791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857709650426217e-05,
+      "loss": 0.7956,
+      "step": 2792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854075002790608e-05,
+      "loss": 0.8049,
+      "step": 2793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850439516038378e-05,
+      "loss": 0.8137,
+      "step": 2794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846803190797991e-05,
+      "loss": 0.793,
+      "step": 2795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843166027698046e-05,
+      "loss": 0.817,
+      "step": 2796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839528027367305e-05,
+      "loss": 0.8036,
+      "step": 2797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835889190434653e-05,
+      "loss": 0.7822,
+      "step": 2798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832249517529136e-05,
+      "loss": 0.8016,
+      "step": 2799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828609009279937e-05,
+      "loss": 0.7558,
+      "step": 2800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824967666316385e-05,
+      "loss": 0.8115,
+      "step": 2801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821325489267955e-05,
+      "loss": 0.8415,
+      "step": 2802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817682478764263e-05,
+      "loss": 0.8247,
+      "step": 2803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814038635435072e-05,
+      "loss": 0.7593,
+      "step": 2804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4810393959910288e-05,
+      "loss": 0.7709,
+      "step": 2805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480674845281996e-05,
+      "loss": 0.8023,
+      "step": 2806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803102114794282e-05,
+      "loss": 0.7907,
+      "step": 2807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799454946463586e-05,
+      "loss": 0.779,
+      "step": 2808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795806948458358e-05,
+      "loss": 0.787,
+      "step": 2809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479215812140922e-05,
+      "loss": 0.7937,
+      "step": 2810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478850846594694e-05,
+      "loss": 0.831,
+      "step": 2811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478485798270243e-05,
+      "loss": 0.7521,
+      "step": 2812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478120667230674e-05,
+      "loss": 0.8063,
+      "step": 2813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777554535391067e-05,
+      "loss": 0.8178,
+      "step": 2814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773901572586753e-05,
+      "loss": 0.8009,
+      "step": 2815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770247784525278e-05,
+      "loss": 0.7704,
+      "step": 2816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766593171838268e-05,
+      "loss": 0.8085,
+      "step": 2817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762937735157488e-05,
+      "loss": 0.8015,
+      "step": 2818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759281475114854e-05,
+      "loss": 0.7461,
+      "step": 2819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755624392342409e-05,
+      "loss": 0.7148,
+      "step": 2820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751966487472353e-05,
+      "loss": 0.7291,
+      "step": 2821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748307761137023e-05,
+      "loss": 0.7696,
+      "step": 2822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744648213968897e-05,
+      "loss": 0.7426,
+      "step": 2823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740987846600594e-05,
+      "loss": 0.8053,
+      "step": 2824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737326659664875e-05,
+      "loss": 0.7435,
+      "step": 2825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733664653794646e-05,
+      "loss": 0.7925,
+      "step": 2826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730001829622951e-05,
+      "loss": 0.8098,
+      "step": 2827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726338187782983e-05,
+      "loss": 0.8167,
+      "step": 2828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722673728908062e-05,
+      "loss": 0.7699,
+      "step": 2829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719008453631659e-05,
+      "loss": 0.8129,
+      "step": 2830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715342362587387e-05,
+      "loss": 0.7367,
+      "step": 2831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711675456409001e-05,
+      "loss": 0.7491,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708007735730383e-05,
+      "loss": 0.2847,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704339201185577e-05,
+      "loss": 0.8759,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700669853408754e-05,
+      "loss": 0.785,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696999693034223e-05,
+      "loss": 0.7523,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693328720696445e-05,
+      "loss": 0.7531,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689656937030014e-05,
+      "loss": 0.8451,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685984342669668e-05,
+      "loss": 0.8348,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682310938250277e-05,
+      "loss": 0.7349,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678636724406863e-05,
+      "loss": 0.7815,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674961701774579e-05,
+      "loss": 0.8544,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671285870988718e-05,
+      "loss": 0.7788,
+      "step": 2843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466760923268472e-05,
+      "loss": 0.85,
+      "step": 2844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663931787498161e-05,
+      "loss": 0.7886,
+      "step": 2845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660253536064747e-05,
+      "loss": 0.798,
+      "step": 2846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465657447902034e-05,
+      "loss": 0.7673,
+      "step": 2847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465289461700093e-05,
+      "loss": 0.7844,
+      "step": 2848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649213950642648e-05,
+      "loss": 0.7968,
+      "step": 2849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645532480581768e-05,
+      "loss": 0.7497,
+      "step": 2850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641850207454699e-05,
+      "loss": 0.8627,
+      "step": 2851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463816713189799e-05,
+      "loss": 0.7983,
+      "step": 2852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463448325454833e-05,
+      "loss": 0.8041,
+      "step": 2853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463079857604254e-05,
+      "loss": 0.8573,
+      "step": 2854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627113097017592e-05,
+      "loss": 0.8305,
+      "step": 2855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623426818110585e-05,
+      "loss": 0.7851,
+      "step": 2856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461973973995876e-05,
+      "loss": 0.8164,
+      "step": 2857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46160518631995e-05,
+      "loss": 0.7519,
+      "step": 2858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612363188470316e-05,
+      "loss": 0.8003,
+      "step": 2859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608673716408875e-05,
+      "loss": 0.7455,
+      "step": 2860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604983447652959e-05,
+      "loss": 0.8228,
+      "step": 2861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601292382840507e-05,
+      "loss": 0.7409,
+      "step": 2862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597600522609581e-05,
+      "loss": 0.8408,
+      "step": 2863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593907867598396e-05,
+      "loss": 0.818,
+      "step": 2864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590214418445285e-05,
+      "loss": 0.8147,
+      "step": 2865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586520175788738e-05,
+      "loss": 0.2785,
+      "step": 2866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4582825140267374e-05,
+      "loss": 0.7639,
+      "step": 2867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579129312519938e-05,
+      "loss": 0.7891,
+      "step": 2868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457543269318533e-05,
+      "loss": 0.7496,
+      "step": 2869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457173528290258e-05,
+      "loss": 0.7724,
+      "step": 2870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568037082310846e-05,
+      "loss": 0.7789,
+      "step": 2871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4564338092049436e-05,
+      "loss": 0.7226,
+      "step": 2872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560638312757791e-05,
+      "loss": 0.7222,
+      "step": 2873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556937745075484e-05,
+      "loss": 0.7811,
+      "step": 2874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553236389642225e-05,
+      "loss": 0.7641,
+      "step": 2875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549534247097862e-05,
+      "loss": 0.8095,
+      "step": 2876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545831318082377e-05,
+      "loss": 0.8289,
+      "step": 2877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542127603235899e-05,
+      "loss": 0.7422,
+      "step": 2878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538423103198673e-05,
+      "loss": 0.7492,
+      "step": 2879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534717818611092e-05,
+      "loss": 0.8468,
+      "step": 2880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531011750113688e-05,
+      "loss": 0.7922,
+      "step": 2881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527304898347123e-05,
+      "loss": 0.667,
+      "step": 2882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452359726395219e-05,
+      "loss": 0.8077,
+      "step": 2883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519888847569822e-05,
+      "loss": 0.8609,
+      "step": 2884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516179649841093e-05,
+      "loss": 0.7875,
+      "step": 2885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512469671407205e-05,
+      "loss": 0.8176,
+      "step": 2886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450875891290949e-05,
+      "loss": 0.7689,
+      "step": 2887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505047374989432e-05,
+      "loss": 0.2442,
+      "step": 2888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450133505828863e-05,
+      "loss": 0.7443,
+      "step": 2889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449762196344883e-05,
+      "loss": 0.7786,
+      "step": 2890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493908091111908e-05,
+      "loss": 0.7407,
+      "step": 2891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490193441919877e-05,
+      "loss": 0.787,
+      "step": 2892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486478016514882e-05,
+      "loss": 0.8275,
+      "step": 2893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482761815539201e-05,
+      "loss": 0.7742,
+      "step": 2894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479044839635253e-05,
+      "loss": 0.7971,
+      "step": 2895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447532708944558e-05,
+      "loss": 0.8315,
+      "step": 2896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471608565612868e-05,
+      "loss": 0.7987,
+      "step": 2897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467889268779935e-05,
+      "loss": 0.779,
+      "step": 2898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464169199589724e-05,
+      "loss": 0.7672,
+      "step": 2899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446044835868532e-05,
+      "loss": 0.2986,
+      "step": 2900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456726746709941e-05,
+      "loss": 0.6832,
+      "step": 2901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453004364306936e-05,
+      "loss": 0.7395,
+      "step": 2902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449281212119783e-05,
+      "loss": 0.7832,
+      "step": 2903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445557290792105e-05,
+      "loss": 0.7501,
+      "step": 2904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441832600967645e-05,
+      "loss": 0.7239,
+      "step": 2905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438107143290286e-05,
+      "loss": 0.699,
+      "step": 2906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434380918404042e-05,
+      "loss": 0.7259,
+      "step": 2907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430653926953064e-05,
+      "loss": 0.7461,
+      "step": 2908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426926169581627e-05,
+      "loss": 0.8051,
+      "step": 2909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423197646934144e-05,
+      "loss": 0.6845,
+      "step": 2910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441946835965516e-05,
+      "loss": 0.7931,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415738308389348e-05,
+      "loss": 0.7493,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412007493781522e-05,
+      "loss": 0.7477,
+      "step": 2913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408275916476617e-05,
+      "loss": 0.7722,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404543577119705e-05,
+      "loss": 0.6987,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400810476355993e-05,
+      "loss": 0.7773,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397076614830816e-05,
+      "loss": 0.2678,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439334199318964e-05,
+      "loss": 0.8702,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389606612078064e-05,
+      "loss": 0.805,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4385870472141818e-05,
+      "loss": 0.819,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382133574026762e-05,
+      "loss": 0.8161,
+      "step": 2921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378395918378887e-05,
+      "loss": 0.7992,
+      "step": 2922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374657505844321e-05,
+      "loss": 0.7229,
+      "step": 2923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437091833706931e-05,
+      "loss": 0.7828,
+      "step": 2924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367178412700242e-05,
+      "loss": 0.7626,
+      "step": 2925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363437733383636e-05,
+      "loss": 0.8379,
+      "step": 2926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359696299766132e-05,
+      "loss": 0.7888,
+      "step": 2927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355954112494509e-05,
+      "loss": 0.7577,
+      "step": 2928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435221117221567e-05,
+      "loss": 0.733,
+      "step": 2929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348467479576656e-05,
+      "loss": 0.8219,
+      "step": 2930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344723035224629e-05,
+      "loss": 0.264,
+      "step": 2931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340977839806888e-05,
+      "loss": 0.8066,
+      "step": 2932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337231893970862e-05,
+      "loss": 0.7765,
+      "step": 2933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333485198364101e-05,
+      "loss": 0.7954,
+      "step": 2934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329737753634296e-05,
+      "loss": 0.7728,
+      "step": 2935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432598956042926e-05,
+      "loss": 0.7132,
+      "step": 2936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322240619396935e-05,
+      "loss": 0.8004,
+      "step": 2937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318490931185397e-05,
+      "loss": 0.7468,
+      "step": 2938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314740496442852e-05,
+      "loss": 0.8068,
+      "step": 2939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310989315817624e-05,
+      "loss": 0.2505,
+      "step": 2940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307237389958179e-05,
+      "loss": 0.8195,
+      "step": 2941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303484719513112e-05,
+      "loss": 0.7664,
+      "step": 2942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299731305131133e-05,
+      "loss": 0.7477,
+      "step": 2943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295977147461091e-05,
+      "loss": 0.8085,
+      "step": 2944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292222247151966e-05,
+      "loss": 0.2812,
+      "step": 2945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288466604852857e-05,
+      "loss": 0.788,
+      "step": 2946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284710221212996e-05,
+      "loss": 0.7831,
+      "step": 2947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280953096881747e-05,
+      "loss": 0.7215,
+      "step": 2948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277195232508598e-05,
+      "loss": 0.7876,
+      "step": 2949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427343662874316e-05,
+      "loss": 0.8036,
+      "step": 2950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269677286235186e-05,
+      "loss": 0.809,
+      "step": 2951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426591720563454e-05,
+      "loss": 0.8534,
+      "step": 2952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262156387591225e-05,
+      "loss": 0.7888,
+      "step": 2953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258394832755367e-05,
+      "loss": 0.8215,
+      "step": 2954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254632541777223e-05,
+      "loss": 0.7043,
+      "step": 2955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250869515307169e-05,
+      "loss": 0.7551,
+      "step": 2956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424710575399572e-05,
+      "loss": 0.7885,
+      "step": 2957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243341258493509e-05,
+      "loss": 0.8277,
+      "step": 2958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239576029451296e-05,
+      "loss": 0.8195,
+      "step": 2959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235810067519974e-05,
+      "loss": 0.8582,
+      "step": 2960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232043373350559e-05,
+      "loss": 0.7421,
+      "step": 2961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422827594759419e-05,
+      "loss": 0.8057,
+      "step": 2962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224507790902142e-05,
+      "loss": 0.7659,
+      "step": 2963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220738903925806e-05,
+      "loss": 0.7903,
+      "step": 2964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216969287316706e-05,
+      "loss": 0.8109,
+      "step": 2965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213198941726487e-05,
+      "loss": 0.728,
+      "step": 2966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209427867806929e-05,
+      "loss": 0.7225,
+      "step": 2967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205656066209928e-05,
+      "loss": 0.7971,
+      "step": 2968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201883537587506e-05,
+      "loss": 0.7275,
+      "step": 2969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198110282591819e-05,
+      "loss": 0.7856,
+      "step": 2970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194336301875147e-05,
+      "loss": 0.7818,
+      "step": 2971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190561596089885e-05,
+      "loss": 0.7493,
+      "step": 2972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186786165888565e-05,
+      "loss": 0.8001,
+      "step": 2973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183010011923839e-05,
+      "loss": 0.2723,
+      "step": 2974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417923313484848e-05,
+      "loss": 0.7849,
+      "step": 2975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175455535315404e-05,
+      "loss": 0.2739,
+      "step": 2976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171677213977626e-05,
+      "loss": 0.7906,
+      "step": 2977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167898171488303e-05,
+      "loss": 0.721,
+      "step": 2978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164118408500717e-05,
+      "loss": 0.7773,
+      "step": 2979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160337925668266e-05,
+      "loss": 0.8472,
+      "step": 2980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156556723644473e-05,
+      "loss": 0.7537,
+      "step": 2981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152774803082994e-05,
+      "loss": 0.7807,
+      "step": 2982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41489921646376e-05,
+      "loss": 0.7959,
+      "step": 2983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145208808962192e-05,
+      "loss": 0.7821,
+      "step": 2984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141424736710796e-05,
+      "loss": 0.7233,
+      "step": 2985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137639948537554e-05,
+      "loss": 0.781,
+      "step": 2986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133854445096738e-05,
+      "loss": 0.6551,
+      "step": 2987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130068227042741e-05,
+      "loss": 0.7357,
+      "step": 2988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126281295030085e-05,
+      "loss": 0.7145,
+      "step": 2989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122493649713408e-05,
+      "loss": 0.6755,
+      "step": 2990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118705291747474e-05,
+      "loss": 0.7792,
+      "step": 2991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114916221787174e-05,
+      "loss": 0.6961,
+      "step": 2992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111126440487516e-05,
+      "loss": 0.8117,
+      "step": 2993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107335948503634e-05,
+      "loss": 0.776,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103544746490787e-05,
+      "loss": 0.7572,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409975283510435e-05,
+      "loss": 0.7367,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409596021499983e-05,
+      "loss": 0.7783,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092166886832842e-05,
+      "loss": 0.7724,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088372851259146e-05,
+      "loss": 0.8042,
+      "step": 2999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084578108934607e-05,
+      "loss": 0.2863,
+      "step": 3000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080782660515211e-05,
+      "loss": 0.8675,
+      "step": 3001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076986506657075e-05,
+      "loss": 0.7949,
+      "step": 3002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073189648016437e-05,
+      "loss": 0.7539,
+      "step": 3003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069392085249649e-05,
+      "loss": 0.7936,
+      "step": 3004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065593819013193e-05,
+      "loss": 0.7791,
+      "step": 3005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061794849963668e-05,
+      "loss": 0.736,
+      "step": 3006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057995178757799e-05,
+      "loss": 0.7919,
+      "step": 3007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054194806052428e-05,
+      "loss": 0.7955,
+      "step": 3008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050393732504517e-05,
+      "loss": 0.848,
+      "step": 3009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046591958771153e-05,
+      "loss": 0.7074,
+      "step": 3010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042789485509547e-05,
+      "loss": 0.69,
+      "step": 3011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038986313377022e-05,
+      "loss": 0.8373,
+      "step": 3012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403518244303103e-05,
+      "loss": 0.8075,
+      "step": 3013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031377875129138e-05,
+      "loss": 0.7445,
+      "step": 3014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027572610329037e-05,
+      "loss": 0.7211,
+      "step": 3015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023766649288537e-05,
+      "loss": 0.8304,
+      "step": 3016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019959992665569e-05,
+      "loss": 0.7675,
+      "step": 3017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016152641118184e-05,
+      "loss": 0.7976,
+      "step": 3018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401234459530455e-05,
+      "loss": 0.805,
+      "step": 3019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008535855882963e-05,
+      "loss": 0.7772,
+      "step": 3020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004726423511831e-05,
+      "loss": 0.7276,
+      "step": 3021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000916298849687e-05,
+      "loss": 0.7672,
+      "step": 3022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399710548255518e-05,
+      "loss": 0.7722,
+      "step": 3023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399329397528708e-05,
+      "loss": 0.7221,
+      "step": 3024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398948177770428e-05,
+      "loss": 0.7689,
+      "step": 3025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398566889046578e-05,
+      "loss": 0.8285,
+      "step": 3026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981855314230718e-05,
+      "loss": 0.7306,
+      "step": 3027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978041049658338e-05,
+      "loss": 0.8299,
+      "step": 3028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974226097408001e-05,
+      "loss": 0.7738,
+      "step": 3029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970410458139201e-05,
+      "loss": 0.8059,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966594132511535e-05,
+      "loss": 0.7862,
+      "step": 3031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962777121184726e-05,
+      "loss": 0.7419,
+      "step": 3032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958959424818615e-05,
+      "loss": 0.7313,
+      "step": 3033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955141044073166e-05,
+      "loss": 0.7325,
+      "step": 3034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951321979608449e-05,
+      "loss": 0.6951,
+      "step": 3035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947502232084671e-05,
+      "loss": 0.7825,
+      "step": 3036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943681802162137e-05,
+      "loss": 0.272,
+      "step": 3037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393986069050128e-05,
+      "loss": 0.7386,
+      "step": 3038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936038897762654e-05,
+      "loss": 0.7269,
+      "step": 3039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932216424606924e-05,
+      "loss": 0.7891,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392839327169487e-05,
+      "loss": 0.7847,
+      "step": 3041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924569439687404e-05,
+      "loss": 0.7866,
+      "step": 3042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392074492924554e-05,
+      "loss": 0.7379,
+      "step": 3043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916919741030417e-05,
+      "loss": 0.8189,
+      "step": 3044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913093875703288e-05,
+      "loss": 0.8017,
+      "step": 3045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909267333925528e-05,
+      "loss": 0.7886,
+      "step": 3046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905440116358624e-05,
+      "loss": 0.7802,
+      "step": 3047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390161222366418e-05,
+      "loss": 0.2723,
+      "step": 3048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389778365650392e-05,
+      "loss": 0.8274,
+      "step": 3049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893954415539683e-05,
+      "loss": 0.7571,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389012450143342e-05,
+      "loss": 0.7496,
+      "step": 3051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886293914847207e-05,
+      "loss": 0.7475,
+      "step": 3052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882462656443227e-05,
+      "loss": 0.7992,
+      "step": 3053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878630726883789e-05,
+      "loss": 0.7676,
+      "step": 3054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874798126831311e-05,
+      "loss": 0.7747,
+      "step": 3055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870964856948326e-05,
+      "loss": 0.8348,
+      "step": 3056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867130917897489e-05,
+      "loss": 0.7228,
+      "step": 3057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863296310341568e-05,
+      "loss": 0.7915,
+      "step": 3058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859461034943443e-05,
+      "loss": 0.7692,
+      "step": 3059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855625092366112e-05,
+      "loss": 0.7503,
+      "step": 3060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851788483272692e-05,
+      "loss": 0.8117,
+      "step": 3061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384795120832641e-05,
+      "loss": 0.7716,
+      "step": 3062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844113268190607e-05,
+      "loss": 0.7781,
+      "step": 3063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840274663528747e-05,
+      "loss": 0.728,
+      "step": 3064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836435395004402e-05,
+      "loss": 0.766,
+      "step": 3065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832595463281257e-05,
+      "loss": 0.7408,
+      "step": 3066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828754869023122e-05,
+      "loss": 0.7047,
+      "step": 3067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824913612893907e-05,
+      "loss": 0.787,
+      "step": 3068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382107169555765e-05,
+      "loss": 0.8103,
+      "step": 3069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817229117678494e-05,
+      "loss": 0.2874,
+      "step": 3070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813385879920704e-05,
+      "loss": 0.681,
+      "step": 3071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809541982948651e-05,
+      "loss": 0.7258,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805697427426822e-05,
+      "loss": 0.6961,
+      "step": 3073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801852214019826e-05,
+      "loss": 0.7438,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798006343392373e-05,
+      "loss": 0.8108,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794159816209293e-05,
+      "loss": 0.2985,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790312633135536e-05,
+      "loss": 0.7225,
+      "step": 3077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786464794836154e-05,
+      "loss": 0.7487,
+      "step": 3078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782616301976312e-05,
+      "loss": 0.7828,
+      "step": 3079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778767155221302e-05,
+      "loss": 0.7336,
+      "step": 3080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774917355236517e-05,
+      "loss": 0.7416,
+      "step": 3081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771066902687464e-05,
+      "loss": 0.7081,
+      "step": 3082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767215798239769e-05,
+      "loss": 0.7709,
+      "step": 3083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763364042559161e-05,
+      "loss": 0.8207,
+      "step": 3084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759511636311493e-05,
+      "loss": 0.753,
+      "step": 3085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755658580162723e-05,
+      "loss": 0.6703,
+      "step": 3086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375180487477892e-05,
+      "loss": 0.8588,
+      "step": 3087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374795052082627e-05,
+      "loss": 0.8002,
+      "step": 3088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374409551897107e-05,
+      "loss": 0.3004,
+      "step": 3089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374023986987973e-05,
+      "loss": 0.8606,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736383574218764e-05,
+      "loss": 0.7326,
+      "step": 3091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732526632654808e-05,
+      "loss": 0.7258,
+      "step": 3092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372866904585461e-05,
+      "loss": 0.8323,
+      "step": 3093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724810814485022e-05,
+      "loss": 0.6965,
+      "step": 3094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720951939213006e-05,
+      "loss": 0.769,
+      "step": 3095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717092420705645e-05,
+      "loss": 0.8152,
+      "step": 3096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713232259630127e-05,
+      "loss": 0.794,
+      "step": 3097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709371456653751e-05,
+      "loss": 0.825,
+      "step": 3098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705510012443932e-05,
+      "loss": 0.7951,
+      "step": 3099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370164792766819e-05,
+      "loss": 0.288,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3697785202994151e-05,
+      "loss": 0.7442,
+      "step": 3101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369392183908957e-05,
+      "loss": 0.7461,
+      "step": 3102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690057836622291e-05,
+      "loss": 0.7709,
+      "step": 3103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686193196260282e-05,
+      "loss": 0.844,
+      "step": 3104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682327918671623e-05,
+      "loss": 0.8305,
+      "step": 3105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678462004524493e-05,
+      "loss": 0.7531,
+      "step": 3106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674595454487186e-05,
+      "loss": 0.8418,
+      "step": 3107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670728269228109e-05,
+      "loss": 0.7673,
+      "step": 3108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666860449415775e-05,
+      "loss": 0.7444,
+      "step": 3109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662991995718808e-05,
+      "loss": 0.8291,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659122908805947e-05,
+      "loss": 0.7072,
+      "step": 3111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655253189346027e-05,
+      "loss": 0.7988,
+      "step": 3112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651382838008005e-05,
+      "loss": 0.7499,
+      "step": 3113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647511855460941e-05,
+      "loss": 0.7362,
+      "step": 3114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364364024237401e-05,
+      "loss": 0.7897,
+      "step": 3115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639767999416486e-05,
+      "loss": 0.7997,
+      "step": 3116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635895127257763e-05,
+      "loss": 0.7549,
+      "step": 3117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632021626567337e-05,
+      "loss": 0.7752,
+      "step": 3118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628147498014811e-05,
+      "loss": 0.7478,
+      "step": 3119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624272742269904e-05,
+      "loss": 0.788,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362039736000244e-05,
+      "loss": 0.7813,
+      "step": 3121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616521351882347e-05,
+      "loss": 0.7499,
+      "step": 3122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612644718579662e-05,
+      "loss": 0.7021,
+      "step": 3123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360876746076454e-05,
+      "loss": 0.8185,
+      "step": 3124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604889579107231e-05,
+      "loss": 0.8314,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36010110742781e-05,
+      "loss": 0.8184,
+      "step": 3126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597131946947621e-05,
+      "loss": 0.2781,
+      "step": 3127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593252197786374e-05,
+      "loss": 0.2746,
+      "step": 3128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589371827465039e-05,
+      "loss": 0.7545,
+      "step": 3129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585490836654412e-05,
+      "loss": 0.7844,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581609226025396e-05,
+      "loss": 0.8036,
+      "step": 3131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577726996248997e-05,
+      "loss": 0.8224,
+      "step": 3132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573844147996333e-05,
+      "loss": 0.8121,
+      "step": 3133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569960681938626e-05,
+      "loss": 0.7683,
+      "step": 3134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35660765987472e-05,
+      "loss": 0.7686,
+      "step": 3135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562191899093495e-05,
+      "loss": 0.758,
+      "step": 3136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558306583649055e-05,
+      "loss": 0.8107,
+      "step": 3137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554420653085523e-05,
+      "loss": 0.7963,
+      "step": 3138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550534108074658e-05,
+      "loss": 0.8241,
+      "step": 3139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354664694928832e-05,
+      "loss": 0.7956,
+      "step": 3140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542759177398477e-05,
+      "loss": 0.7297,
+      "step": 3141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538870793077199e-05,
+      "loss": 0.7152,
+      "step": 3142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353498179699667e-05,
+      "loss": 0.2997,
+      "step": 3143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531092189829172e-05,
+      "loss": 0.6999,
+      "step": 3144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527201972247095e-05,
+      "loss": 0.817,
+      "step": 3145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523311144922938e-05,
+      "loss": 0.7663,
+      "step": 3146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35194197085293e-05,
+      "loss": 0.7674,
+      "step": 3147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515527663738885e-05,
+      "loss": 0.7171,
+      "step": 3148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511635011224512e-05,
+      "loss": 0.7335,
+      "step": 3149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507741751659092e-05,
+      "loss": 0.7213,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350384788571565e-05,
+      "loss": 0.6946,
+      "step": 3151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499953414067312e-05,
+      "loss": 0.6862,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349605833738731e-05,
+      "loss": 0.7746,
+      "step": 3153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492162656348979e-05,
+      "loss": 0.7921,
+      "step": 3154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3488266371625758e-05,
+      "loss": 0.7812,
+      "step": 3155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484369483891196e-05,
+      "loss": 0.7342,
+      "step": 3156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480471993818937e-05,
+      "loss": 0.7047,
+      "step": 3157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476573902082737e-05,
+      "loss": 0.8073,
+      "step": 3158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472675209356457e-05,
+      "loss": 0.7228,
+      "step": 3159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468775916314053e-05,
+      "loss": 0.7288,
+      "step": 3160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464876023629585e-05,
+      "loss": 0.7543,
+      "step": 3161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460975531977234e-05,
+      "loss": 0.7967,
+      "step": 3162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3457074442031266e-05,
+      "loss": 0.7708,
+      "step": 3163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453172754466053e-05,
+      "loss": 0.7765,
+      "step": 3164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449270469956081e-05,
+      "loss": 0.8164,
+      "step": 3165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445367589175926e-05,
+      "loss": 0.7298,
+      "step": 3166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3441464112800275e-05,
+      "loss": 0.7491,
+      "step": 3167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437560041503919e-05,
+      "loss": 0.7564,
+      "step": 3168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433655375961747e-05,
+      "loss": 0.7907,
+      "step": 3169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429750116848747e-05,
+      "loss": 0.7918,
+      "step": 3170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425844264840025e-05,
+      "loss": 0.7807,
+      "step": 3171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421937820610771e-05,
+      "loss": 0.7345,
+      "step": 3172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418030784836293e-05,
+      "loss": 0.7171,
+      "step": 3173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414123158191989e-05,
+      "loss": 0.7257,
+      "step": 3174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341021494135337e-05,
+      "loss": 0.7885,
+      "step": 3175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340630613499604e-05,
+      "loss": 0.7855,
+      "step": 3176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340239673979571e-05,
+      "loss": 0.7532,
+      "step": 3177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398486756428187e-05,
+      "loss": 0.7273,
+      "step": 3178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339457618556939e-05,
+      "loss": 0.8006,
+      "step": 3179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390665027895329e-05,
+      "loss": 0.7167,
+      "step": 3180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386753284082121e-05,
+      "loss": 0.8126,
+      "step": 3181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382840954805987e-05,
+      "loss": 0.7884,
+      "step": 3182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337892804074324e-05,
+      "loss": 0.2568,
+      "step": 3183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33750145425703e-05,
+      "loss": 0.7595,
+      "step": 3184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371100460963693e-05,
+      "loss": 0.7607,
+      "step": 3185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367185796600033e-05,
+      "loss": 0.7828,
+      "step": 3186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363270550156049e-05,
+      "loss": 0.8256,
+      "step": 3187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359354722308562e-05,
+      "loss": 0.7469,
+      "step": 3188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335543831373449e-05,
+      "loss": 0.7364,
+      "step": 3189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351521325110864e-05,
+      "loss": 0.8026,
+      "step": 3190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347603757114802e-05,
+      "loss": 0.7856,
+      "step": 3191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334368561042353e-05,
+      "loss": 0.7085,
+      "step": 3192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339766885714376e-05,
+      "loss": 0.7559,
+      "step": 3193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335847583664758e-05,
+      "loss": 0.8358,
+      "step": 3194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33319277049522e-05,
+      "loss": 0.7582,
+      "step": 3195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328007250254332e-05,
+      "loss": 0.8181,
+      "step": 3196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324086220248873e-05,
+      "loss": 0.7122,
+      "step": 3197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320164615613644e-05,
+      "loss": 0.7812,
+      "step": 3198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316242437026569e-05,
+      "loss": 0.7547,
+      "step": 3199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331231968516567e-05,
+      "loss": 0.8042,
+      "step": 3200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308396360709064e-05,
+      "loss": 0.6818,
+      "step": 3201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304472464334972e-05,
+      "loss": 0.7909,
+      "step": 3202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300547996721714e-05,
+      "loss": 0.767,
+      "step": 3203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296622958547702e-05,
+      "loss": 0.7411,
+      "step": 3204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292697350491455e-05,
+      "loss": 0.7819,
+      "step": 3205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288771173231587e-05,
+      "loss": 0.7053,
+      "step": 3206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284844427446809e-05,
+      "loss": 0.7802,
+      "step": 3207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328091711381593e-05,
+      "loss": 0.7367,
+      "step": 3208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276989233017863e-05,
+      "loss": 0.7855,
+      "step": 3209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273060785731615e-05,
+      "loss": 0.7157,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269131772636283e-05,
+      "loss": 0.7958,
+      "step": 3211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326520219441108e-05,
+      "loss": 0.266,
+      "step": 3212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261272051735302e-05,
+      "loss": 0.8153,
+      "step": 3213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325734134528834e-05,
+      "loss": 0.8313,
+      "step": 3214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253410075749699e-05,
+      "loss": 0.2845,
+      "step": 3215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324947824379897e-05,
+      "loss": 0.7742,
+      "step": 3216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245545850115837e-05,
+      "loss": 0.6945,
+      "step": 3217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241612895380094e-05,
+      "loss": 0.7665,
+      "step": 3218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323767938027162e-05,
+      "loss": 0.7502,
+      "step": 3219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233745305470401e-05,
+      "loss": 0.7737,
+      "step": 3220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229810671656508e-05,
+      "loss": 0.7313,
+      "step": 3221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225875479510117e-05,
+      "loss": 0.7885,
+      "step": 3222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221939729711504e-05,
+      "loss": 0.7902,
+      "step": 3223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218003422941028e-05,
+      "loss": 0.8103,
+      "step": 3224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214066559879162e-05,
+      "loss": 0.7738,
+      "step": 3225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210129141206455e-05,
+      "loss": 0.7828,
+      "step": 3226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320619116760357e-05,
+      "loss": 0.7544,
+      "step": 3227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202252639751259e-05,
+      "loss": 0.7262,
+      "step": 3228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198313558330364e-05,
+      "loss": 0.7357,
+      "step": 3229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194373924021828e-05,
+      "loss": 0.7966,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190433737506698e-05,
+      "loss": 0.8394,
+      "step": 3231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.31864929994661e-05,
+      "loss": 0.799,
+      "step": 3232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182551710581266e-05,
+      "loss": 0.7506,
+      "step": 3233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178609871533522e-05,
+      "loss": 0.7922,
+      "step": 3234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174667483004286e-05,
+      "loss": 0.6802,
+      "step": 3235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170724545675074e-05,
+      "loss": 0.7229,
+      "step": 3236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166781060227494e-05,
+      "loss": 0.7375,
+      "step": 3237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316283702734325e-05,
+      "loss": 0.7842,
+      "step": 3238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158892447704142e-05,
+      "loss": 0.7792,
+      "step": 3239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154947321992067e-05,
+      "loss": 0.7716,
+      "step": 3240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151001650889008e-05,
+      "loss": 0.7839,
+      "step": 3241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147055435077047e-05,
+      "loss": 0.7975,
+      "step": 3242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143108675238363e-05,
+      "loss": 0.8234,
+      "step": 3243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139161372055227e-05,
+      "loss": 0.7844,
+      "step": 3244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135213526210003e-05,
+      "loss": 0.7413,
+      "step": 3245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131265138385145e-05,
+      "loss": 0.7244,
+      "step": 3246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312731620926321e-05,
+      "loss": 0.6886,
+      "step": 3247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123366739526842e-05,
+      "loss": 0.797,
+      "step": 3248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3119416729858779e-05,
+      "loss": 0.7892,
+      "step": 3249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115466180941854e-05,
+      "loss": 0.7807,
+      "step": 3250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311151509345899e-05,
+      "loss": 0.7626,
+      "step": 3251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107563468093209e-05,
+      "loss": 0.2899,
+      "step": 3252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310361130552762e-05,
+      "loss": 0.8379,
+      "step": 3253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099658606445428e-05,
+      "loss": 0.8734,
+      "step": 3254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095705371529933e-05,
+      "loss": 0.8176,
+      "step": 3255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309175160146452e-05,
+      "loss": 0.7445,
+      "step": 3256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087797296932674e-05,
+      "loss": 0.811,
+      "step": 3257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308384245861797e-05,
+      "loss": 0.7595,
+      "step": 3258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3079887087204072e-05,
+      "loss": 0.7513,
+      "step": 3259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075931183374743e-05,
+      "loss": 0.7249,
+      "step": 3260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071974747813834e-05,
+      "loss": 0.8423,
+      "step": 3261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068017781205284e-05,
+      "loss": 0.7695,
+      "step": 3262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064060284233132e-05,
+      "loss": 0.2423,
+      "step": 3263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060102257581499e-05,
+      "loss": 0.7353,
+      "step": 3264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056143701934609e-05,
+      "loss": 0.7765,
+      "step": 3265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052184617976771e-05,
+      "loss": 0.798,
+      "step": 3266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048225006392383e-05,
+      "loss": 0.7732,
+      "step": 3267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044264867865937e-05,
+      "loss": 0.8234,
+      "step": 3268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040304203082021e-05,
+      "loss": 0.7659,
+      "step": 3269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036343012725303e-05,
+      "loss": 0.7765,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303238129748055e-05,
+      "loss": 0.8021,
+      "step": 3271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028419058032622e-05,
+      "loss": 0.7666,
+      "step": 3272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024456295066458e-05,
+      "loss": 0.7541,
+      "step": 3273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30204930092671e-05,
+      "loss": 0.7474,
+      "step": 3274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016529201319675e-05,
+      "loss": 0.7266,
+      "step": 3275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012564871909398e-05,
+      "loss": 0.2889,
+      "step": 3276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008600021721578e-05,
+      "loss": 0.7445,
+      "step": 3277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300463465144161e-05,
+      "loss": 0.7745,
+      "step": 3278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000668761754988e-05,
+      "loss": 0.7796,
+      "step": 3279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996702353347286e-05,
+      "loss": 0.7631,
+      "step": 3280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992735426904168e-05,
+      "loss": 0.6772,
+      "step": 3281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988767983111394e-05,
+      "loss": 0.7137,
+      "step": 3282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298480002265481e-05,
+      "loss": 0.7818,
+      "step": 3283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980831546220349e-05,
+      "loss": 0.7465,
+      "step": 3284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976862554494041e-05,
+      "loss": 0.8137,
+      "step": 3285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972893048161992e-05,
+      "loss": 0.7066,
+      "step": 3286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296892302791041e-05,
+      "loss": 0.8013,
+      "step": 3287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296495249442558e-05,
+      "loss": 0.8336,
+      "step": 3288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960981448393892e-05,
+      "loss": 0.7855,
+      "step": 3289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957009890501807e-05,
+      "loss": 0.7525,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953037821435888e-05,
+      "loss": 0.7491,
+      "step": 3291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949065241882774e-05,
+      "loss": 0.8129,
+      "step": 3292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945092152529204e-05,
+      "loss": 0.7431,
+      "step": 3293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941118554061998e-05,
+      "loss": 0.8053,
+      "step": 3294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937144447168067e-05,
+      "loss": 0.8178,
+      "step": 3295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933169832534405e-05,
+      "loss": 0.7584,
+      "step": 3296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929194710848103e-05,
+      "loss": 0.8032,
+      "step": 3297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925219082796332e-05,
+      "loss": 0.7862,
+      "step": 3298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292124294906635e-05,
+      "loss": 0.7683,
+      "step": 3299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917266310345509e-05,
+      "loss": 0.8249,
+      "step": 3300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913289167321246e-05,
+      "loss": 0.7666,
+      "step": 3301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290931152068108e-05,
+      "loss": 0.7033,
+      "step": 3302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290533337111262e-05,
+      "loss": 0.7729,
+      "step": 3303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901354719303568e-05,
+      "loss": 0.7499,
+      "step": 3304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.28973755659417e-05,
+      "loss": 0.7794,
+      "step": 3305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893395911714893e-05,
+      "loss": 0.7428,
+      "step": 3306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889415757311103e-05,
+      "loss": 0.827,
+      "step": 3307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885435103418372e-05,
+      "loss": 0.3112,
+      "step": 3308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881453950724825e-05,
+      "loss": 0.2712,
+      "step": 3309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877472299918683e-05,
+      "loss": 0.7987,
+      "step": 3310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873490151688248e-05,
+      "loss": 0.7021,
+      "step": 3311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869507506721903e-05,
+      "loss": 0.8015,
+      "step": 3312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865524365708133e-05,
+      "loss": 0.7226,
+      "step": 3313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861540729335486e-05,
+      "loss": 0.8295,
+      "step": 3314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857556598292613e-05,
+      "loss": 0.7984,
+      "step": 3315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853571973268243e-05,
+      "loss": 0.7938,
+      "step": 3316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849586854951192e-05,
+      "loss": 0.7727,
+      "step": 3317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845601244030364e-05,
+      "loss": 0.693,
+      "step": 3318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841615141194741e-05,
+      "loss": 0.7414,
+      "step": 3319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837628547133398e-05,
+      "loss": 0.8274,
+      "step": 3320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283364146253549e-05,
+      "loss": 0.7494,
+      "step": 3321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829653888090259e-05,
+      "loss": 0.8235,
+      "step": 3322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825665824487028e-05,
+      "loss": 0.8136,
+      "step": 3323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821677272415213e-05,
+      "loss": 0.8118,
+      "step": 3324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817688232564305e-05,
+      "loss": 0.7817,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813698705623887e-05,
+      "loss": 0.7403,
+      "step": 3326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809708692283618e-05,
+      "loss": 0.6903,
+      "step": 3327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805718193233248e-05,
+      "loss": 0.7739,
+      "step": 3328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801727209162608e-05,
+      "loss": 0.6802,
+      "step": 3329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797735740761614e-05,
+      "loss": 0.8082,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793743788720263e-05,
+      "loss": 0.6963,
+      "step": 3331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789751353728641e-05,
+      "loss": 0.7021,
+      "step": 3332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785758436476913e-05,
+      "loss": 0.7388,
+      "step": 3333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781765037655327e-05,
+      "loss": 0.7631,
+      "step": 3334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777771157954217e-05,
+      "loss": 0.8023,
+      "step": 3335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773776798064e-05,
+      "loss": 0.6918,
+      "step": 3336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2769781958675172e-05,
+      "loss": 0.7444,
+      "step": 3337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765786640478319e-05,
+      "loss": 0.8142,
+      "step": 3338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761790844164104e-05,
+      "loss": 0.82,
+      "step": 3339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275779457042327e-05,
+      "loss": 0.8073,
+      "step": 3340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753797819946655e-05,
+      "loss": 0.7981,
+      "step": 3341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749800593425163e-05,
+      "loss": 0.7678,
+      "step": 3342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745802891549795e-05,
+      "loss": 0.811,
+      "step": 3343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741804715011624e-05,
+      "loss": 0.7613,
+      "step": 3344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737806064501808e-05,
+      "loss": 0.8081,
+      "step": 3345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273380694071159e-05,
+      "loss": 0.7312,
+      "step": 3346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729807344332292e-05,
+      "loss": 0.7596,
+      "step": 3347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272580727605532e-05,
+      "loss": 0.7074,
+      "step": 3348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721806736572155e-05,
+      "loss": 0.729,
+      "step": 3349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717805726574368e-05,
+      "loss": 0.8041,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713804246753609e-05,
+      "loss": 0.7584,
+      "step": 3351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709802297801604e-05,
+      "loss": 0.298,
+      "step": 3352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705799880410164e-05,
+      "loss": 0.2737,
+      "step": 3353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701796995271182e-05,
+      "loss": 0.8351,
+      "step": 3354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697793643076632e-05,
+      "loss": 0.7804,
+      "step": 3355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693789824518566e-05,
+      "loss": 0.7841,
+      "step": 3356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689785540289119e-05,
+      "loss": 0.7673,
+      "step": 3357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685780791080502e-05,
+      "loss": 0.7742,
+      "step": 3358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681775577585014e-05,
+      "loss": 0.8216,
+      "step": 3359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267776990049503e-05,
+      "loss": 0.8556,
+      "step": 3360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673763760503003e-05,
+      "loss": 0.7748,
+      "step": 3361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669757158301472e-05,
+      "loss": 0.7663,
+      "step": 3362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266575009458305e-05,
+      "loss": 0.7307,
+      "step": 3363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661742570040432e-05,
+      "loss": 0.7154,
+      "step": 3364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657734585366393e-05,
+      "loss": 0.8053,
+      "step": 3365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653726141253786e-05,
+      "loss": 0.7268,
+      "step": 3366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649717238395548e-05,
+      "loss": 0.8253,
+      "step": 3367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645707877484685e-05,
+      "loss": 0.8421,
+      "step": 3368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641698059214295e-05,
+      "loss": 0.7587,
+      "step": 3369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637687784277552e-05,
+      "loss": 0.7804,
+      "step": 3370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633677053367702e-05,
+      "loss": 0.7427,
+      "step": 3371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629665867178072e-05,
+      "loss": 0.7206,
+      "step": 3372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625654226402074e-05,
+      "loss": 0.8126,
+      "step": 3373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262164213173319e-05,
+      "loss": 0.747,
+      "step": 3374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617629583864988e-05,
+      "loss": 0.7788,
+      "step": 3375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261361658349111e-05,
+      "loss": 0.7515,
+      "step": 3376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609603131305279e-05,
+      "loss": 0.7516,
+      "step": 3377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605589228001291e-05,
+      "loss": 0.7764,
+      "step": 3378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601574874273027e-05,
+      "loss": 0.8131,
+      "step": 3379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597560070814443e-05,
+      "loss": 0.7858,
+      "step": 3380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593544818319563e-05,
+      "loss": 0.7759,
+      "step": 3381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589529117482506e-05,
+      "loss": 0.7837,
+      "step": 3382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585512968997461e-05,
+      "loss": 0.77,
+      "step": 3383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581496373558693e-05,
+      "loss": 0.7797,
+      "step": 3384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577479331860538e-05,
+      "loss": 0.6956,
+      "step": 3385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573461844597423e-05,
+      "loss": 0.7385,
+      "step": 3386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569443912463843e-05,
+      "loss": 0.7316,
+      "step": 3387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565425536154371e-05,
+      "loss": 0.7202,
+      "step": 3388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561406716363657e-05,
+      "loss": 0.8282,
+      "step": 3389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255738745378643e-05,
+      "loss": 0.7188,
+      "step": 3390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553367749117496e-05,
+      "loss": 0.7639,
+      "step": 3391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549347603051728e-05,
+      "loss": 0.7889,
+      "step": 3392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254532701628409e-05,
+      "loss": 0.7316,
+      "step": 3393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541305989509612e-05,
+      "loss": 0.7604,
+      "step": 3394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25372845234234e-05,
+      "loss": 0.795,
+      "step": 3395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533262618720646e-05,
+      "loss": 0.7872,
+      "step": 3396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529240276096604e-05,
+      "loss": 0.7449,
+      "step": 3397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525217496246614e-05,
+      "loss": 0.8161,
+      "step": 3398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521194279866084e-05,
+      "loss": 0.8476,
+      "step": 3399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2517170627650503e-05,
+      "loss": 0.7866,
+      "step": 3400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513146540295435e-05,
+      "loss": 0.7649,
+      "step": 3401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250912201849652e-05,
+      "loss": 0.7219,
+      "step": 3402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505097062949464e-05,
+      "loss": 0.7743,
+      "step": 3403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250107167435006e-05,
+      "loss": 0.7912,
+      "step": 3404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497045853394174e-05,
+      "loss": 0.789,
+      "step": 3405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493019600777738e-05,
+      "loss": 0.7862,
+      "step": 3406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488992917196762e-05,
+      "loss": 0.8012,
+      "step": 3407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484965803347342e-05,
+      "loss": 0.7095,
+      "step": 3408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480938259925633e-05,
+      "loss": 0.7227,
+      "step": 3409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476910287627869e-05,
+      "loss": 0.7874,
+      "step": 3410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2472881887150362e-05,
+      "loss": 0.7218,
+      "step": 3411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468853059189497e-05,
+      "loss": 0.7681,
+      "step": 3412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464823804441726e-05,
+      "loss": 0.7729,
+      "step": 3413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460794123603586e-05,
+      "loss": 0.7574,
+      "step": 3414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245676401737168e-05,
+      "loss": 0.7644,
+      "step": 3415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452733486442682e-05,
+      "loss": 0.7602,
+      "step": 3416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448702531513353e-05,
+      "loss": 0.3032,
+      "step": 3417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244467115328051e-05,
+      "loss": 0.7814,
+      "step": 3418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440639352441052e-05,
+      "loss": 0.7861,
+      "step": 3419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436607129691957e-05,
+      "loss": 0.7008,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432574485730262e-05,
+      "loss": 0.8293,
+      "step": 3421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428541421253088e-05,
+      "loss": 0.7056,
+      "step": 3422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424507936957624e-05,
+      "loss": 0.7482,
+      "step": 3423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242047403354113e-05,
+      "loss": 0.7705,
+      "step": 3424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416439711700943e-05,
+      "loss": 0.6853,
+      "step": 3425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412404972134471e-05,
+      "loss": 0.8023,
+      "step": 3426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240836981553919e-05,
+      "loss": 0.7935,
+      "step": 3427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404334242612658e-05,
+      "loss": 0.714,
+      "step": 3428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400298254052488e-05,
+      "loss": 0.7847,
+      "step": 3429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396261850556384e-05,
+      "loss": 0.8159,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392225032822111e-05,
+      "loss": 0.7397,
+      "step": 3431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388187801547504e-05,
+      "loss": 0.755,
+      "step": 3432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238415015743048e-05,
+      "loss": 0.7364,
+      "step": 3433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380112101169015e-05,
+      "loss": 0.6996,
+      "step": 3434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376073633461162e-05,
+      "loss": 0.7885,
+      "step": 3435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372034755005044e-05,
+      "loss": 0.6703,
+      "step": 3436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367995466498861e-05,
+      "loss": 0.7366,
+      "step": 3437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2363955768640872e-05,
+      "loss": 0.7736,
+      "step": 3438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359915662129418e-05,
+      "loss": 0.8095,
+      "step": 3439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355875147662906e-05,
+      "loss": 0.7846,
+      "step": 3440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351834225939812e-05,
+      "loss": 0.7933,
+      "step": 3441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347792897658686e-05,
+      "loss": 0.7476,
+      "step": 3442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343751163518143e-05,
+      "loss": 0.7836,
+      "step": 3443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339709024216873e-05,
+      "loss": 0.7807,
+      "step": 3444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335666480453639e-05,
+      "loss": 0.7832,
+      "step": 3445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331623532927262e-05,
+      "loss": 0.738,
+      "step": 3446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327580182336643e-05,
+      "loss": 0.8259,
+      "step": 3447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323536429380752e-05,
+      "loss": 0.7509,
+      "step": 3448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2319492274758624e-05,
+      "loss": 0.8166,
+      "step": 3449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315447719169367e-05,
+      "loss": 0.8329,
+      "step": 3450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311402763312157e-05,
+      "loss": 0.8894,
+      "step": 3451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230735740788624e-05,
+      "loss": 0.6981,
+      "step": 3452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303311653590932e-05,
+      "loss": 0.7362,
+      "step": 3453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229926550112561e-05,
+      "loss": 0.8584,
+      "step": 3454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295218951189734e-05,
+      "loss": 0.8548,
+      "step": 3455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291172004482824e-05,
+      "loss": 0.6689,
+      "step": 3456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287124661704461e-05,
+      "loss": 0.6991,
+      "step": 3457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283076923554314e-05,
+      "loss": 0.7591,
+      "step": 3458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279028790732104e-05,
+      "loss": 0.8001,
+      "step": 3459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274980263937623e-05,
+      "loss": 0.7417,
+      "step": 3460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270931343870743e-05,
+      "loss": 0.7773,
+      "step": 3461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266882031231382e-05,
+      "loss": 0.705,
+      "step": 3462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262832326719547e-05,
+      "loss": 0.757,
+      "step": 3463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258782231035307e-05,
+      "loss": 0.7412,
+      "step": 3464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225473174487879e-05,
+      "loss": 0.8308,
+      "step": 3465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2250680868950196e-05,
+      "loss": 0.7585,
+      "step": 3466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246629603949797e-05,
+      "loss": 0.7973,
+      "step": 3467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2242577950577932e-05,
+      "loss": 0.7204,
+      "step": 3468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238525909535e-05,
+      "loss": 0.742,
+      "step": 3469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234473481521472e-05,
+      "loss": 0.7694,
+      "step": 3470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230420667237885e-05,
+      "loss": 0.7572,
+      "step": 3471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226367467384844e-05,
+      "loss": 0.6917,
+      "step": 3472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222313882663018e-05,
+      "loss": 0.8033,
+      "step": 3473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218259913773142e-05,
+      "loss": 0.7383,
+      "step": 3474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214205561416029e-05,
+      "loss": 0.7797,
+      "step": 3475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210150826292534e-05,
+      "loss": 0.7867,
+      "step": 3476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220609570910361e-05,
+      "loss": 0.6342,
+      "step": 3477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202040210550243e-05,
+      "loss": 0.8301,
+      "step": 3478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219798433133351e-05,
+      "loss": 0.74,
+      "step": 3479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193928072154546e-05,
+      "loss": 0.722,
+      "step": 3480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189871433714544e-05,
+      "loss": 0.7458,
+      "step": 3481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185814416714771e-05,
+      "loss": 0.7725,
+      "step": 3482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181757021856562e-05,
+      "loss": 0.7701,
+      "step": 3483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177699249841308e-05,
+      "loss": 0.7957,
+      "step": 3484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173641101370466e-05,
+      "loss": 0.7364,
+      "step": 3485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169582577145574e-05,
+      "loss": 0.7712,
+      "step": 3486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165523677868215e-05,
+      "loss": 0.7219,
+      "step": 3487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161464404240042e-05,
+      "loss": 0.7482,
+      "step": 3488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157404756962783e-05,
+      "loss": 0.8161,
+      "step": 3489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153344736738217e-05,
+      "loss": 0.7728,
+      "step": 3490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149284344268197e-05,
+      "loss": 0.8,
+      "step": 3491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145223580254637e-05,
+      "loss": 0.7731,
+      "step": 3492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141162445399515e-05,
+      "loss": 0.757,
+      "step": 3493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213710094040487e-05,
+      "loss": 0.7824,
+      "step": 3494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133039065972812e-05,
+      "loss": 0.2533,
+      "step": 3495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128976822805509e-05,
+      "loss": 0.7269,
+      "step": 3496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124914211605195e-05,
+      "loss": 0.7552,
+      "step": 3497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120851233074167e-05,
+      "loss": 0.3028,
+      "step": 3498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211678788791479e-05,
+      "loss": 0.7493,
+      "step": 3499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112724176829484e-05,
+      "loss": 0.7644,
+      "step": 3500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108660100520738e-05,
+      "loss": 0.7818,
+      "step": 3501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104595659691103e-05,
+      "loss": 0.7679,
+      "step": 3502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100530855043193e-05,
+      "loss": 0.7874,
+      "step": 3503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096465687279682e-05,
+      "loss": 0.7629,
+      "step": 3504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092400157103315e-05,
+      "loss": 0.7834,
+      "step": 3505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088334265216887e-05,
+      "loss": 0.7713,
+      "step": 3506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084268012323265e-05,
+      "loss": 0.8276,
+      "step": 3507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208020139912538e-05,
+      "loss": 0.766,
+      "step": 3508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2076134426326218e-05,
+      "loss": 0.3108,
+      "step": 3509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072067094628832e-05,
+      "loss": 0.7969,
+      "step": 3510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067999404736333e-05,
+      "loss": 0.7956,
+      "step": 3511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063931357351899e-05,
+      "loss": 0.7503,
+      "step": 3512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059862953178765e-05,
+      "loss": 0.7928,
+      "step": 3513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055794192920232e-05,
+      "loss": 0.7057,
+      "step": 3514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051725077279658e-05,
+      "loss": 0.7146,
+      "step": 3515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047655606960466e-05,
+      "loss": 0.3075,
+      "step": 3516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043585782666141e-05,
+      "loss": 0.7229,
+      "step": 3517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039515605100225e-05,
+      "loss": 0.9151,
+      "step": 3518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035445074966324e-05,
+      "loss": 0.7253,
+      "step": 3519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031374192968102e-05,
+      "loss": 0.7945,
+      "step": 3520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027302959809292e-05,
+      "loss": 0.7527,
+      "step": 3521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202323137619368e-05,
+      "loss": 0.7065,
+      "step": 3522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019159442825111e-05,
+      "loss": 0.2846,
+      "step": 3523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015087160407497e-05,
+      "loss": 0.706,
+      "step": 3524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011014529644807e-05,
+      "loss": 0.7648,
+      "step": 3525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006941551241069e-05,
+      "loss": 0.7897,
+      "step": 3526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002868225900374e-05,
+      "loss": 0.7146,
+      "step": 3527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998794554326873e-05,
+      "loss": 0.7735,
+      "step": 3528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994720537224774e-05,
+      "loss": 0.8039,
+      "step": 3529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990646175298343e-05,
+      "loss": 0.2711,
+      "step": 3530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986571469251915e-05,
+      "loss": 0.7498,
+      "step": 3531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982496419789872e-05,
+      "loss": 0.7548,
+      "step": 3532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978421027616667e-05,
+      "loss": 0.8533,
+      "step": 3533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974345293436804e-05,
+      "loss": 0.2474,
+      "step": 3534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970269217954853e-05,
+      "loss": 0.7372,
+      "step": 3535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966192801875431e-05,
+      "loss": 0.7267,
+      "step": 3536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196211604590323e-05,
+      "loss": 0.772,
+      "step": 3537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958038950742987e-05,
+      "loss": 0.8026,
+      "step": 3538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953961517099505e-05,
+      "loss": 0.833,
+      "step": 3539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949883745677646e-05,
+      "loss": 0.7492,
+      "step": 3540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945805637182326e-05,
+      "loss": 0.7317,
+      "step": 3541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941727192318522e-05,
+      "loss": 0.759,
+      "step": 3542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937648411791268e-05,
+      "loss": 0.7288,
+      "step": 3543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933569296305658e-05,
+      "loss": 0.7433,
+      "step": 3544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192948984656684e-05,
+      "loss": 0.6492,
+      "step": 3545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925410063280026e-05,
+      "loss": 0.794,
+      "step": 3546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921329947150481e-05,
+      "loss": 0.6993,
+      "step": 3547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917249498883525e-05,
+      "loss": 0.7219,
+      "step": 3548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913168719184544e-05,
+      "loss": 0.7768,
+      "step": 3549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909087608758973e-05,
+      "loss": 0.787,
+      "step": 3550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905006168312308e-05,
+      "loss": 0.7733,
+      "step": 3551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19009243985501e-05,
+      "loss": 0.7694,
+      "step": 3552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896842300177962e-05,
+      "loss": 0.7799,
+      "step": 3553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892759873901556e-05,
+      "loss": 0.7733,
+      "step": 3554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888677120426608e-05,
+      "loss": 0.7141,
+      "step": 3555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884594040458896e-05,
+      "loss": 0.7493,
+      "step": 3556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188051063470426e-05,
+      "loss": 0.759,
+      "step": 3557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876426903868583e-05,
+      "loss": 0.7778,
+      "step": 3558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872342848657822e-05,
+      "loss": 0.8192,
+      "step": 3559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186825846977798e-05,
+      "loss": 0.7369,
+      "step": 3560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864173767935115e-05,
+      "loss": 0.6904,
+      "step": 3561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860088743835343e-05,
+      "loss": 0.8127,
+      "step": 3562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856003398184836e-05,
+      "loss": 0.6858,
+      "step": 3563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851917731689824e-05,
+      "loss": 0.7609,
+      "step": 3564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847831745056587e-05,
+      "loss": 0.7411,
+      "step": 3565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843745438991465e-05,
+      "loss": 0.7626,
+      "step": 3566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839658814200852e-05,
+      "loss": 0.8589,
+      "step": 3567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835571871391195e-05,
+      "loss": 0.8002,
+      "step": 3568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831484611269002e-05,
+      "loss": 0.7314,
+      "step": 3569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827397034540826e-05,
+      "loss": 0.7506,
+      "step": 3570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823309141913284e-05,
+      "loss": 0.7302,
+      "step": 3571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819220934093042e-05,
+      "loss": 0.7177,
+      "step": 3572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815132411786818e-05,
+      "loss": 0.8431,
+      "step": 3573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811043575701398e-05,
+      "loss": 0.7321,
+      "step": 3574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806954426543608e-05,
+      "loss": 0.8045,
+      "step": 3575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180286496502033e-05,
+      "loss": 0.8233,
+      "step": 3576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179877519183851e-05,
+      "loss": 0.7815,
+      "step": 3577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794685107705135e-05,
+      "loss": 0.7563,
+      "step": 3578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790594713327255e-05,
+      "loss": 0.7231,
+      "step": 3579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786504009411968e-05,
+      "loss": 0.7699,
+      "step": 3580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178241299666643e-05,
+      "loss": 0.849,
+      "step": 3581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778321675797849e-05,
+      "loss": 0.7166,
+      "step": 3582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177423004751348e-05,
+      "loss": 0.7359,
+      "step": 3583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770138112520641e-05,
+      "loss": 0.8402,
+      "step": 3584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766045871526701e-05,
+      "loss": 0.8118,
+      "step": 3585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761953325239073e-05,
+      "loss": 0.7772,
+      "step": 3586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757860474365232e-05,
+      "loss": 0.7618,
+      "step": 3587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753767319612708e-05,
+      "loss": 0.7782,
+      "step": 3588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174967386168907e-05,
+      "loss": 0.8109,
+      "step": 3589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745580101301952e-05,
+      "loss": 0.7563,
+      "step": 3590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741486039159037e-05,
+      "loss": 0.7649,
+      "step": 3591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173739167596806e-05,
+      "loss": 0.7965,
+      "step": 3592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733297012436807e-05,
+      "loss": 0.7633,
+      "step": 3593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729202049273113e-05,
+      "loss": 0.7291,
+      "step": 3594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172510678718487e-05,
+      "loss": 0.8229,
+      "step": 3595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1721011226880021e-05,
+      "loss": 0.8349,
+      "step": 3596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716915369066559e-05,
+      "loss": 0.7972,
+      "step": 3597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712819214452526e-05,
+      "loss": 0.7861,
+      "step": 3598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708722763746022e-05,
+      "loss": 0.7859,
+      "step": 3599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704626017655195e-05,
+      "loss": 0.7532,
+      "step": 3600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700528976888237e-05,
+      "loss": 0.7721,
+      "step": 3601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16964316421534e-05,
+      "loss": 0.8308,
+      "step": 3602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692334014158989e-05,
+      "loss": 0.8411,
+      "step": 3603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168823609361335e-05,
+      "loss": 0.7664,
+      "step": 3604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684137881224884e-05,
+      "loss": 0.7745,
+      "step": 3605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680039377702045e-05,
+      "loss": 0.685,
+      "step": 3606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675940583753336e-05,
+      "loss": 0.2369,
+      "step": 3607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671841500087303e-05,
+      "loss": 0.2926,
+      "step": 3608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667742127412556e-05,
+      "loss": 0.784,
+      "step": 3609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663642466437745e-05,
+      "loss": 0.8109,
+      "step": 3610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659542517871569e-05,
+      "loss": 0.7063,
+      "step": 3611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655442282422784e-05,
+      "loss": 0.7404,
+      "step": 3612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651341760800187e-05,
+      "loss": 0.7434,
+      "step": 3613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647240953712636e-05,
+      "loss": 0.7658,
+      "step": 3614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643139861869024e-05,
+      "loss": 0.7201,
+      "step": 3615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639038485978307e-05,
+      "loss": 0.8527,
+      "step": 3616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634936826749479e-05,
+      "loss": 0.7328,
+      "step": 3617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630834884891588e-05,
+      "loss": 0.7473,
+      "step": 3618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626732661113733e-05,
+      "loss": 0.7205,
+      "step": 3619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622630156125059e-05,
+      "loss": 0.7227,
+      "step": 3620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618527370634756e-05,
+      "loss": 0.8131,
+      "step": 3621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614424305352071e-05,
+      "loss": 0.7693,
+      "step": 3622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610320960986291e-05,
+      "loss": 0.7917,
+      "step": 3623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606217338246757e-05,
+      "loss": 0.7635,
+      "step": 3624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602113437842856e-05,
+      "loss": 0.7402,
+      "step": 3625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598009260484025e-05,
+      "loss": 0.7763,
+      "step": 3626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593904806879742e-05,
+      "loss": 0.7712,
+      "step": 3627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158980007773954e-05,
+      "loss": 0.7413,
+      "step": 3628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1585695073773002e-05,
+      "loss": 0.7141,
+      "step": 3629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581589795689745e-05,
+      "loss": 0.7099,
+      "step": 3630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577484244199448e-05,
+      "loss": 0.8208,
+      "step": 3631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573378420011834e-05,
+      "loss": 0.7156,
+      "step": 3632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569272323836662e-05,
+      "loss": 0.7792,
+      "step": 3633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565165956383755e-05,
+      "loss": 0.7327,
+      "step": 3634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156105931836297e-05,
+      "loss": 0.7072,
+      "step": 3635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556952410484214e-05,
+      "loss": 0.6986,
+      "step": 3636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552845233457446e-05,
+      "loss": 0.778,
+      "step": 3637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548737787992668e-05,
+      "loss": 0.7969,
+      "step": 3638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544630074799924e-05,
+      "loss": 0.7543,
+      "step": 3639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154052209458931e-05,
+      "loss": 0.8053,
+      "step": 3640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536413848070968e-05,
+      "loss": 0.7622,
+      "step": 3641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532305335955083e-05,
+      "loss": 0.8001,
+      "step": 3642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528196558951883e-05,
+      "loss": 0.7314,
+      "step": 3643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524087517771654e-05,
+      "loss": 0.6883,
+      "step": 3644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519978213124714e-05,
+      "loss": 0.7571,
+      "step": 3645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515868645721436e-05,
+      "loss": 0.713,
+      "step": 3646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151175881627223e-05,
+      "loss": 0.7669,
+      "step": 3647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150764872548756e-05,
+      "loss": 0.8208,
+      "step": 3648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150353837407793e-05,
+      "loss": 0.7988,
+      "step": 3649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499427762753892e-05,
+      "loss": 0.7568,
+      "step": 3650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149531689222604e-05,
+      "loss": 0.7528,
+      "step": 3651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491205763205008e-05,
+      "loss": 0.7134,
+      "step": 3652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487094376401493e-05,
+      "loss": 0.7162,
+      "step": 3653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482982732526216e-05,
+      "loss": 0.7465,
+      "step": 3654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478870832289948e-05,
+      "loss": 0.7893,
+      "step": 3655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474758676403514e-05,
+      "loss": 0.6997,
+      "step": 3656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470646265577774e-05,
+      "loss": 0.7376,
+      "step": 3657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146653360052363e-05,
+      "loss": 0.727,
+      "step": 3658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462420681952038e-05,
+      "loss": 0.8153,
+      "step": 3659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458307510573989e-05,
+      "loss": 0.6689,
+      "step": 3660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454194087100525e-05,
+      "loss": 0.7773,
+      "step": 3661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1450080412242719e-05,
+      "loss": 0.785,
+      "step": 3662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445966486711703e-05,
+      "loss": 0.7377,
+      "step": 3663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441852311218641e-05,
+      "loss": 0.7908,
+      "step": 3664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437737886474747e-05,
+      "loss": 0.812,
+      "step": 3665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433623213191276e-05,
+      "loss": 0.7775,
+      "step": 3666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429508292079522e-05,
+      "loss": 0.8517,
+      "step": 3667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425393123850825e-05,
+      "loss": 0.7894,
+      "step": 3668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142127770921657e-05,
+      "loss": 0.7557,
+      "step": 3669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417162048888186e-05,
+      "loss": 0.8131,
+      "step": 3670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413046143577133e-05,
+      "loss": 0.7838,
+      "step": 3671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408929993994928e-05,
+      "loss": 0.7794,
+      "step": 3672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404813600853121e-05,
+      "loss": 0.8152,
+      "step": 3673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400696964863307e-05,
+      "loss": 0.7715,
+      "step": 3674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396580086737123e-05,
+      "loss": 0.8261,
+      "step": 3675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392462967186247e-05,
+      "loss": 0.7385,
+      "step": 3676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388345606922397e-05,
+      "loss": 0.7332,
+      "step": 3677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138422800665734e-05,
+      "loss": 0.7161,
+      "step": 3678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380110167102875e-05,
+      "loss": 0.8145,
+      "step": 3679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375992088970849e-05,
+      "loss": 0.8466,
+      "step": 3680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371873772973148e-05,
+      "loss": 0.7158,
+      "step": 3681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367755219821694e-05,
+      "loss": 0.7788,
+      "step": 3682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363636430228464e-05,
+      "loss": 0.7786,
+      "step": 3683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359517404905457e-05,
+      "loss": 0.8401,
+      "step": 3684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355398144564734e-05,
+      "loss": 0.7346,
+      "step": 3685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351278649918376e-05,
+      "loss": 0.7739,
+      "step": 3686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347158921678516e-05,
+      "loss": 0.7726,
+      "step": 3687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134303896055733e-05,
+      "loss": 0.7631,
+      "step": 3688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1338918767267025e-05,
+      "loss": 0.8055,
+      "step": 3689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334798342519847e-05,
+      "loss": 0.7327,
+      "step": 3690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330677687028102e-05,
+      "loss": 0.7843,
+      "step": 3691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326556801504112e-05,
+      "loss": 0.696,
+      "step": 3692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322435686660245e-05,
+      "loss": 0.7444,
+      "step": 3693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318314343208919e-05,
+      "loss": 0.7999,
+      "step": 3694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314192771862582e-05,
+      "loss": 0.7525,
+      "step": 3695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310070973333725e-05,
+      "loss": 0.7084,
+      "step": 3696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305948948334875e-05,
+      "loss": 0.7571,
+      "step": 3697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301826697578602e-05,
+      "loss": 0.7742,
+      "step": 3698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129770422177751e-05,
+      "loss": 0.7412,
+      "step": 3699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129358152164425e-05,
+      "loss": 0.8208,
+      "step": 3700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289458597891505e-05,
+      "loss": 0.7999,
+      "step": 3701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285335451231996e-05,
+      "loss": 0.763,
+      "step": 3702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128121208237849e-05,
+      "loss": 0.7201,
+      "step": 3703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277088492043783e-05,
+      "loss": 0.7345,
+      "step": 3704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272964680940715e-05,
+      "loss": 0.8185,
+      "step": 3705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268840649782163e-05,
+      "loss": 0.6924,
+      "step": 3706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264716399281046e-05,
+      "loss": 0.7317,
+      "step": 3707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260591930150312e-05,
+      "loss": 0.6964,
+      "step": 3708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256467243102951e-05,
+      "loss": 0.7017,
+      "step": 3709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252342338851997e-05,
+      "loss": 0.3068,
+      "step": 3710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248217218110513e-05,
+      "loss": 0.7567,
+      "step": 3711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244091881591602e-05,
+      "loss": 0.7511,
+      "step": 3712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239966330008405e-05,
+      "loss": 0.7403,
+      "step": 3713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235840564074098e-05,
+      "loss": 0.7491,
+      "step": 3714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231714584501898e-05,
+      "loss": 0.7144,
+      "step": 3715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122758839200506e-05,
+      "loss": 0.713,
+      "step": 3716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223461987296864e-05,
+      "loss": 0.7325,
+      "step": 3717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219335371090644e-05,
+      "loss": 0.7292,
+      "step": 3718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215208544099758e-05,
+      "loss": 0.7309,
+      "step": 3719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211081507037605e-05,
+      "loss": 0.7847,
+      "step": 3720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206954260617617e-05,
+      "loss": 0.7149,
+      "step": 3721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202826805553271e-05,
+      "loss": 0.7735,
+      "step": 3722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198699142558072e-05,
+      "loss": 0.7093,
+      "step": 3723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194571272345555e-05,
+      "loss": 0.7507,
+      "step": 3724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119044319562931e-05,
+      "loss": 0.7734,
+      "step": 3725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1186314913122944e-05,
+      "loss": 0.7766,
+      "step": 3726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118218642554011e-05,
+      "loss": 0.7864,
+      "step": 3727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178057733594495e-05,
+      "loss": 0.7583,
+      "step": 3728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173928837999812e-05,
+      "loss": 0.7511,
+      "step": 3729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169799739469827e-05,
+      "loss": 0.7415,
+      "step": 3730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165670438718323e-05,
+      "loss": 0.7694,
+      "step": 3731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161540936459133e-05,
+      "loss": 0.7561,
+      "step": 3732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157411233406113e-05,
+      "loss": 0.8138,
+      "step": 3733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153281330273157e-05,
+      "loss": 0.7229,
+      "step": 3734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149151227774203e-05,
+      "loss": 0.7445,
+      "step": 3735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145020926623206e-05,
+      "loss": 0.7294,
+      "step": 3736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114089042753417e-05,
+      "loss": 0.7889,
+      "step": 3737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113675973122113e-05,
+      "loss": 0.7405,
+      "step": 3738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113262883839815e-05,
+      "loss": 0.6515,
+      "step": 3739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128497749779328e-05,
+      "loss": 0.7822,
+      "step": 3740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124366466078809e-05,
+      "loss": 0.8197,
+      "step": 3741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120234988010752e-05,
+      "loss": 0.7353,
+      "step": 3742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116103316289363e-05,
+      "loss": 0.8668,
+      "step": 3743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111971451628883e-05,
+      "loss": 0.7831,
+      "step": 3744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107839394743571e-05,
+      "loss": 0.8072,
+      "step": 3745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103707146347738e-05,
+      "loss": 0.7651,
+      "step": 3746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099574707155714e-05,
+      "loss": 0.8063,
+      "step": 3747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095442077881871e-05,
+      "loss": 0.7591,
+      "step": 3748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091309259240608e-05,
+      "loss": 0.7824,
+      "step": 3749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087176251946361e-05,
+      "loss": 0.7791,
+      "step": 3750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083043056713599e-05,
+      "loss": 0.837,
+      "step": 3751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078909674256813e-05,
+      "loss": 0.8428,
+      "step": 3752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074776105290541e-05,
+      "loss": 0.7732,
+      "step": 3753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070642350529347e-05,
+      "loss": 0.2759,
+      "step": 3754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066508410687826e-05,
+      "loss": 0.7984,
+      "step": 3755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062374286480603e-05,
+      "loss": 0.8069,
+      "step": 3756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058239978622345e-05,
+      "loss": 0.8234,
+      "step": 3757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054105487827738e-05,
+      "loss": 0.7018,
+      "step": 3758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049970814811501e-05,
+      "loss": 0.8025,
+      "step": 3759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045835960288399e-05,
+      "loss": 0.7455,
+      "step": 3760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104170092497321e-05,
+      "loss": 0.8038,
+      "step": 3761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037565709580755e-05,
+      "loss": 0.7485,
+      "step": 3762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033430314825882e-05,
+      "loss": 0.3125,
+      "step": 3763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029294741423471e-05,
+      "loss": 0.7413,
+      "step": 3764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1025158990088429e-05,
+      "loss": 0.698,
+      "step": 3765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10210230615357e-05,
+      "loss": 0.8042,
+      "step": 3766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016886956480258e-05,
+      "loss": 0.7162,
+      "step": 3767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10127506756371e-05,
+      "loss": 0.7285,
+      "step": 3768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008614219721261e-05,
+      "loss": 0.7432,
+      "step": 3769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004477589447807e-05,
+      "loss": 0.7522,
+      "step": 3770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000340785531824e-05,
+      "loss": 0.7398,
+      "step": 3771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099620380868844e-05,
+      "loss": 0.6824,
+      "step": 3772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099206665963281e-05,
+      "loss": 0.7349,
+      "step": 3773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987929339080107e-05,
+      "loss": 0.7385,
+      "step": 3774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983791847745555e-05,
+      "loss": 0.759,
+      "step": 3775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097965418634439e-05,
+      "loss": 0.7476,
+      "step": 3776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975516355591886e-05,
+      "loss": 0.6861,
+      "step": 3777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971378356203339e-05,
+      "loss": 0.7334,
+      "step": 3778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967240188894083e-05,
+      "loss": 0.6872,
+      "step": 3779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963101854379477e-05,
+      "loss": 0.7401,
+      "step": 3780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958963353374904e-05,
+      "loss": 0.6932,
+      "step": 3781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954824686595784e-05,
+      "loss": 0.7772,
+      "step": 3782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950685854757563e-05,
+      "loss": 0.806,
+      "step": 3783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946546858575711e-05,
+      "loss": 0.7477,
+      "step": 3784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942407698765733e-05,
+      "loss": 0.8328,
+      "step": 3785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093826837604316e-05,
+      "loss": 0.6882,
+      "step": 3786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934128891123546e-05,
+      "loss": 0.7617,
+      "step": 3787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092998924472248e-05,
+      "loss": 0.737,
+      "step": 3788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925849437555578e-05,
+      "loss": 0.6911,
+      "step": 3789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921709470338482e-05,
+      "loss": 0.7455,
+      "step": 3790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917569343786859e-05,
+      "loss": 0.7235,
+      "step": 3791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091342905861641e-05,
+      "loss": 0.6676,
+      "step": 3792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909288615542857e-05,
+      "loss": 0.7502,
+      "step": 3793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905148015281953e-05,
+      "loss": 0.7368,
+      "step": 3794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901007258549479e-05,
+      "loss": 0.7281,
+      "step": 3795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896866346061239e-05,
+      "loss": 0.7635,
+      "step": 3796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892725278533064e-05,
+      "loss": 0.7891,
+      "step": 3797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088858405668082e-05,
+      "loss": 0.6032,
+      "step": 3798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884442681220393e-05,
+      "loss": 0.7297,
+      "step": 3799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880301152867691e-05,
+      "loss": 0.7917,
+      "step": 3800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087615947233866e-05,
+      "loss": 0.7729,
+      "step": 3801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872017640349265e-05,
+      "loss": 0.7785,
+      "step": 3802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867875657615492e-05,
+      "loss": 0.7344,
+      "step": 3803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863733524853369e-05,
+      "loss": 0.7483,
+      "step": 3804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085959124277893e-05,
+      "loss": 0.7506,
+      "step": 3805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855448812108252e-05,
+      "loss": 0.7286,
+      "step": 3806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085130623355743e-05,
+      "loss": 0.7686,
+      "step": 3807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847163507842583e-05,
+      "loss": 0.8046,
+      "step": 3808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843020635679859e-05,
+      "loss": 0.7177,
+      "step": 3809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838877617785428e-05,
+      "loss": 0.7534,
+      "step": 3810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834734454875493e-05,
+      "loss": 0.7655,
+      "step": 3811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830591147666269e-05,
+      "loss": 0.8223,
+      "step": 3812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082644769687401e-05,
+      "loss": 0.7315,
+      "step": 3813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822304103214982e-05,
+      "loss": 0.7255,
+      "step": 3814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818160367405487e-05,
+      "loss": 0.7341,
+      "step": 3815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814016490161843e-05,
+      "loss": 0.7733,
+      "step": 3816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809872472200395e-05,
+      "loss": 0.7278,
+      "step": 3817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080572831423752e-05,
+      "loss": 0.7965,
+      "step": 3818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801584016989602e-05,
+      "loss": 0.2994,
+      "step": 3819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797439581173069e-05,
+      "loss": 0.7341,
+      "step": 3820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793295007504355e-05,
+      "loss": 0.7614,
+      "step": 3821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789150296699929e-05,
+      "loss": 0.6914,
+      "step": 3822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785005449476286e-05,
+      "loss": 0.7919,
+      "step": 3823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780860466549936e-05,
+      "loss": 0.8021,
+      "step": 3824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776715348637412e-05,
+      "loss": 0.7151,
+      "step": 3825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772570096455281e-05,
+      "loss": 0.746,
+      "step": 3826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768424710720124e-05,
+      "loss": 0.7124,
+      "step": 3827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764279192148544e-05,
+      "loss": 0.6964,
+      "step": 3828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760133541457176e-05,
+      "loss": 0.8591,
+      "step": 3829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075598775936267e-05,
+      "loss": 0.7984,
+      "step": 3830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751841846581697e-05,
+      "loss": 0.738,
+      "step": 3831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747695803830962e-05,
+      "loss": 0.7755,
+      "step": 3832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074354963182718e-05,
+      "loss": 0.7665,
+      "step": 3833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739403331287097e-05,
+      "loss": 0.7471,
+      "step": 3834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735256902927474e-05,
+      "loss": 0.7005,
+      "step": 3835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07311103474651e-05,
+      "loss": 0.704,
+      "step": 3836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726963665616787e-05,
+      "loss": 0.7707,
+      "step": 3837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722816858099359e-05,
+      "loss": 0.7248,
+      "step": 3838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718669925629673e-05,
+      "loss": 0.7504,
+      "step": 3839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714522868924601e-05,
+      "loss": 0.8416,
+      "step": 3840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710375688701037e-05,
+      "loss": 0.7737,
+      "step": 3841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706228385675903e-05,
+      "loss": 0.7613,
+      "step": 3842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702080960566134e-05,
+      "loss": 0.8077,
+      "step": 3843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697933414088683e-05,
+      "loss": 0.7588,
+      "step": 3844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693785746960541e-05,
+      "loss": 0.6632,
+      "step": 3845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689637959898705e-05,
+      "loss": 0.8052,
+      "step": 3846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685490053620193e-05,
+      "loss": 0.8535,
+      "step": 3847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681342028842052e-05,
+      "loss": 0.6894,
+      "step": 3848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677193886281341e-05,
+      "loss": 0.7385,
+      "step": 3849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673045626655145e-05,
+      "loss": 0.8072,
+      "step": 3850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668897250680569e-05,
+      "loss": 0.7424,
+      "step": 3851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664748759074733e-05,
+      "loss": 0.7184,
+      "step": 3852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066060015255478e-05,
+      "loss": 0.7746,
+      "step": 3853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656451431837879e-05,
+      "loss": 0.7608,
+      "step": 3854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652302597641211e-05,
+      "loss": 0.765,
+      "step": 3855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648153650681974e-05,
+      "loss": 0.7454,
+      "step": 3856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644004591677392e-05,
+      "loss": 0.7836,
+      "step": 3857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639855421344713e-05,
+      "loss": 0.7198,
+      "step": 3858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635706140401192e-05,
+      "loss": 0.7138,
+      "step": 3859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631556749564107e-05,
+      "loss": 0.8249,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062740724955076e-05,
+      "loss": 0.744,
+      "step": 3861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0623257641078473e-05,
+      "loss": 0.7943,
+      "step": 3862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619107924864574e-05,
+      "loss": 0.729,
+      "step": 3863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614958101626424e-05,
+      "loss": 0.7789,
+      "step": 3864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610808172081394e-05,
+      "loss": 0.77,
+      "step": 3865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606658136946877e-05,
+      "loss": 0.7955,
+      "step": 3866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602507996940286e-05,
+      "loss": 0.7154,
+      "step": 3867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598357752779043e-05,
+      "loss": 0.6881,
+      "step": 3868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594207405180599e-05,
+      "loss": 0.7167,
+      "step": 3869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059005695486242e-05,
+      "loss": 0.7158,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585906402541986e-05,
+      "loss": 0.7684,
+      "step": 3871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581755748936792e-05,
+      "loss": 0.6513,
+      "step": 3872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577604994764363e-05,
+      "loss": 0.7286,
+      "step": 3873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573454140742233e-05,
+      "loss": 0.7954,
+      "step": 3874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569303187587945e-05,
+      "loss": 0.8208,
+      "step": 3875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565152136019077e-05,
+      "loss": 0.727,
+      "step": 3876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561000986753213e-05,
+      "loss": 0.7514,
+      "step": 3877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556849740507954e-05,
+      "loss": 0.7593,
+      "step": 3878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552698398000918e-05,
+      "loss": 0.7692,
+      "step": 3879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548546959949749e-05,
+      "loss": 0.7903,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0544395427072094e-05,
+      "loss": 0.7348,
+      "step": 3881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540243800085618e-05,
+      "loss": 0.7173,
+      "step": 3882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536092079708018e-05,
+      "loss": 0.7811,
+      "step": 3883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531940266656988e-05,
+      "loss": 0.7864,
+      "step": 3884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527788361650244e-05,
+      "loss": 0.7195,
+      "step": 3885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523636365405525e-05,
+      "loss": 0.7267,
+      "step": 3886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519484278640578e-05,
+      "loss": 0.7517,
+      "step": 3887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515332102073164e-05,
+      "loss": 0.7622,
+      "step": 3888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051117983642107e-05,
+      "loss": 0.6711,
+      "step": 3889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050702748240209e-05,
+      "loss": 0.3056,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502875040734034e-05,
+      "loss": 0.7655,
+      "step": 3891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498722512134724e-05,
+      "loss": 0.7394,
+      "step": 3892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494569897322011e-05,
+      "loss": 0.7445,
+      "step": 3893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490417197013746e-05,
+      "loss": 0.732,
+      "step": 3894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486264411927797e-05,
+      "loss": 0.7288,
+      "step": 3895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482111542782056e-05,
+      "loss": 0.6653,
+      "step": 3896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477958590294422e-05,
+      "loss": 0.7467,
+      "step": 3897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473805555182804e-05,
+      "loss": 0.7111,
+      "step": 3898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469652438165134e-05,
+      "loss": 0.8047,
+      "step": 3899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046549923995936e-05,
+      "loss": 0.7182,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461345961283432e-05,
+      "loss": 0.7209,
+      "step": 3901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457192602855323e-05,
+      "loss": 0.7969,
+      "step": 3902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045303916539302e-05,
+      "loss": 0.7539,
+      "step": 3903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448885649614517e-05,
+      "loss": 0.8132,
+      "step": 3904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444732056237834e-05,
+      "loss": 0.8558,
+      "step": 3905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440578385980992e-05,
+      "loss": 0.7215,
+      "step": 3906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436424639562027e-05,
+      "loss": 0.7211,
+      "step": 3907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432270817698993e-05,
+      "loss": 0.7311,
+      "step": 3908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428116921109956e-05,
+      "loss": 0.6674,
+      "step": 3909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423962950512992e-05,
+      "loss": 0.716,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0419808906626194e-05,
+      "loss": 0.7497,
+      "step": 3911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415654790167665e-05,
+      "loss": 0.7986,
+      "step": 3912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411500601855515e-05,
+      "loss": 0.7712,
+      "step": 3913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407346342407879e-05,
+      "loss": 0.7829,
+      "step": 3914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403192012542895e-05,
+      "loss": 0.7571,
+      "step": 3915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399037612978719e-05,
+      "loss": 0.7784,
+      "step": 3916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394883144433512e-05,
+      "loss": 0.8025,
+      "step": 3917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390728607625451e-05,
+      "loss": 0.7971,
+      "step": 3918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386574003272727e-05,
+      "loss": 0.7888,
+      "step": 3919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382419332093534e-05,
+      "loss": 0.7888,
+      "step": 3920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378264594806094e-05,
+      "loss": 0.7312,
+      "step": 3921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374109792128621e-05,
+      "loss": 0.8482,
+      "step": 3922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369954924779354e-05,
+      "loss": 0.7716,
+      "step": 3923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365799993476535e-05,
+      "loss": 0.7719,
+      "step": 3924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361644998938428e-05,
+      "loss": 0.8312,
+      "step": 3925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357489941883287e-05,
+      "loss": 0.7771,
+      "step": 3926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353334823029407e-05,
+      "loss": 0.6985,
+      "step": 3927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034917964309507e-05,
+      "loss": 0.7029,
+      "step": 3928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345024402798569e-05,
+      "loss": 0.7496,
+      "step": 3929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340869102858224e-05,
+      "loss": 0.8317,
+      "step": 3930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336713743992353e-05,
+      "loss": 0.6753,
+      "step": 3931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332558326919281e-05,
+      "loss": 0.8201,
+      "step": 3932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328402852357357e-05,
+      "loss": 0.8035,
+      "step": 3933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324247321024927e-05,
+      "loss": 0.807,
+      "step": 3934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.032009173364035e-05,
+      "loss": 0.7002,
+      "step": 3935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315936090921998e-05,
+      "loss": 0.679,
+      "step": 3936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031178039358825e-05,
+      "loss": 0.7488,
+      "step": 3937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307624642357498e-05,
+      "loss": 0.7911,
+      "step": 3938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303468837948139e-05,
+      "loss": 0.7757,
+      "step": 3939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029931298107858e-05,
+      "loss": 0.7588,
+      "step": 3940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295157072467236e-05,
+      "loss": 0.721,
+      "step": 3941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291001112832535e-05,
+      "loss": 0.7856,
+      "step": 3942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028684510289291e-05,
+      "loss": 0.6955,
+      "step": 3943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0282689043366807e-05,
+      "loss": 0.7863,
+      "step": 3944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278532934972671e-05,
+      "loss": 0.7646,
+      "step": 3945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274376778428971e-05,
+      "loss": 0.7788,
+      "step": 3946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0270220574454172e-05,
+      "loss": 0.8127,
+      "step": 3947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026606432376675e-05,
+      "loss": 0.77,
+      "step": 3948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261908027085186e-05,
+      "loss": 0.732,
+      "step": 3949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025775168512798e-05,
+      "loss": 0.7146,
+      "step": 3950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253595298613626e-05,
+      "loss": 0.2791,
+      "step": 3951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024943886826064e-05,
+      "loss": 0.7582,
+      "step": 3952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245282394787531e-05,
+      "loss": 0.7006,
+      "step": 3953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241125878912824e-05,
+      "loss": 0.6665,
+      "step": 3954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023696932135505e-05,
+      "loss": 0.805,
+      "step": 3955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232812722832748e-05,
+      "loss": 0.698,
+      "step": 3956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228656084064461e-05,
+      "loss": 0.7403,
+      "step": 3957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224499405768744e-05,
+      "loss": 0.7278,
+      "step": 3958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220342688664154e-05,
+      "loss": 0.7703,
+      "step": 3959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216185933469252e-05,
+      "loss": 0.7306,
+      "step": 3960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212029140902614e-05,
+      "loss": 0.7897,
+      "step": 3961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020787231168282e-05,
+      "loss": 0.6945,
+      "step": 3962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203715446528454e-05,
+      "loss": 0.2826,
+      "step": 3963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199558546158106e-05,
+      "loss": 0.7548,
+      "step": 3964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0195401611290375e-05,
+      "loss": 0.7806,
+      "step": 3965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019124464264386e-05,
+      "loss": 0.7036,
+      "step": 3966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018708764093717e-05,
+      "loss": 0.6633,
+      "step": 3967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182930606888926e-05,
+      "loss": 0.7397,
+      "step": 3968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178773541217744e-05,
+      "loss": 0.7315,
+      "step": 3969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174616444642247e-05,
+      "loss": 0.7305,
+      "step": 3970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017045931788107e-05,
+      "loss": 0.7582,
+      "step": 3971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166302161652848e-05,
+      "loss": 0.7048,
+      "step": 3972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016214497667622e-05,
+      "loss": 0.7588,
+      "step": 3973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157987763669836e-05,
+      "loss": 0.6598,
+      "step": 3974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153830523352349e-05,
+      "loss": 0.7551,
+      "step": 3975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149673256442408e-05,
+      "loss": 0.7867,
+      "step": 3976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145515963658678e-05,
+      "loss": 0.7425,
+      "step": 3977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141358645719825e-05,
+      "loss": 0.7475,
+      "step": 3978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137201303344512e-05,
+      "loss": 0.7582,
+      "step": 3979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133043937251421e-05,
+      "loss": 0.703,
+      "step": 3980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128886548159224e-05,
+      "loss": 0.7532,
+      "step": 3981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124729136786603e-05,
+      "loss": 0.8222,
+      "step": 3982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120571703852244e-05,
+      "loss": 0.6984,
+      "step": 3983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116414250074835e-05,
+      "loss": 0.7916,
+      "step": 3984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112256776173076e-05,
+      "loss": 0.7161,
+      "step": 3985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108099282865651e-05,
+      "loss": 0.7596,
+      "step": 3986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010394177087127e-05,
+      "loss": 0.7501,
+      "step": 3987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099784240908632e-05,
+      "loss": 0.7361,
+      "step": 3988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095626693696441e-05,
+      "loss": 0.6984,
+      "step": 3989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009146912995341e-05,
+      "loss": 0.7427,
+      "step": 3990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087311550398246e-05,
+      "loss": 0.7852,
+      "step": 3991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083153955749667e-05,
+      "loss": 0.717,
+      "step": 3992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078996346726391e-05,
+      "loss": 0.7187,
+      "step": 3993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074838724047136e-05,
+      "loss": 0.7447,
+      "step": 3994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070681088430615e-05,
+      "loss": 0.7614,
+      "step": 3995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006652344059557e-05,
+      "loss": 0.72,
+      "step": 3996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062365781260717e-05,
+      "loss": 0.7651,
+      "step": 3997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058208111144785e-05,
+      "loss": 0.8276,
+      "step": 3998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054050430966504e-05,
+      "loss": 0.7606,
+      "step": 3999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049892741444607e-05,
+      "loss": 0.6874,
+      "step": 4000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045735043297826e-05,
+      "loss": 0.921,
+      "step": 4001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041577337244898e-05,
+      "loss": 0.7051,
+      "step": 4002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037419624004561e-05,
+      "loss": 0.81,
+      "step": 4003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033261904295547e-05,
+      "loss": 0.8063,
+      "step": 4004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029104178836597e-05,
+      "loss": 0.7134,
+      "step": 4005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024946448346455e-05,
+      "loss": 0.762,
+      "step": 4006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020788713543854e-05,
+      "loss": 0.7154,
+      "step": 4007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016630975147541e-05,
+      "loss": 0.6622,
+      "step": 4008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012473233876256e-05,
+      "loss": 0.7352,
+      "step": 4009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008315490448744e-05,
+      "loss": 0.7502,
+      "step": 4010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004157745583742e-05,
+      "loss": 0.7812,
+      "step": 4011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7325,
+      "step": 4012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995842254416258e-06,
+      "loss": 0.7161,
+      "step": 4013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991684509551258e-06,
+      "loss": 0.7948,
+      "step": 4014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987526766123743e-06,
+      "loss": 0.7356,
+      "step": 4015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983369024852464e-06,
+      "loss": 0.7238,
+      "step": 4016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979211286456149e-06,
+      "loss": 0.7942,
+      "step": 4017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975053551653551e-06,
+      "loss": 0.7371,
+      "step": 4018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970895821163405e-06,
+      "loss": 0.733,
+      "step": 4019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966738095704457e-06,
+      "loss": 0.7177,
+      "step": 4020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962580375995442e-06,
+      "loss": 0.7783,
+      "step": 4021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958422662755107e-06,
+      "loss": 0.7561,
+      "step": 4022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954264956702176e-06,
+      "loss": 0.3065,
+      "step": 4023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950107258555393e-06,
+      "loss": 0.6111,
+      "step": 4024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945949569033498e-06,
+      "loss": 0.7415,
+      "step": 4025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941791888855217e-06,
+      "loss": 0.8035,
+      "step": 4026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937634218739285e-06,
+      "loss": 0.6924,
+      "step": 4027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93347655940443e-06,
+      "loss": 0.7106,
+      "step": 4028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929318911569386e-06,
+      "loss": 0.7523,
+      "step": 4029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925161275952868e-06,
+      "loss": 0.2787,
+      "step": 4030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921003653273614e-06,
+      "loss": 0.6745,
+      "step": 4031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916846044250335e-06,
+      "loss": 0.7536,
+      "step": 4032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912688449601758e-06,
+      "loss": 0.7713,
+      "step": 4033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908530870046592e-06,
+      "loss": 0.8224,
+      "step": 4034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904373306303562e-06,
+      "loss": 0.6643,
+      "step": 4035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900215759091371e-06,
+      "loss": 0.664,
+      "step": 4036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89605822912873e-06,
+      "loss": 0.7533,
+      "step": 4037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891900717134352e-06,
+      "loss": 0.733,
+      "step": 4038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887743223826928e-06,
+      "loss": 0.7551,
+      "step": 4039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883585749925167e-06,
+      "loss": 0.7816,
+      "step": 4040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879428296147759e-06,
+      "loss": 0.2883,
+      "step": 4041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8752708632134e-06,
+      "loss": 0.7196,
+      "step": 4042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87111345184078e-06,
+      "loss": 0.6959,
+      "step": 4043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866956062748585e-06,
+      "loss": 0.766,
+      "step": 4044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86279869665549e-06,
+      "loss": 0.7431,
+      "step": 4045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858641354280181e-06,
+      "loss": 0.7154,
+      "step": 4046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854484036341323e-06,
+      "loss": 0.7842,
+      "step": 4047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850326743557594e-06,
+      "loss": 0.7571,
+      "step": 4048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.846169476647655e-06,
+      "loss": 0.7614,
+      "step": 4049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842012236330162e-06,
+      "loss": 0.7768,
+      "step": 4050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837855023323784e-06,
+      "loss": 0.7346,
+      "step": 4051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833697838347153e-06,
+      "loss": 0.7359,
+      "step": 4052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829540682118935e-06,
+      "loss": 0.6865,
+      "step": 4053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825383555357756e-06,
+      "loss": 0.8034,
+      "step": 4054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821226458782261e-06,
+      "loss": 0.6865,
+      "step": 4055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817069393111075e-06,
+      "loss": 0.7998,
+      "step": 4056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812912359062835e-06,
+      "loss": 0.7052,
+      "step": 4057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808755357356143e-06,
+      "loss": 0.7364,
+      "step": 4058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804598388709627e-06,
+      "loss": 0.7327,
+      "step": 4059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800441453841897e-06,
+      "loss": 0.7485,
+      "step": 4060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.796284553471547e-06,
+      "loss": 0.7311,
+      "step": 4061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792127688317182e-06,
+      "loss": 0.729,
+      "step": 4062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787970859097386e-06,
+      "loss": 0.7606,
+      "step": 4063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783814066530751e-06,
+      "loss": 0.2191,
+      "step": 4064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77965731133585e-06,
+      "loss": 0.7615,
+      "step": 4065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775500594231261e-06,
+      "loss": 0.7785,
+      "step": 4066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771343915935542e-06,
+      "loss": 0.7428,
+      "step": 4067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767187277167256e-06,
+      "loss": 0.7607,
+      "step": 4068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763030678644952e-06,
+      "loss": 0.687,
+      "step": 4069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758874121087178e-06,
+      "loss": 0.705,
+      "step": 4070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754717605212472e-06,
+      "loss": 0.7665,
+      "step": 4071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750561131739362e-06,
+      "loss": 0.6854,
+      "step": 4072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746404701386377e-06,
+      "loss": 0.7947,
+      "step": 4073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742248314872022e-06,
+      "loss": 0.7557,
+      "step": 4074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738091972914818e-06,
+      "loss": 0.7795,
+      "step": 4075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733935676233255e-06,
+      "loss": 0.281,
+      "step": 4076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729779425545833e-06,
+      "loss": 0.7656,
+      "step": 4077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72562322157103e-06,
+      "loss": 0.7143,
+      "step": 4078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721467065027332e-06,
+      "loss": 0.737,
+      "step": 4079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717310956633197e-06,
+      "loss": 0.6908,
+      "step": 4080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713154897107094e-06,
+      "loss": 0.74,
+      "step": 4081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708998887167468e-06,
+      "loss": 0.7681,
+      "step": 4082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704842927532765e-06,
+      "loss": 0.7734,
+      "step": 4083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700687018921425e-06,
+      "loss": 0.8143,
+      "step": 4084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696531162051863e-06,
+      "loss": 0.2682,
+      "step": 4085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692375357642503e-06,
+      "loss": 0.7004,
+      "step": 4086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68821960641175e-06,
+      "loss": 0.7771,
+      "step": 4087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684063909078005e-06,
+      "loss": 0.7322,
+      "step": 4088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679908266359653e-06,
+      "loss": 0.703,
+      "step": 4089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675752678975078e-06,
+      "loss": 0.2903,
+      "step": 4090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671597147642645e-06,
+      "loss": 0.7741,
+      "step": 4091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66744167308072e-06,
+      "loss": 0.7648,
+      "step": 4092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66328625600765e-06,
+      "loss": 0.764,
+      "step": 4093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659130897141776e-06,
+      "loss": 0.7978,
+      "step": 4094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654975597201433e-06,
+      "loss": 0.7731,
+      "step": 4095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650820356904932e-06,
+      "loss": 0.7226,
+      "step": 4096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646665176970595e-06,
+      "loss": 0.7756,
+      "step": 4097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642510058116713e-06,
+      "loss": 0.7343,
+      "step": 4098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638355001061577e-06,
+      "loss": 0.7382,
+      "step": 4099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634200006523467e-06,
+      "loss": 0.7799,
+      "step": 4100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630045075220653e-06,
+      "loss": 0.7487,
+      "step": 4101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625890207871382e-06,
+      "loss": 0.7464,
+      "step": 4102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621735405193911e-06,
+      "loss": 0.6867,
+      "step": 4103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617580667906468e-06,
+      "loss": 0.7523,
+      "step": 4104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613425996727278e-06,
+      "loss": 0.7377,
+      "step": 4105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609271392374552e-06,
+      "loss": 0.7339,
+      "step": 4106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60511685556649e-06,
+      "loss": 0.8034,
+      "step": 4107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600962387021283e-06,
+      "loss": 0.7583,
+      "step": 4108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596807987457103e-06,
+      "loss": 0.6902,
+      "step": 4109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592653657592125e-06,
+      "loss": 0.7252,
+      "step": 4110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588499398144488e-06,
+      "loss": 0.697,
+      "step": 4111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58434520983234e-06,
+      "loss": 0.6499,
+      "step": 4112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58019109337381e-06,
+      "loss": 0.7596,
+      "step": 4113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576037049487011e-06,
+      "loss": 0.7672,
+      "step": 4114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571883078890046e-06,
+      "loss": 0.7814,
+      "step": 4115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567729182301012e-06,
+      "loss": 0.7125,
+      "step": 4116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563575360437977e-06,
+      "loss": 0.714,
+      "step": 4117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559421614019011e-06,
+      "loss": 0.6904,
+      "step": 4118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555267943762168e-06,
+      "loss": 0.6701,
+      "step": 4119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551114350385483e-06,
+      "loss": 0.2443,
+      "step": 4120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546960834606983e-06,
+      "loss": 0.7938,
+      "step": 4121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542807397144679e-06,
+      "loss": 0.8021,
+      "step": 4122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538654038716575e-06,
+      "loss": 0.704,
+      "step": 4123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534500760040644e-06,
+      "loss": 0.803,
+      "step": 4124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53034756183487e-06,
+      "loss": 0.2756,
+      "step": 4125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5261944448172e-06,
+      "loss": 0.2747,
+      "step": 4126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522041409705585e-06,
+      "loss": 0.8069,
+      "step": 4127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517888457217946e-06,
+      "loss": 0.7172,
+      "step": 4128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513735588072203e-06,
+      "loss": 0.7093,
+      "step": 4129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509582802986257e-06,
+      "loss": 0.7943,
+      "step": 4130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505430102677989e-06,
+      "loss": 0.7712,
+      "step": 4131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501277487865277e-06,
+      "loss": 0.2698,
+      "step": 4132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497124959265971e-06,
+      "loss": 0.7922,
+      "step": 4133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492972517597914e-06,
+      "loss": 0.6995,
+      "step": 4134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488820163578933e-06,
+      "loss": 0.6873,
+      "step": 4135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484667897926838e-06,
+      "loss": 0.6992,
+      "step": 4136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480515721359426e-06,
+      "loss": 0.2598,
+      "step": 4137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47636363459448e-06,
+      "loss": 0.7682,
+      "step": 4138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47221163834976e-06,
+      "loss": 0.7792,
+      "step": 4139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468059733343016e-06,
+      "loss": 0.737,
+      "step": 4140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463907920291985e-06,
+      "loss": 0.7487,
+      "step": 4141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45975619991438e-06,
+      "loss": 0.7041,
+      "step": 4142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45560457292791e-06,
+      "loss": 0.7797,
+      "step": 4143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451453040050253e-06,
+      "loss": 0.6978,
+      "step": 4144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447301601999083e-06,
+      "loss": 0.7067,
+      "step": 4145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44315025949205e-06,
+      "loss": 0.7562,
+      "step": 4146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438999013246792e-06,
+      "loss": 0.7972,
+      "step": 4147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434847863980925e-06,
+      "loss": 0.7216,
+      "step": 4148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430696812412058e-06,
+      "loss": 0.7498,
+      "step": 4149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426545859257772e-06,
+      "loss": 0.8062,
+      "step": 4150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422395005235642e-06,
+      "loss": 0.7636,
+      "step": 4151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418244251063211e-06,
+      "loss": 0.7327,
+      "step": 4152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414093597458016e-06,
+      "loss": 0.7138,
+      "step": 4153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409943045137583e-06,
+      "loss": 0.7861,
+      "step": 4154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405792594819403e-06,
+      "loss": 0.7603,
+      "step": 4155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40164224722096e-06,
+      "loss": 0.7496,
+      "step": 4156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397492003059718e-06,
+      "loss": 0.7813,
+      "step": 4157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393341863053126e-06,
+      "loss": 0.7522,
+      "step": 4158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389191827918607e-06,
+      "loss": 0.7966,
+      "step": 4159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38504189837358e-06,
+      "loss": 0.7003,
+      "step": 4160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380892075135429e-06,
+      "loss": 0.7271,
+      "step": 4161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.376742358921532e-06,
+      "loss": 0.8093,
+      "step": 4162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372592750449242e-06,
+      "loss": 0.7309,
+      "step": 4163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368443250435895e-06,
+      "loss": 0.8183,
+      "step": 4164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364293859598812e-06,
+      "loss": 0.7019,
+      "step": 4165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360144578655289e-06,
+      "loss": 0.7434,
+      "step": 4166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35599540832261e-06,
+      "loss": 0.6683,
+      "step": 4167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351846349318029e-06,
+      "loss": 0.7769,
+      "step": 4168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347697402358794e-06,
+      "loss": 0.7034,
+      "step": 4169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343548568162123e-06,
+      "loss": 0.7359,
+      "step": 4170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339399847445222e-06,
+      "loss": 0.7754,
+      "step": 4171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335251240925269e-06,
+      "loss": 0.782,
+      "step": 4172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331102749319436e-06,
+      "loss": 0.6468,
+      "step": 4173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326954373344859e-06,
+      "loss": 0.7882,
+      "step": 4174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322806113718659e-06,
+      "loss": 0.7394,
+      "step": 4175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318657971157952e-06,
+      "loss": 0.6692,
+      "step": 4176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314509946379809e-06,
+      "loss": 0.7632,
+      "step": 4177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310362040101297e-06,
+      "loss": 0.721,
+      "step": 4178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306214253039459e-06,
+      "loss": 0.6824,
+      "step": 4179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30206658591132e-06,
+      "loss": 0.6367,
+      "step": 4180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29791903943387e-06,
+      "loss": 0.7466,
+      "step": 4181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2937716143241e-06,
+      "loss": 0.7166,
+      "step": 4182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289624311298964e-06,
+      "loss": 0.7022,
+      "step": 4183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285477131075402e-06,
+      "loss": 0.7458,
+      "step": 4184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281330074370329e-06,
+      "loss": 0.7848,
+      "step": 4185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277183141900641e-06,
+      "loss": 0.7548,
+      "step": 4186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273036334383215e-06,
+      "loss": 0.7591,
+      "step": 4187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268889652534898e-06,
+      "loss": 0.7266,
+      "step": 4188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264743097072528e-06,
+      "loss": 0.698,
+      "step": 4189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260596668712906e-06,
+      "loss": 0.7819,
+      "step": 4190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.256450368172822e-06,
+      "loss": 0.7971,
+      "step": 4191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252304196169041e-06,
+      "loss": 0.8454,
+      "step": 4192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248158153418305e-06,
+      "loss": 0.6438,
+      "step": 4193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244012240637333e-06,
+      "loss": 0.789,
+      "step": 4194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23986645854283e-06,
+      "loss": 0.7687,
+      "step": 4195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23572080785146e-06,
+      "loss": 0.7486,
+      "step": 4196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231575289279881e-06,
+      "loss": 0.7672,
+      "step": 4197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22742990354472e-06,
+      "loss": 0.765,
+      "step": 4198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223284651362588e-06,
+      "loss": 0.6815,
+      "step": 4199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219139533450067e-06,
+      "loss": 0.6984,
+      "step": 4200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214994550523715e-06,
+      "loss": 0.6998,
+      "step": 4201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210849703300073e-06,
+      "loss": 0.7455,
+      "step": 4202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206704992495647e-06,
+      "loss": 0.6409,
+      "step": 4203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202560418826938e-06,
+      "loss": 0.7351,
+      "step": 4204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198415983010402e-06,
+      "loss": 0.7808,
+      "step": 4205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194271685762485e-06,
+      "loss": 0.8242,
+      "step": 4206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190127527799607e-06,
+      "loss": 0.7304,
+      "step": 4207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185983509838164e-06,
+      "loss": 0.7643,
+      "step": 4208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181839632594516e-06,
+      "loss": 0.7483,
+      "step": 4209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177695896785018e-06,
+      "loss": 0.7601,
+      "step": 4210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173552303125994e-06,
+      "loss": 0.7582,
+      "step": 4211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169408852333733e-06,
+      "loss": 0.6878,
+      "step": 4212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16526554512451e-06,
+      "loss": 0.727,
+      "step": 4213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161122382214572e-06,
+      "loss": 0.6857,
+      "step": 4214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156979364320144e-06,
+      "loss": 0.6968,
+      "step": 4215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152836492157418e-06,
+      "loss": 0.283,
+      "step": 4216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148693766442575e-06,
+      "loss": 0.744,
+      "step": 4217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14455118789175e-06,
+      "loss": 0.7553,
+      "step": 4218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140408757221073e-06,
+      "loss": 0.7626,
+      "step": 4219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136266475146636e-06,
+      "loss": 0.7713,
+      "step": 4220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132124342384508e-06,
+      "loss": 0.7445,
+      "step": 4221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127982359650737e-06,
+      "loss": 0.7445,
+      "step": 4222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12384052766134e-06,
+      "loss": 0.8016,
+      "step": 4223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11969884713231e-06,
+      "loss": 0.7844,
+      "step": 4224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115557318779608e-06,
+      "loss": 0.7538,
+      "step": 4225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111415943319182e-06,
+      "loss": 0.7295,
+      "step": 4226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107274721466937e-06,
+      "loss": 0.7362,
+      "step": 4227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103133653938766e-06,
+      "loss": 0.7085,
+      "step": 4228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098992741450524e-06,
+      "loss": 0.7497,
+      "step": 4229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094851984718052e-06,
+      "loss": 0.7184,
+      "step": 4230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090711384457146e-06,
+      "loss": 0.8125,
+      "step": 4231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086570941383597e-06,
+      "loss": 0.7744,
+      "step": 4232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082430656213145e-06,
+      "loss": 0.6767,
+      "step": 4233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078290529661522e-06,
+      "loss": 0.6882,
+      "step": 4234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074150562444424e-06,
+      "loss": 0.7822,
+      "step": 4235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070010755277522e-06,
+      "loss": 0.6831,
+      "step": 4236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065871108876458e-06,
+      "loss": 0.8048,
+      "step": 4237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061731623956842e-06,
+      "loss": 0.7356,
+      "step": 4238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057592301234272e-06,
+      "loss": 0.2521,
+      "step": 4239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053453141424292e-06,
+      "loss": 0.6848,
+      "step": 4240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049314145242442e-06,
+      "loss": 0.758,
+      "step": 4241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045175313404218e-06,
+      "loss": 0.7569,
+      "step": 4242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041036646625098e-06,
+      "loss": 0.6802,
+      "step": 4243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036898145620526e-06,
+      "loss": 0.73,
+      "step": 4244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032759811105917e-06,
+      "loss": 0.7089,
+      "step": 4245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028621643796664e-06,
+      "loss": 0.7739,
+      "step": 4246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024483644408116e-06,
+      "loss": 0.6921,
+      "step": 4247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020345813655613e-06,
+      "loss": 0.7231,
+      "step": 4248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016208152254447e-06,
+      "loss": 0.7519,
+      "step": 4249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012070660919894e-06,
+      "loss": 0.8032,
+      "step": 4250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007933340367194e-06,
+      "loss": 0.6921,
+      "step": 4251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003796191311565e-06,
+      "loss": 0.7742,
+      "step": 4252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999659214468179e-06,
+      "loss": 0.7471,
+      "step": 4253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9955224105522e-06,
+      "loss": 0.7873,
+      "step": 4254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991385780278742e-06,
+      "loss": 0.7237,
+      "step": 4255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987249324362902e-06,
+      "loss": 0.6846,
+      "step": 4256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983113043519746e-06,
+      "loss": 0.7307,
+      "step": 4257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9789769384643e-06,
+      "loss": 0.7653,
+      "step": 4258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974841009911573e-06,
+      "loss": 0.6722,
+      "step": 4259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97070525857653e-06,
+      "loss": 0.7438,
+      "step": 4260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966569685174121e-06,
+      "loss": 0.667,
+      "step": 4261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962434290419248e-06,
+      "loss": 0.8005,
+      "step": 4262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958299075026793e-06,
+      "loss": 0.7828,
+      "step": 4263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954164039711603e-06,
+      "loss": 0.686,
+      "step": 4264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9500291851885e-06,
+      "loss": 0.6664,
+      "step": 4265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945894512172267e-06,
+      "loss": 0.7358,
+      "step": 4266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941760021377655e-06,
+      "loss": 0.7374,
+      "step": 4267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937625713519398e-06,
+      "loss": 0.7651,
+      "step": 4268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933491589312177e-06,
+      "loss": 0.7692,
+      "step": 4269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929357649470656e-06,
+      "loss": 0.7105,
+      "step": 4270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92522389470946e-06,
+      "loss": 0.7354,
+      "step": 4271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92109032574319e-06,
+      "loss": 0.7521,
+      "step": 4272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916956943286404e-06,
+      "loss": 0.7593,
+      "step": 4273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912823748053642e-06,
+      "loss": 0.7467,
+      "step": 4274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908690740759395e-06,
+      "loss": 0.8767,
+      "step": 4275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904557922118134e-06,
+      "loss": 0.7982,
+      "step": 4276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90042529284429e-06,
+      "loss": 0.7247,
+      "step": 4277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896292853652267e-06,
+      "loss": 0.7572,
+      "step": 4278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89216060525643e-06,
+      "loss": 0.6666,
+      "step": 4279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888028548371119e-06,
+      "loss": 0.7351,
+      "step": 4280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883896683710638e-06,
+      "loss": 0.7111,
+      "step": 4281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879765011989247e-06,
+      "loss": 0.7513,
+      "step": 4282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875633533921195e-06,
+      "loss": 0.7715,
+      "step": 4283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871502250220673e-06,
+      "loss": 0.283,
+      "step": 4284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867371161601855e-06,
+      "loss": 0.7336,
+      "step": 4285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863240268778873e-06,
+      "loss": 0.7283,
+      "step": 4286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859109572465833e-06,
+      "loss": 0.6671,
+      "step": 4287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854979073376796e-06,
+      "loss": 0.7648,
+      "step": 4288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850848772225804e-06,
+      "loss": 0.6423,
+      "step": 4289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846718669726846e-06,
+      "loss": 0.7823,
+      "step": 4290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84258876659389e-06,
+      "loss": 0.7468,
+      "step": 4291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83845906354087e-06,
+      "loss": 0.7806,
+      "step": 4292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834329561281679e-06,
+      "loss": 0.7555,
+      "step": 4293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830200260530176e-06,
+      "loss": 0.7441,
+      "step": 4294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826071162000188e-06,
+      "loss": 0.7411,
+      "step": 4295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821942266405512e-06,
+      "loss": 0.6856,
+      "step": 4296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817813574459892e-06,
+      "loss": 0.7447,
+      "step": 4297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81368508687706e-06,
+      "loss": 0.7373,
+      "step": 4298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.809556804370694e-06,
+      "loss": 0.8353,
+      "step": 4299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805428727654447e-06,
+      "loss": 0.6593,
+      "step": 4300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801300857441932e-06,
+      "loss": 0.7343,
+      "step": 4301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797173194446729e-06,
+      "loss": 0.7442,
+      "step": 4302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793045739382385e-06,
+      "loss": 0.7323,
+      "step": 4303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788918492962397e-06,
+      "loss": 0.2816,
+      "step": 4304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784791455900245e-06,
+      "loss": 0.7552,
+      "step": 4305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780664628909358e-06,
+      "loss": 0.7582,
+      "step": 4306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776538012703138e-06,
+      "loss": 0.7244,
+      "step": 4307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772411607994945e-06,
+      "loss": 0.72,
+      "step": 4308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768285415498107e-06,
+      "loss": 0.7611,
+      "step": 4309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764159435925904e-06,
+      "loss": 0.6837,
+      "step": 4310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7600336699916e-06,
+      "loss": 0.7128,
+      "step": 4311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755908118408401e-06,
+      "loss": 0.7692,
+      "step": 4312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75178278188949e-06,
+      "loss": 0.785,
+      "step": 4313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747657661148004e-06,
+      "loss": 0.6614,
+      "step": 4314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743532756897049e-06,
+      "loss": 0.6685,
+      "step": 4315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739408069849691e-06,
+      "loss": 0.768,
+      "step": 4316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735283600718955e-06,
+      "loss": 0.7164,
+      "step": 4317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731159350217839e-06,
+      "loss": 0.7014,
+      "step": 4318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727035319059288e-06,
+      "loss": 0.7353,
+      "step": 4319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722911507956222e-06,
+      "loss": 0.6996,
+      "step": 4320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718787917621513e-06,
+      "loss": 0.7173,
+      "step": 4321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714664548768008e-06,
+      "loss": 0.7177,
+      "step": 4322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710541402108498e-06,
+      "loss": 0.7916,
+      "step": 4323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706418478355754e-06,
+      "loss": 0.7229,
+      "step": 4324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.702295778222492e-06,
+      "loss": 0.749,
+      "step": 4325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698173302421398e-06,
+      "loss": 0.7835,
+      "step": 4326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694051051665128e-06,
+      "loss": 0.7949,
+      "step": 4327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689929026666277e-06,
+      "loss": 0.6996,
+      "step": 4328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685807228137421e-06,
+      "loss": 0.6839,
+      "step": 4329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681685656791083e-06,
+      "loss": 0.773,
+      "step": 4330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.677564313339759e-06,
+      "loss": 0.7438,
+      "step": 4331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673443198495893e-06,
+      "loss": 0.7499,
+      "step": 4332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669322312971903e-06,
+      "loss": 0.7261,
+      "step": 4333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665201657480154e-06,
+      "loss": 0.8174,
+      "step": 4334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661081232732982e-06,
+      "loss": 0.7058,
+      "step": 4335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656961039442672e-06,
+      "loss": 0.7108,
+      "step": 4336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652841078321483e-06,
+      "loss": 0.7393,
+      "step": 4337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648721350081626e-06,
+      "loss": 0.7518,
+      "step": 4338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644601855435268e-06,
+      "loss": 0.669,
+      "step": 4339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640482595094544e-06,
+      "loss": 0.7663,
+      "step": 4340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63636356977154e-06,
+      "loss": 0.7599,
+      "step": 4341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63224478017831e-06,
+      "loss": 0.7582,
+      "step": 4342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628126227026857e-06,
+      "loss": 0.7367,
+      "step": 4343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624007911029156e-06,
+      "loss": 0.7621,
+      "step": 4344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619889832897126e-06,
+      "loss": 0.7,
+      "step": 4345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615771993342665e-06,
+      "loss": 0.7447,
+      "step": 4346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611654393077606e-06,
+      "loss": 0.7205,
+      "step": 4347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607537032813758e-06,
+      "loss": 0.7057,
+      "step": 4348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60341991326288e-06,
+      "loss": 0.7042,
+      "step": 4349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599303035136694e-06,
+      "loss": 0.7619,
+      "step": 4350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59518639914688e-06,
+      "loss": 0.7393,
+      "step": 4351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591070006005072e-06,
+      "loss": 0.8247,
+      "step": 4352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58695385642287e-06,
+      "loss": 0.767,
+      "step": 4353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582837951111815e-06,
+      "loss": 0.7265,
+      "step": 4354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578722290783433e-06,
+      "loss": 0.7212,
+      "step": 4355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574606876149177e-06,
+      "loss": 0.7546,
+      "step": 4356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570491707920483e-06,
+      "loss": 0.705,
+      "step": 4357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566376786808728e-06,
+      "loss": 0.6715,
+      "step": 4358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562262113525258e-06,
+      "loss": 0.653,
+      "step": 4359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55814768878136e-06,
+      "loss": 0.7775,
+      "step": 4360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554033513288298e-06,
+      "loss": 0.7195,
+      "step": 4361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549919587757283e-06,
+      "loss": 0.7227,
+      "step": 4362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545805912899479e-06,
+      "loss": 0.7371,
+      "step": 4363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541692489426013e-06,
+      "loss": 0.7232,
+      "step": 4364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537579318047963e-06,
+      "loss": 0.2927,
+      "step": 4365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533466399476372e-06,
+      "loss": 0.7464,
+      "step": 4366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529353734422229e-06,
+      "loss": 0.6779,
+      "step": 4367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525241323596491e-06,
+      "loss": 0.6882,
+      "step": 4368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521129167710055e-06,
+      "loss": 0.7205,
+      "step": 4369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51701726747379e-06,
+      "loss": 0.712,
+      "step": 4370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51290562359851e-06,
+      "loss": 0.6978,
+      "step": 4371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50879423679499e-06,
+      "loss": 0.7284,
+      "step": 4372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504683107773964e-06,
+      "loss": 0.7553,
+      "step": 4373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500572237246108e-06,
+      "loss": 0.7015,
+      "step": 4374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496461625922072e-06,
+      "loss": 0.7472,
+      "step": 4375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.492351274512443e-06,
+      "loss": 0.7213,
+      "step": 4376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488241183727774e-06,
+      "loss": 0.7416,
+      "step": 4377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484131354278568e-06,
+      "loss": 0.7776,
+      "step": 4378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48002178687529e-06,
+      "loss": 0.7292,
+      "step": 4379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47591248222835e-06,
+      "loss": 0.7562,
+      "step": 4380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471803441048122e-06,
+      "loss": 0.6734,
+      "step": 4381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46769466404492e-06,
+      "loss": 0.7236,
+      "step": 4382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463586151929033e-06,
+      "loss": 0.672,
+      "step": 4383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459477905410693e-06,
+      "loss": 0.6991,
+      "step": 4384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455369925200077e-06,
+      "loss": 0.7813,
+      "step": 4385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451262212007335e-06,
+      "loss": 0.7455,
+      "step": 4386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447154766542554e-06,
+      "loss": 0.7851,
+      "step": 4387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443047589515789e-06,
+      "loss": 0.66,
+      "step": 4388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438940681637033e-06,
+      "loss": 0.6966,
+      "step": 4389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434834043616251e-06,
+      "loss": 0.7159,
+      "step": 4390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430727676163341e-06,
+      "loss": 0.8113,
+      "step": 4391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426621579988173e-06,
+      "loss": 0.744,
+      "step": 4392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422515755800554e-06,
+      "loss": 0.8006,
+      "step": 4393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418410204310257e-06,
+      "loss": 0.7185,
+      "step": 4394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414304926227002e-06,
+      "loss": 0.7487,
+      "step": 4395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410199922260459e-06,
+      "loss": 0.7362,
+      "step": 4396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406095193120261e-06,
+      "loss": 0.7179,
+      "step": 4397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401990739515978e-06,
+      "loss": 0.7594,
+      "step": 4398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397886562157146e-06,
+      "loss": 0.7364,
+      "step": 4399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393782661753245e-06,
+      "loss": 0.7349,
+      "step": 4400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389679039013712e-06,
+      "loss": 0.7157,
+      "step": 4401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385575694647932e-06,
+      "loss": 0.7691,
+      "step": 4402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381472629365249e-06,
+      "loss": 0.7419,
+      "step": 4403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377369843874945e-06,
+      "loss": 0.7357,
+      "step": 4404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37326733888627e-06,
+      "loss": 0.6807,
+      "step": 4405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369165115108413e-06,
+      "loss": 0.703,
+      "step": 4406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365063173250523e-06,
+      "loss": 0.7673,
+      "step": 4407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360961514021695e-06,
+      "loss": 0.7065,
+      "step": 4408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356860138130975e-06,
+      "loss": 0.7221,
+      "step": 4409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35275904628737e-06,
+      "loss": 0.7697,
+      "step": 4410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348658239199811e-06,
+      "loss": 0.6819,
+      "step": 4411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344557717577221e-06,
+      "loss": 0.7437,
+      "step": 4412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340457482128434e-06,
+      "loss": 0.7382,
+      "step": 4413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33635753356226e-06,
+      "loss": 0.7197,
+      "step": 4414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332257872587446e-06,
+      "loss": 0.7479,
+      "step": 4415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328158499912699e-06,
+      "loss": 0.7445,
+      "step": 4416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324059416246668e-06,
+      "loss": 0.7414,
+      "step": 4417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319960622297956e-06,
+      "loss": 0.7353,
+      "step": 4418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31586211877512e-06,
+      "loss": 0.6994,
+      "step": 4419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311763906386653e-06,
+      "loss": 0.7325,
+      "step": 4420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307665985841013e-06,
+      "loss": 0.7269,
+      "step": 4421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3035683578466e-06,
+      "loss": 0.7424,
+      "step": 4422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299471023111766e-06,
+      "loss": 0.7094,
+      "step": 4423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295373982344807e-06,
+      "loss": 0.7198,
+      "step": 4424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.291277236253983e-06,
+      "loss": 0.7475,
+      "step": 4425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287180785547476e-06,
+      "loss": 0.7404,
+      "step": 4426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283084630933446e-06,
+      "loss": 0.7039,
+      "step": 4427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278988773119982e-06,
+      "loss": 0.7677,
+      "step": 4428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274893212815134e-06,
+      "loss": 0.7445,
+      "step": 4429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270797950726889e-06,
+      "loss": 0.7716,
+      "step": 4430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266702987563195e-06,
+      "loss": 0.697,
+      "step": 4431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262608324031944e-06,
+      "loss": 0.8019,
+      "step": 4432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258513960840961e-06,
+      "loss": 0.2834,
+      "step": 4433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254419898698051e-06,
+      "loss": 0.7752,
+      "step": 4434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250326138310933e-06,
+      "loss": 0.6883,
+      "step": 4435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246232680387297e-06,
+      "loss": 0.6747,
+      "step": 4436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24213952563477e-06,
+      "loss": 0.7683,
+      "step": 4437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238046674760934e-06,
+      "loss": 0.7635,
+      "step": 4438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.233954128473302e-06,
+      "loss": 0.7965,
+      "step": 4439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229861887479364e-06,
+      "loss": 0.7703,
+      "step": 4440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225769952486523e-06,
+      "loss": 0.7387,
+      "step": 4441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221678324202155e-06,
+      "loss": 0.7356,
+      "step": 4442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217587003333573e-06,
+      "loss": 0.6809,
+      "step": 4443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213495990588033e-06,
+      "loss": 0.6917,
+      "step": 4444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209405286672746e-06,
+      "loss": 0.7974,
+      "step": 4445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205314892294865e-06,
+      "loss": 0.8044,
+      "step": 4446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201224808161495e-06,
+      "loss": 0.847,
+      "step": 4447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197135034979671e-06,
+      "loss": 0.7438,
+      "step": 4448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193045573456397e-06,
+      "loss": 0.7597,
+      "step": 4449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188956424298604e-06,
+      "loss": 0.7924,
+      "step": 4450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184867588213184e-06,
+      "loss": 0.7034,
+      "step": 4451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180779065906962e-06,
+      "loss": 0.7602,
+      "step": 4452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176690858086717e-06,
+      "loss": 0.7934,
+      "step": 4453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172602965459177e-06,
+      "loss": 0.7696,
+      "step": 4454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168515388730998e-06,
+      "loss": 0.2664,
+      "step": 4455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164428128608806e-06,
+      "loss": 0.7616,
+      "step": 4456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16034118579915e-06,
+      "loss": 0.801,
+      "step": 4457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156254561008538e-06,
+      "loss": 0.781,
+      "step": 4458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152168254943416e-06,
+      "loss": 0.7563,
+      "step": 4459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148082268310183e-06,
+      "loss": 0.6241,
+      "step": 4460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143996601815167e-06,
+      "loss": 0.7441,
+      "step": 4461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139911256164664e-06,
+      "loss": 0.7499,
+      "step": 4462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13582623206489e-06,
+      "loss": 0.6782,
+      "step": 4463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131741530222021e-06,
+      "loss": 0.7196,
+      "step": 4464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12765715134218e-06,
+      "loss": 0.7852,
+      "step": 4465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123573096131417e-06,
+      "loss": 0.7125,
+      "step": 4466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119489365295743e-06,
+      "loss": 0.7385,
+      "step": 4467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115405959541102e-06,
+      "loss": 0.7693,
+      "step": 4468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111322879573395e-06,
+      "loss": 0.754,
+      "step": 4469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107240126098446e-06,
+      "loss": 0.7204,
+      "step": 4470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103157699822041e-06,
+      "loss": 0.7251,
+      "step": 4471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099075601449903e-06,
+      "loss": 0.7913,
+      "step": 4472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094993831687696e-06,
+      "loss": 0.7693,
+      "step": 4473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09091239124103e-06,
+      "loss": 0.7968,
+      "step": 4474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08683128081546e-06,
+      "loss": 0.7079,
+      "step": 4475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082750501116479e-06,
+      "loss": 0.6701,
+      "step": 4476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07867005284952e-06,
+      "loss": 0.7374,
+      "step": 4477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074589936719976e-06,
+      "loss": 0.6805,
+      "step": 4478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070510153433162e-06,
+      "loss": 0.7191,
+      "step": 4479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066430703694346e-06,
+      "loss": 0.2716,
+      "step": 4480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062351588208734e-06,
+      "loss": 0.2586,
+      "step": 4481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058272807681483e-06,
+      "loss": 0.6798,
+      "step": 4482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054194362817676e-06,
+      "loss": 0.7058,
+      "step": 4483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050116254322359e-06,
+      "loss": 0.7176,
+      "step": 4484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046038482900498e-06,
+      "loss": 0.6971,
+      "step": 4485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041961049257018e-06,
+      "loss": 0.8154,
+      "step": 4486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037883954096775e-06,
+      "loss": 0.7418,
+      "step": 4487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03380719812457e-06,
+      "loss": 0.7422,
+      "step": 4488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029730782045152e-06,
+      "loss": 0.6931,
+      "step": 4489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025654706563195e-06,
+      "loss": 0.773,
+      "step": 4490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021578972383337e-06,
+      "loss": 0.7392,
+      "step": 4491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017503580210129e-06,
+      "loss": 0.7415,
+      "step": 4492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013428530748089e-06,
+      "loss": 0.7293,
+      "step": 4493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009353824701659e-06,
+      "loss": 0.7222,
+      "step": 4494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005279462775231e-06,
+      "loss": 0.8414,
+      "step": 4495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001205445673129e-06,
+      "loss": 0.6966,
+      "step": 4496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99713177409963e-06,
+      "loss": 0.7413,
+      "step": 4497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993058448758933e-06,
+      "loss": 0.7703,
+      "step": 4498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.988985470355195e-06,
+      "loss": 0.6953,
+      "step": 4499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984912839592506e-06,
+      "loss": 0.7691,
+      "step": 4500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98084055717489e-06,
+      "loss": 0.7186,
+      "step": 4501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976768623806323e-06,
+      "loss": 0.7721,
+      "step": 4502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97269704019071e-06,
+      "loss": 0.713,
+      "step": 4503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968625807031901e-06,
+      "loss": 0.7279,
+      "step": 4504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964554925033681e-06,
+      "loss": 0.6494,
+      "step": 4505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960484394899781e-06,
+      "loss": 0.6877,
+      "step": 4506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956414217333864e-06,
+      "loss": 0.7732,
+      "step": 4507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952344393039539e-06,
+      "loss": 0.7359,
+      "step": 4508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948274922720344e-06,
+      "loss": 0.745,
+      "step": 4509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944205807079775e-06,
+      "loss": 0.7781,
+      "step": 4510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940137046821237e-06,
+      "loss": 0.7341,
+      "step": 4511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936068642648103e-06,
+      "loss": 0.7578,
+      "step": 4512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93200059526367e-06,
+      "loss": 0.768,
+      "step": 4513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92793290537117e-06,
+      "loss": 0.7847,
+      "step": 4514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923865573673783e-06,
+      "loss": 0.6523,
+      "step": 4515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919798600874622e-06,
+      "loss": 0.7412,
+      "step": 4516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915731987676737e-06,
+      "loss": 0.6782,
+      "step": 4517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911665734783115e-06,
+      "loss": 0.7398,
+      "step": 4518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907599842896691e-06,
+      "loss": 0.666,
+      "step": 4519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903534312720322e-06,
+      "loss": 0.6975,
+      "step": 4520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899469144956812e-06,
+      "loss": 0.7662,
+      "step": 4521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8954043403089e-06,
+      "loss": 0.6827,
+      "step": 4522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891339899479264e-06,
+      "loss": 0.7665,
+      "step": 4523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887275823170518e-06,
+      "loss": 0.7283,
+      "step": 4524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88321211208521e-06,
+      "loss": 0.7434,
+      "step": 4525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879148766925835e-06,
+      "loss": 0.7822,
+      "step": 4526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875085788394807e-06,
+      "loss": 0.742,
+      "step": 4527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871023177194496e-06,
+      "loss": 0.756,
+      "step": 4528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866960934027192e-06,
+      "loss": 0.7543,
+      "step": 4529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862899059595134e-06,
+      "loss": 0.7208,
+      "step": 4530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858837554600488e-06,
+      "loss": 0.7726,
+      "step": 4531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854776419745369e-06,
+      "loss": 0.7318,
+      "step": 4532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.850715655731805e-06,
+      "loss": 0.7314,
+      "step": 4533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846655263261781e-06,
+      "loss": 0.7527,
+      "step": 4534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84259524303722e-06,
+      "loss": 0.7098,
+      "step": 4535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83853559575996e-06,
+      "loss": 0.7133,
+      "step": 4536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83447632213179e-06,
+      "loss": 0.7592,
+      "step": 4537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830417422854428e-06,
+      "loss": 0.7347,
+      "step": 4538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826358898629536e-06,
+      "loss": 0.7968,
+      "step": 4539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822300750158695e-06,
+      "loss": 0.6969,
+      "step": 4540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818242978143443e-06,
+      "loss": 0.7299,
+      "step": 4541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81418558328523e-06,
+      "loss": 0.73,
+      "step": 4542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81012856628546e-06,
+      "loss": 0.7676,
+      "step": 4543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806071927845457e-06,
+      "loss": 0.7063,
+      "step": 4544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802015668666491e-06,
+      "loss": 0.7084,
+      "step": 4545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797959789449759e-06,
+      "loss": 0.7222,
+      "step": 4546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793904290896392e-06,
+      "loss": 0.7039,
+      "step": 4547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789849173707467e-06,
+      "loss": 0.7624,
+      "step": 4548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785794438583976e-06,
+      "loss": 0.6858,
+      "step": 4549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78174008622686e-06,
+      "loss": 0.2661,
+      "step": 4550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777686117336986e-06,
+      "loss": 0.6878,
+      "step": 4551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773632532615161e-06,
+      "loss": 0.7658,
+      "step": 4552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769579332762118e-06,
+      "loss": 0.7063,
+      "step": 4553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765526518478533e-06,
+      "loss": 0.742,
+      "step": 4554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761474090465005e-06,
+      "loss": 0.7426,
+      "step": 4555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757422049422072e-06,
+      "loss": 0.7127,
+      "step": 4556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753370396050204e-06,
+      "loss": 0.7662,
+      "step": 4557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.749319131049804e-06,
+      "loss": 0.777,
+      "step": 4558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745268255121214e-06,
+      "loss": 0.6947,
+      "step": 4559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741217768964695e-06,
+      "loss": 0.7981,
+      "step": 4560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737167673280454e-06,
+      "loss": 0.631,
+      "step": 4561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733117968768618e-06,
+      "loss": 0.7042,
+      "step": 4562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729068656129264e-06,
+      "loss": 0.7516,
+      "step": 4563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725019736062378e-06,
+      "loss": 0.7321,
+      "step": 4564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7209712092679e-06,
+      "loss": 0.7372,
+      "step": 4565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71692307644569e-06,
+      "loss": 0.709,
+      "step": 4566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712875338295544e-06,
+      "loss": 0.7322,
+      "step": 4567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708827995517181e-06,
+      "loss": 0.7405,
+      "step": 4568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704781048810266e-06,
+      "loss": 0.8288,
+      "step": 4569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700734498874393e-06,
+      "loss": 0.6832,
+      "step": 4570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696688346409071e-06,
+      "loss": 0.7428,
+      "step": 4571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692642592113762e-06,
+      "loss": 0.7726,
+      "step": 4572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688597236687844e-06,
+      "loss": 0.7201,
+      "step": 4573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.684552280830634e-06,
+      "loss": 0.7911,
+      "step": 4574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.680507725241378e-06,
+      "loss": 0.7636,
+      "step": 4575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676463570619253e-06,
+      "loss": 0.7732,
+      "step": 4576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67241981766336e-06,
+      "loss": 0.8136,
+      "step": 4577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668376467072743e-06,
+      "loss": 0.7528,
+      "step": 4578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664333519546365e-06,
+      "loss": 0.6727,
+      "step": 4579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660290975783127e-06,
+      "loss": 0.7403,
+      "step": 4580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656248836481859e-06,
+      "loss": 0.7179,
+      "step": 4581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652207102341316e-06,
+      "loss": 0.7008,
+      "step": 4582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648165774060191e-06,
+      "loss": 0.7353,
+      "step": 4583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644124852337094e-06,
+      "loss": 0.6914,
+      "step": 4584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640084337870585e-06,
+      "loss": 0.6842,
+      "step": 4585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63604423135913e-06,
+      "loss": 0.7283,
+      "step": 4586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632004533501144e-06,
+      "loss": 0.7706,
+      "step": 4587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627965244994958e-06,
+      "loss": 0.699,
+      "step": 4588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623926366538845e-06,
+      "loss": 0.6499,
+      "step": 4589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619887898830989e-06,
+      "loss": 0.7591,
+      "step": 4590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615849842569526e-06,
+      "loss": 0.7113,
+      "step": 4591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611812198452499e-06,
+      "loss": 0.2666,
+      "step": 4592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607774967177892e-06,
+      "loss": 0.737,
+      "step": 4593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603738149443619e-06,
+      "loss": 0.7242,
+      "step": 4594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599701745947514e-06,
+      "loss": 0.7043,
+      "step": 4595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5956657573873475e-06,
+      "loss": 0.8147,
+      "step": 4596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591630184460811e-06,
+      "loss": 0.808,
+      "step": 4597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5875950278655345e-06,
+      "loss": 0.6935,
+      "step": 4598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58356028829906e-06,
+      "loss": 0.8299,
+      "step": 4599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579525966458874e-06,
+      "loss": 0.8094,
+      "step": 4600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57549206304238e-06,
+      "loss": 0.7418,
+      "step": 4601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571458578746915e-06,
+      "loss": 0.2793,
+      "step": 4602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567425514269739e-06,
+      "loss": 0.6778,
+      "step": 4603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563392870308044e-06,
+      "loss": 0.7164,
+      "step": 4604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5593606475589495e-06,
+      "loss": 0.7688,
+      "step": 4605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555328846719491e-06,
+      "loss": 0.6586,
+      "step": 4606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5512974684866515e-06,
+      "loss": 0.7933,
+      "step": 4607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5472665135573186e-06,
+      "loss": 0.767,
+      "step": 4608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543235982628325e-06,
+      "loss": 0.7092,
+      "step": 4609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539205876396416e-06,
+      "loss": 0.7443,
+      "step": 4610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535176195558279e-06,
+      "loss": 0.7117,
+      "step": 4611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5311469408105066e-06,
+      "loss": 0.7618,
+      "step": 4612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527118112849643e-06,
+      "loss": 0.7445,
+      "step": 4613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523089712372135e-06,
+      "loss": 0.7581,
+      "step": 4614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519061740074371e-06,
+      "loss": 0.7287,
+      "step": 4615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515034196652661e-06,
+      "loss": 0.7428,
+      "step": 4616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511007082803237e-06,
+      "loss": 0.7519,
+      "step": 4617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506980399222266e-06,
+      "loss": 0.7284,
+      "step": 4618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502954146605828e-06,
+      "loss": 0.7314,
+      "step": 4619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4989283256499416e-06,
+      "loss": 0.7845,
+      "step": 4620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494902937050537e-06,
+      "loss": 0.7502,
+      "step": 4621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490877981503485e-06,
+      "loss": 0.7706,
+      "step": 4622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486853459704566e-06,
+      "loss": 0.6944,
+      "step": 4623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4828293723495e-06,
+      "loss": 0.7576,
+      "step": 4624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478805720133919e-06,
+      "loss": 0.71,
+      "step": 4625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474782503753392e-06,
+      "loss": 0.7544,
+      "step": 4626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4707597239034e-06,
+      "loss": 0.6923,
+      "step": 4627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4667373812793565e-06,
+      "loss": 0.7193,
+      "step": 4628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462715476576601e-06,
+      "loss": 0.7056,
+      "step": 4629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45869401049039e-06,
+      "loss": 0.662,
+      "step": 4630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454672983715913e-06,
+      "loss": 0.7474,
+      "step": 4631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450652396948272e-06,
+      "loss": 0.7004,
+      "step": 4632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446632250882511e-06,
+      "loss": 0.7625,
+      "step": 4633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442612546213573e-06,
+      "loss": 0.7801,
+      "step": 4634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438593283636346e-06,
+      "loss": 0.6821,
+      "step": 4635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434574463845632e-06,
+      "loss": 0.749,
+      "step": 4636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43055608753616e-06,
+      "loss": 0.7199,
+      "step": 4637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426538155402578e-06,
+      "loss": 0.6976,
+      "step": 4638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422520668139462e-06,
+      "loss": 0.7477,
+      "step": 4639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4185036264413105e-06,
+      "loss": 0.751,
+      "step": 4640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414487031002538e-06,
+      "loss": 0.764,
+      "step": 4641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410470882517495e-06,
+      "loss": 0.6876,
+      "step": 4642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406455181680439e-06,
+      "loss": 0.277,
+      "step": 4643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402439929185563e-06,
+      "loss": 0.6546,
+      "step": 4644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398425125726976e-06,
+      "loss": 0.6967,
+      "step": 4645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394410771998711e-06,
+      "loss": 0.7094,
+      "step": 4646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3903968686947225e-06,
+      "loss": 0.809,
+      "step": 4647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386383416508893e-06,
+      "loss": 0.2949,
+      "step": 4648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3823704161350154e-06,
+      "loss": 0.6557,
+      "step": 4649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378357868266812e-06,
+      "loss": 0.6456,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3743457735979305e-06,
+      "loss": 0.2734,
+      "step": 4651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37033413282193e-06,
+      "loss": 0.7532,
+      "step": 4652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366322946632302e-06,
+      "loss": 0.7394,
+      "step": 4653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362312215722449e-06,
+      "loss": 0.664,
+      "step": 4654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358301940785707e-06,
+      "loss": 0.7284,
+      "step": 4655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354292122515317e-06,
+      "loss": 0.6865,
+      "step": 4656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350282761604458e-06,
+      "loss": 0.7733,
+      "step": 4657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3462738587462175e-06,
+      "loss": 0.7543,
+      "step": 4658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342265414633611e-06,
+      "loss": 0.6528,
+      "step": 4659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33825742995957e-06,
+      "loss": 0.701,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33424990541695e-06,
+      "loss": 0.7458,
+      "step": 4661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330242841698531e-06,
+      "loss": 0.707,
+      "step": 4662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326236239496996e-06,
+      "loss": 0.7067,
+      "step": 4663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322230099504972e-06,
+      "loss": 0.7525,
+      "step": 4664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318224422414988e-06,
+      "loss": 0.7923,
+      "step": 4665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314219208919502e-06,
+      "loss": 0.7301,
+      "step": 4666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310214459710885e-06,
+      "loss": 0.7734,
+      "step": 4667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306210175481441e-06,
+      "loss": 0.7642,
+      "step": 4668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30220635692337e-06,
+      "loss": 0.7337,
+      "step": 4669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298203004728823e-06,
+      "loss": 0.7118,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29420011958984e-06,
+      "loss": 0.6032,
+      "step": 4671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290197702198402e-06,
+      "loss": 0.7131,
+      "step": 4672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286195753246394e-06,
+      "loss": 0.7677,
+      "step": 4673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282194273425633e-06,
+      "loss": 0.2557,
+      "step": 4674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278193263427846e-06,
+      "loss": 0.7555,
+      "step": 4675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274192723944682e-06,
+      "loss": 0.7309,
+      "step": 4676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270192655667711e-06,
+      "loss": 0.7904,
+      "step": 4677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266193059288413e-06,
+      "loss": 0.7634,
+      "step": 4678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262193935498196e-06,
+      "loss": 0.7611,
+      "step": 4679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25819528498838e-06,
+      "loss": 0.791,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.254197108450209e-06,
+      "loss": 0.7265,
+      "step": 4681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250199406574839e-06,
+      "loss": 0.7549,
+      "step": 4682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246202180053352e-06,
+      "loss": 0.7351,
+      "step": 4683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242205429576733e-06,
+      "loss": 0.706,
+      "step": 4684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238209155835898e-06,
+      "loss": 0.7791,
+      "step": 4685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234213359521684e-06,
+      "loss": 0.7852,
+      "step": 4686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230218041324829e-06,
+      "loss": 0.7772,
+      "step": 4687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226223201936004e-06,
+      "loss": 0.6803,
+      "step": 4688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222228842045784e-06,
+      "loss": 0.686,
+      "step": 4689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218234962344678e-06,
+      "loss": 0.7823,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214241563523089e-06,
+      "loss": 0.6751,
+      "step": 4691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210248646271363e-06,
+      "loss": 0.7723,
+      "step": 4692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20625621127974e-06,
+      "loss": 0.7114,
+      "step": 4693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20226425923839e-06,
+      "loss": 0.7019,
+      "step": 4694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198272790837395e-06,
+      "loss": 0.7893,
+      "step": 4695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194281806766754e-06,
+      "loss": 0.7017,
+      "step": 4696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1902913077163835e-06,
+      "loss": 0.711,
+      "step": 4697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186301294376115e-06,
+      "loss": 0.7393,
+      "step": 4698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1823117674356975e-06,
+      "loss": 0.7289,
+      "step": 4699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17832272758479e-06,
+      "loss": 0.7157,
+      "step": 4700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174334175512975e-06,
+      "loss": 0.7104,
+      "step": 4701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1703461119097454e-06,
+      "loss": 0.7223,
+      "step": 4702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166358537464514e-06,
+      "loss": 0.7654,
+      "step": 4703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162371452866606e-06,
+      "loss": 0.7415,
+      "step": 4704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158384858805264e-06,
+      "loss": 0.7689,
+      "step": 4705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1543987559696404e-06,
+      "loss": 0.6936,
+      "step": 4706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150413145048811e-06,
+      "loss": 0.6943,
+      "step": 4707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14642802673176e-06,
+      "loss": 0.7892,
+      "step": 4708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142443401707388e-06,
+      "loss": 0.7121,
+      "step": 4709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138459270664515e-06,
+      "loss": 0.7279,
+      "step": 4710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1344756342918685e-06,
+      "loss": 0.7594,
+      "step": 4711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130492493278098e-06,
+      "loss": 0.6995,
+      "step": 4712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1265098483117535e-06,
+      "loss": 0.7143,
+      "step": 4713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122527700081321e-06,
+      "loss": 0.7667,
+      "step": 4714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118546049275179e-06,
+      "loss": 0.2985,
+      "step": 4715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114564896581635e-06,
+      "loss": 0.6563,
+      "step": 4716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110584242688901e-06,
+      "loss": 0.7262,
+      "step": 4717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106604088285112e-06,
+      "loss": 0.8134,
+      "step": 4718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1026244340583e-06,
+      "loss": 0.7203,
+      "step": 4719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098645280696435e-06,
+      "loss": 0.6716,
+      "step": 4720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0946666288873836e-06,
+      "loss": 0.7342,
+      "step": 4721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090688479318923e-06,
+      "loss": 0.7915,
+      "step": 4722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086710832678758e-06,
+      "loss": 0.7452,
+      "step": 4723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082733689654492e-06,
+      "loss": 0.7415,
+      "step": 4724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078757050933654e-06,
+      "loss": 0.6818,
+      "step": 4725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074780917203672e-06,
+      "loss": 0.6395,
+      "step": 4726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070805289151903e-06,
+      "loss": 0.6939,
+      "step": 4727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0668301674655984e-06,
+      "loss": 0.6961,
+      "step": 4728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.062855552831938e-06,
+      "loss": 0.703,
+      "step": 4729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058881445938005e-06,
+      "loss": 0.6906,
+      "step": 4730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054907847470798e-06,
+      "loss": 0.7483,
+      "step": 4731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0509347581172274e-06,
+      "loss": 0.7001,
+      "step": 4732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046962178564114e-06,
+      "loss": 0.6959,
+      "step": 4733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042990109498196e-06,
+      "loss": 0.7065,
+      "step": 4734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039018551606108e-06,
+      "loss": 0.7922,
+      "step": 4735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0350475055744216e-06,
+      "loss": 0.7923,
+      "step": 4736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031076972089594e-06,
+      "loss": 0.6971,
+      "step": 4737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0271069518380116e-06,
+      "loss": 0.7045,
+      "step": 4738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023137445505963e-06,
+      "loss": 0.724,
+      "step": 4739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019168453779654e-06,
+      "loss": 0.6663,
+      "step": 4740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015199977345192e-06,
+      "loss": 0.6719,
+      "step": 4741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0112320168886095e-06,
+      "loss": 0.715,
+      "step": 4742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007264573095836e-06,
+      "loss": 0.7075,
+      "step": 4743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003297646652717e-06,
+      "loss": 0.7464,
+      "step": 4744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999331238245015e-06,
+      "loss": 0.7444,
+      "step": 4745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99536534855839e-06,
+      "loss": 0.7412,
+      "step": 4746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991399978278426e-06,
+      "loss": 0.7229,
+      "step": 4747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987435128090605e-06,
+      "loss": 0.7225,
+      "step": 4748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9834707986803295e-06,
+      "loss": 0.7342,
+      "step": 4749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979506990732903e-06,
+      "loss": 0.7641,
+      "step": 4750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975543704933545e-06,
+      "loss": 0.8174,
+      "step": 4751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971580941967382e-06,
+      "loss": 0.7672,
+      "step": 4752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967618702519452e-06,
+      "loss": 0.7124,
+      "step": 4753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9636569872746986e-06,
+      "loss": 0.7253,
+      "step": 4754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9596957969179805e-06,
+      "loss": 0.6906,
+      "step": 4755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955735132134065e-06,
+      "loss": 0.6649,
+      "step": 4756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9517749936076186e-06,
+      "loss": 0.6834,
+      "step": 4757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947815382023232e-06,
+      "loss": 0.7844,
+      "step": 4758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943856298065391e-06,
+      "loss": 0.7589,
+      "step": 4759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939897742418502e-06,
+      "loss": 0.7363,
+      "step": 4760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935939715766871e-06,
+      "loss": 0.6832,
+      "step": 4761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9319822187947206e-06,
+      "loss": 0.7533,
+      "step": 4762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92802525218617e-06,
+      "loss": 0.7609,
+      "step": 4763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92406881662526e-06,
+      "loss": 0.6994,
+      "step": 4764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920112912795929e-06,
+      "loss": 0.7561,
+      "step": 4765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916157541382031e-06,
+      "loss": 0.7263,
+      "step": 4766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912202703067328e-06,
+      "loss": 0.6703,
+      "step": 4767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908248398535482e-06,
+      "loss": 0.6859,
+      "step": 4768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90429462847007e-06,
+      "loss": 0.6953,
+      "step": 4769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900341393554572e-06,
+      "loss": 0.7467,
+      "step": 4770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896388694472385e-06,
+      "loss": 0.7029,
+      "step": 4771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8924365319067944e-06,
+      "loss": 0.7048,
+      "step": 4772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888484906541014e-06,
+      "loss": 0.7098,
+      "step": 4773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88453381905815e-06,
+      "loss": 0.7294,
+      "step": 4774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8805832701412244e-06,
+      "loss": 0.6804,
+      "step": 4775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87663326047316e-06,
+      "loss": 0.7788,
+      "step": 4776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872683790736789e-06,
+      "loss": 0.67,
+      "step": 4777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868734861614859e-06,
+      "loss": 0.7384,
+      "step": 4778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864786473790001e-06,
+      "loss": 0.7327,
+      "step": 4779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8608386279447755e-06,
+      "loss": 0.7766,
+      "step": 4780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856891324761638e-06,
+      "loss": 0.6587,
+      "step": 4781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852944564922956e-06,
+      "loss": 0.7381,
+      "step": 4782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848998349110996e-06,
+      "loss": 0.7813,
+      "step": 4783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.845052678007939e-06,
+      "loss": 0.7176,
+      "step": 4784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84110755229586e-06,
+      "loss": 0.7723,
+      "step": 4785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837162972656755e-06,
+      "loss": 0.6828,
+      "step": 4786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83321893977251e-06,
+      "loss": 0.7513,
+      "step": 4787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82927545432493e-06,
+      "loss": 0.776,
+      "step": 4788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825332516995716e-06,
+      "loss": 0.8162,
+      "step": 4789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821390128466478e-06,
+      "loss": 0.7644,
+      "step": 4790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817448289418737e-06,
+      "loss": 0.6776,
+      "step": 4791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8135070005339e-06,
+      "loss": 0.7184,
+      "step": 4792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809566262493306e-06,
+      "loss": 0.7939,
+      "step": 4793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805626075978171e-06,
+      "loss": 0.7267,
+      "step": 4794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80168644166964e-06,
+      "loss": 0.7443,
+      "step": 4795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797747360248744e-06,
+      "loss": 0.7669,
+      "step": 4796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793808832396433e-06,
+      "loss": 0.7283,
+      "step": 4797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789870858793546e-06,
+      "loss": 0.7422,
+      "step": 4798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785933440120844e-06,
+      "loss": 0.7882,
+      "step": 4799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781996577058973e-06,
+      "loss": 0.7076,
+      "step": 4800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778060270288499e-06,
+      "loss": 0.7593,
+      "step": 4801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774124520489884e-06,
+      "loss": 0.2607,
+      "step": 4802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770189328343495e-06,
+      "loss": 0.6979,
+      "step": 4803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766254694529604e-06,
+      "loss": 0.8561,
+      "step": 4804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76232061972838e-06,
+      "loss": 0.7227,
+      "step": 4805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75838710461991e-06,
+      "loss": 0.6985,
+      "step": 4806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754454149884164e-06,
+      "loss": 0.7743,
+      "step": 4807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750521756201034e-06,
+      "loss": 0.6697,
+      "step": 4808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746589924250302e-06,
+      "loss": 0.2913,
+      "step": 4809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742658654711663e-06,
+      "loss": 0.7049,
+      "step": 4810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7387279482647026e-06,
+      "loss": 0.7211,
+      "step": 4811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734797805588921e-06,
+      "loss": 0.7082,
+      "step": 4812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730868227363718e-06,
+      "loss": 0.7772,
+      "step": 4813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726939214268387e-06,
+      "loss": 0.7495,
+      "step": 4814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723010766982139e-06,
+      "loss": 0.7645,
+      "step": 4815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719082886184072e-06,
+      "loss": 0.8082,
+      "step": 4816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715155572553196e-06,
+      "loss": 0.6831,
+      "step": 4817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711228826768416e-06,
+      "loss": 0.7917,
+      "step": 4818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70730264950855e-06,
+      "loss": 0.7246,
+      "step": 4819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7033770414523004e-06,
+      "loss": 0.6927,
+      "step": 4820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699452003278293e-06,
+      "loss": 0.7676,
+      "step": 4821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695527535665032e-06,
+      "loss": 0.6817,
+      "step": 4822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69160363929094e-06,
+      "loss": 0.7104,
+      "step": 4823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687680314834334e-06,
+      "loss": 0.7501,
+      "step": 4824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683757562973432e-06,
+      "loss": 0.7394,
+      "step": 4825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679835384386358e-06,
+      "loss": 0.7017,
+      "step": 4826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6759137797511285e-06,
+      "loss": 0.7009,
+      "step": 4827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6719927497456716e-06,
+      "loss": 0.6909,
+      "step": 4828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6680722950478e-06,
+      "loss": 0.6559,
+      "step": 4829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664152416335247e-06,
+      "loss": 0.7459,
+      "step": 4830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660233114285628e-06,
+      "loss": 0.8125,
+      "step": 4831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6563143895764725e-06,
+      "loss": 0.7343,
+      "step": 4832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652396242885201e-06,
+      "loss": 0.6824,
+      "step": 4833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648478674889143e-06,
+      "loss": 0.7225,
+      "step": 4834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6445616862655135e-06,
+      "loss": 0.7403,
+      "step": 4835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64064527769144e-06,
+      "loss": 0.7089,
+      "step": 4836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.636729449843953e-06,
+      "loss": 0.6369,
+      "step": 4837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632814203399967e-06,
+      "loss": 0.7712,
+      "step": 4838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628899539036311e-06,
+      "loss": 0.7369,
+      "step": 4839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6249854574297e-06,
+      "loss": 0.6703,
+      "step": 4840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621071959256767e-06,
+      "loss": 0.7245,
+      "step": 4841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617159045194016e-06,
+      "loss": 0.7004,
+      "step": 4842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613246715917883e-06,
+      "loss": 0.6695,
+      "step": 4843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609334972104675e-06,
+      "loss": 0.7534,
+      "step": 4844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605423814430615e-06,
+      "loss": 0.2575,
+      "step": 4845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601513243571815e-06,
+      "loss": 0.6898,
+      "step": 4846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597603260204293e-06,
+      "loss": 0.7651,
+      "step": 4847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593693865003963e-06,
+      "loss": 0.728,
+      "step": 4848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589785058646631e-06,
+      "loss": 0.7103,
+      "step": 4849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585876841808014e-06,
+      "loss": 0.7338,
+      "step": 4850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581969215163709e-06,
+      "loss": 0.6911,
+      "step": 4851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578062179389232e-06,
+      "loss": 0.7425,
+      "step": 4852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574155735159978e-06,
+      "loss": 0.6102,
+      "step": 4853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5702498831512555e-06,
+      "loss": 0.7595,
+      "step": 4854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5663446240382565e-06,
+      "loss": 0.6911,
+      "step": 4855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5624399584960855e-06,
+      "loss": 0.6575,
+      "step": 4856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558535887199726e-06,
+      "loss": 0.7457,
+      "step": 4857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554632410824075e-06,
+      "loss": 0.703,
+      "step": 4858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550729530043922e-06,
+      "loss": 0.7016,
+      "step": 4859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546827245533947e-06,
+      "loss": 0.751,
+      "step": 4860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542925557968737e-06,
+      "loss": 0.6523,
+      "step": 4861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.539024468022766e-06,
+      "loss": 0.7377,
+      "step": 4862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535123976370417e-06,
+      "loss": 0.2498,
+      "step": 4863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531224083685951e-06,
+      "loss": 0.6873,
+      "step": 4864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527324790643547e-06,
+      "loss": 0.6937,
+      "step": 4865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523426097917265e-06,
+      "loss": 0.7119,
+      "step": 4866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519528006181066e-06,
+      "loss": 0.2924,
+      "step": 4867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515630516108808e-06,
+      "loss": 0.742,
+      "step": 4868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511733628374248e-06,
+      "loss": 0.6786,
+      "step": 4869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507837343651025e-06,
+      "loss": 0.6927,
+      "step": 4870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503941662612691e-06,
+      "loss": 0.7015,
+      "step": 4871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500046585932691e-06,
+      "loss": 0.7571,
+      "step": 4872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496152114284352e-06,
+      "loss": 0.7279,
+      "step": 4873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49225824834091e-06,
+      "loss": 0.7765,
+      "step": 4874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488364988775491e-06,
+      "loss": 0.6736,
+      "step": 4875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484472336261117e-06,
+      "loss": 0.7023,
+      "step": 4876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480580291470704e-06,
+      "loss": 0.7255,
+      "step": 4877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476688855077068e-06,
+      "loss": 0.747,
+      "step": 4878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472798027752909e-06,
+      "loss": 0.7128,
+      "step": 4879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.468907810170832e-06,
+      "loss": 0.7383,
+      "step": 4880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465018203003332e-06,
+      "loss": 0.7593,
+      "step": 4881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461129206922803e-06,
+      "loss": 0.7529,
+      "step": 4882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457240822601527e-06,
+      "loss": 0.6931,
+      "step": 4883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453353050711682e-06,
+      "loss": 0.7842,
+      "step": 4884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449465891925346e-06,
+      "loss": 0.7528,
+      "step": 4885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445579346914479e-06,
+      "loss": 0.7016,
+      "step": 4886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441693416350949e-06,
+      "loss": 0.7255,
+      "step": 4887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437808100906506e-06,
+      "loss": 0.6623,
+      "step": 4888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433923401252803e-06,
+      "loss": 0.7636,
+      "step": 4889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430039318061378e-06,
+      "loss": 0.7491,
+      "step": 4890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426155852003672e-06,
+      "loss": 0.2661,
+      "step": 4891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4222730037510064e-06,
+      "loss": 0.7062,
+      "step": 4892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418390773974605e-06,
+      "loss": 0.7229,
+      "step": 4893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4145091633455925e-06,
+      "loss": 0.7045,
+      "step": 4894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410628172534965e-06,
+      "loss": 0.7378,
+      "step": 4895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406747802213631e-06,
+      "loss": 0.7965,
+      "step": 4896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402868053052379e-06,
+      "loss": 0.7662,
+      "step": 4897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3989889257219e-06,
+      "loss": 0.7665,
+      "step": 4898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395110420892769e-06,
+      "loss": 0.7051,
+      "step": 4899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391232539235465e-06,
+      "loss": 0.7137,
+      "step": 4900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387355281420341e-06,
+      "loss": 0.2644,
+      "step": 4901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383478648117659e-06,
+      "loss": 0.759,
+      "step": 4902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379602639997564e-06,
+      "loss": 0.6683,
+      "step": 4903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375727257730097e-06,
+      "loss": 0.7782,
+      "step": 4904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371852501985189e-06,
+      "loss": 0.7543,
+      "step": 4905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367978373432664e-06,
+      "loss": 0.6618,
+      "step": 4906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36410487274224e-06,
+      "loss": 0.6895,
+      "step": 4907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360232000583515e-06,
+      "loss": 0.7103,
+      "step": 4908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356359757625994e-06,
+      "loss": 0.7369,
+      "step": 4909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352488144539059e-06,
+      "loss": 0.701,
+      "step": 4910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348617161991999e-06,
+      "loss": 0.7174,
+      "step": 4911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344746810653975e-06,
+      "loss": 0.6965,
+      "step": 4912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34087709119406e-06,
+      "loss": 0.7303,
+      "step": 4913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337008004281193e-06,
+      "loss": 0.7553,
+      "step": 4914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333139550584228e-06,
+      "loss": 0.645,
+      "step": 4915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329271730771895e-06,
+      "loss": 0.76,
+      "step": 4916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325404545512815e-06,
+      "loss": 0.7277,
+      "step": 4917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321537995475509e-06,
+      "loss": 0.6833,
+      "step": 4918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317672081328377e-06,
+      "loss": 0.7609,
+      "step": 4919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31380680373972e-06,
+      "loss": 0.283,
+      "step": 4920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309942163377709e-06,
+      "loss": 0.6941,
+      "step": 4921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306078160910436e-06,
+      "loss": 0.7484,
+      "step": 4922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302214797005852e-06,
+      "loss": 0.6971,
+      "step": 4923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298352072331817e-06,
+      "loss": 0.6892,
+      "step": 4924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294489987556072e-06,
+      "loss": 0.7396,
+      "step": 4925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290628543346253e-06,
+      "loss": 0.7592,
+      "step": 4926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286767740369875e-06,
+      "loss": 0.7657,
+      "step": 4927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282907579294355e-06,
+      "loss": 0.2643,
+      "step": 4928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279048060786998e-06,
+      "loss": 0.742,
+      "step": 4929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275189185514981e-06,
+      "loss": 0.697,
+      "step": 4930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271330954145391e-06,
+      "loss": 0.697,
+      "step": 4931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2674733673451915e-06,
+      "loss": 0.7181,
+      "step": 4932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263616425781238e-06,
+      "loss": 0.781,
+      "step": 4933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259760130120273e-06,
+      "loss": 0.7508,
+      "step": 4934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255904481028935e-06,
+      "loss": 0.7595,
+      "step": 4935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252049479173734e-06,
+      "loss": 0.7503,
+      "step": 4936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248195125221086e-06,
+      "loss": 0.683,
+      "step": 4937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244341419837281e-06,
+      "loss": 0.7523,
+      "step": 4938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240488363688511e-06,
+      "loss": 0.638,
+      "step": 4939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23663595744084e-06,
+      "loss": 0.7047,
+      "step": 4940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232784201760233e-06,
+      "loss": 0.7361,
+      "step": 4941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22893309731254e-06,
+      "loss": 0.699,
+      "step": 4942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2250826447634825e-06,
+      "loss": 0.7284,
+      "step": 4943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221232844778701e-06,
+      "loss": 0.7662,
+      "step": 4944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217383698023689e-06,
+      "loss": 0.6649,
+      "step": 4945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2135352051638505e-06,
+      "loss": 0.729,
+      "step": 4946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209687366864465e-06,
+      "loss": 0.2774,
+      "step": 4947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20584018379071e-06,
+      "loss": 0.6964,
+      "step": 4948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201993656607629e-06,
+      "loss": 0.7336,
+      "step": 4949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19814778598018e-06,
+      "loss": 0.6544,
+      "step": 4950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19430257257318e-06,
+      "loss": 0.2528,
+      "step": 4951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1904580170513504e-06,
+      "loss": 0.7372,
+      "step": 4952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186614120079298e-06,
+      "loss": 0.7018,
+      "step": 4953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182770882321506e-06,
+      "loss": 0.8026,
+      "step": 4954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178928304442352e-06,
+      "loss": 0.7233,
+      "step": 4955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175086387106094e-06,
+      "loss": 0.699,
+      "step": 4956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171245130976884e-06,
+      "loss": 0.7349,
+      "step": 4957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1674045367187455e-06,
+      "loss": 0.725,
+      "step": 4958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163564604995602e-06,
+      "loss": 0.7428,
+      "step": 4959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159725336471255e-06,
+      "loss": 0.7456,
+      "step": 4960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155886731809395e-06,
+      "loss": 0.2885,
+      "step": 4961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152048791673593e-06,
+      "loss": 0.7373,
+      "step": 4962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14821151672731e-06,
+      "loss": 0.6643,
+      "step": 4963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144374907633891e-06,
+      "loss": 0.7555,
+      "step": 4964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140538965056557e-06,
+      "loss": 0.7509,
+      "step": 4965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136703689658435e-06,
+      "loss": 0.7343,
+      "step": 4966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132869082102513e-06,
+      "loss": 0.7435,
+      "step": 4967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129035143051676e-06,
+      "loss": 0.7319,
+      "step": 4968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125201873168691e-06,
+      "loss": 0.7952,
+      "step": 4969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121369273116216e-06,
+      "loss": 0.6649,
+      "step": 4970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117537343556774e-06,
+      "loss": 0.724,
+      "step": 4971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1137060851528e-06,
+      "loss": 0.7461,
+      "step": 4972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109875498566584e-06,
+      "loss": 0.6863,
+      "step": 4973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106045584460321e-06,
+      "loss": 0.7789,
+      "step": 4974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102216343496081e-06,
+      "loss": 0.7586,
+      "step": 4975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.09838777633582e-06,
+      "loss": 0.7417,
+      "step": 4976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094559883641378e-06,
+      "loss": 0.7114,
+      "step": 4977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0907326660744725e-06,
+      "loss": 0.6584,
+      "step": 4978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086906124296714e-06,
+      "loss": 0.6677,
+      "step": 4979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083080258969587e-06,
+      "loss": 0.7532,
+      "step": 4980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079255070754464e-06,
+      "loss": 0.738,
+      "step": 4981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075430560312599e-06,
+      "loss": 0.8093,
+      "step": 4982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071606728305133e-06,
+      "loss": 0.692,
+      "step": 4983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06778357539308e-06,
+      "loss": 0.7726,
+      "step": 4984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063961102237351e-06,
+      "loss": 0.7398,
+      "step": 4985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060139309498723e-06,
+      "loss": 0.7069,
+      "step": 4986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056318197837863e-06,
+      "loss": 0.7028,
+      "step": 4987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052497767915332e-06,
+      "loss": 0.645,
+      "step": 4988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04867802039155e-06,
+      "loss": 0.7294,
+      "step": 4989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044858955926838e-06,
+      "loss": 0.6671,
+      "step": 4990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041040575181386e-06,
+      "loss": 0.8027,
+      "step": 4991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037222878815281e-06,
+      "loss": 0.7271,
+      "step": 4992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0334058674884686e-06,
+      "loss": 0.6977,
+      "step": 4993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029589541860806e-06,
+      "loss": 0.7619,
+      "step": 4994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025773902592002e-06,
+      "loss": 0.7297,
+      "step": 4995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0219589503416665e-06,
+      "loss": 0.7305,
+      "step": 4996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018144685769284e-06,
+      "loss": 0.7423,
+      "step": 4997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01433110953422e-06,
+      "loss": 0.6961,
+      "step": 4998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0105182222957235e-06,
+      "loss": 0.7053,
+      "step": 4999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006706024712921e-06,
+      "loss": 0.6662,
+      "step": 5000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002894517444824e-06,
+      "loss": 0.6887,
+      "step": 5001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999083701150315e-06,
+      "loss": 0.6516,
+      "step": 5002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995273576488172e-06,
+      "loss": 0.7161,
+      "step": 5003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991464144117039e-06,
+      "loss": 0.6763,
+      "step": 5004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987655404695454e-06,
+      "loss": 0.2562,
+      "step": 5005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98384735888182e-06,
+      "loss": 0.6792,
+      "step": 5006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980040007334437e-06,
+      "loss": 0.7175,
+      "step": 5007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976233350711467e-06,
+      "loss": 0.6531,
+      "step": 5008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972427389670965e-06,
+      "loss": 0.8029,
+      "step": 5009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968622124870864e-06,
+      "loss": 0.7033,
+      "step": 5010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.964817556968972e-06,
+      "loss": 0.6088,
+      "step": 5011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961013686622979e-06,
+      "loss": 0.7447,
+      "step": 5012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957210514490454e-06,
+      "loss": 0.7016,
+      "step": 5013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95340804122885e-06,
+      "loss": 0.7697,
+      "step": 5014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949606267495487e-06,
+      "loss": 0.7415,
+      "step": 5015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945805193947578e-06,
+      "loss": 0.7098,
+      "step": 5016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9420048212422045e-06,
+      "loss": 0.8095,
+      "step": 5017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938205150036335e-06,
+      "loss": 0.7849,
+      "step": 5018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934406180986809e-06,
+      "loss": 0.7325,
+      "step": 5019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930607914750357e-06,
+      "loss": 0.7589,
+      "step": 5020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926810351983566e-06,
+      "loss": 0.7486,
+      "step": 5021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923013493342925e-06,
+      "loss": 0.7324,
+      "step": 5022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919217339484793e-06,
+      "loss": 0.7031,
+      "step": 5023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915421891065396e-06,
+      "loss": 0.7398,
+      "step": 5024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911627148740855e-06,
+      "loss": 0.704,
+      "step": 5025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9078331131671575e-06,
+      "loss": 0.7651,
+      "step": 5026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904039785000175e-06,
+      "loss": 0.7195,
+      "step": 5027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900247164895652e-06,
+      "loss": 0.673,
+      "step": 5028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896455253509219e-06,
+      "loss": 0.8202,
+      "step": 5029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892664051496369e-06,
+      "loss": 0.7345,
+      "step": 5030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888873559512488e-06,
+      "loss": 0.7199,
+      "step": 5031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885083778212828e-06,
+      "loss": 0.7433,
+      "step": 5032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8812947082525266e-06,
+      "loss": 0.6599,
+      "step": 5033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877506350286595e-06,
+      "loss": 0.8349,
+      "step": 5034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873718704969915e-06,
+      "loss": 0.7441,
+      "step": 5035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869931772957262e-06,
+      "loss": 0.723,
+      "step": 5036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866145554903265e-06,
+      "loss": 0.6578,
+      "step": 5037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86236005146245e-06,
+      "loss": 0.6587,
+      "step": 5038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8585752632892065e-06,
+      "loss": 0.7321,
+      "step": 5039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85479119103781e-06,
+      "loss": 0.7336,
+      "step": 5040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8510078353624015e-06,
+      "loss": 0.6517,
+      "step": 5041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847225196917012e-06,
+      "loss": 0.6744,
+      "step": 5042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843443276355532e-06,
+      "loss": 0.6799,
+      "step": 5043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839662074331736e-06,
+      "loss": 0.6887,
+      "step": 5044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835881591499286e-06,
+      "loss": 0.2705,
+      "step": 5045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832101828511696e-06,
+      "loss": 0.2517,
+      "step": 5046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828322786022378e-06,
+      "loss": 0.6902,
+      "step": 5047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8245444646846e-06,
+      "loss": 0.733,
+      "step": 5048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82076686515152e-06,
+      "loss": 0.6659,
+      "step": 5049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816989988076165e-06,
+      "loss": 0.7021,
+      "step": 5050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81321383411144e-06,
+      "loss": 0.741,
+      "step": 5051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809438403910119e-06,
+      "loss": 0.7524,
+      "step": 5052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805663698124861e-06,
+      "loss": 0.7639,
+      "step": 5053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8018897174081845e-06,
+      "loss": 0.6876,
+      "step": 5054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798116462412498e-06,
+      "loss": 0.7468,
+      "step": 5055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7943439337900765e-06,
+      "loss": 0.6374,
+      "step": 5056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790572132193072e-06,
+      "loss": 0.7501,
+      "step": 5057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786801058273514e-06,
+      "loss": 0.7224,
+      "step": 5058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783030712683297e-06,
+      "loss": 0.7011,
+      "step": 5059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7792610960742e-06,
+      "loss": 0.7081,
+      "step": 5060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7754922090978594e-06,
+      "loss": 0.7789,
+      "step": 5061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771724052405812e-06,
+      "loss": 0.7057,
+      "step": 5062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767956626649444e-06,
+      "loss": 0.6343,
+      "step": 5063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764189932480031e-06,
+      "loss": 0.7648,
+      "step": 5064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760423970548707e-06,
+      "loss": 0.6654,
+      "step": 5065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756658741506499e-06,
+      "loss": 0.7616,
+      "step": 5066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752894246004281e-06,
+      "loss": 0.5976,
+      "step": 5067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7491304846928295e-06,
+      "loss": 0.5848,
+      "step": 5068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74536745822278e-06,
+      "loss": 0.7906,
+      "step": 5069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741605167244634e-06,
+      "loss": 0.7571,
+      "step": 5070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73784361240878e-06,
+      "loss": 0.7393,
+      "step": 5071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734082794365463e-06,
+      "loss": 0.6993,
+      "step": 5072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730322713764818e-06,
+      "loss": 0.7505,
+      "step": 5073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726563371256839e-06,
+      "loss": 0.7428,
+      "step": 5074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7228047674914075e-06,
+      "loss": 0.7466,
+      "step": 5075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719046903118255e-06,
+      "loss": 0.2926,
+      "step": 5076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715289778787009e-06,
+      "loss": 0.7328,
+      "step": 5077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711533395147149e-06,
+      "loss": 0.6914,
+      "step": 5078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707777752848038e-06,
+      "loss": 0.721,
+      "step": 5079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70402285253891e-06,
+      "loss": 0.6986,
+      "step": 5080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7002686948688665e-06,
+      "loss": 0.7143,
+      "step": 5081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6965152804868914e-06,
+      "loss": 0.6969,
+      "step": 5082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69276261004182e-06,
+      "loss": 0.6825,
+      "step": 5083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689010684182381e-06,
+      "loss": 0.6838,
+      "step": 5084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6852595035571535e-06,
+      "loss": 0.2388,
+      "step": 5085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.681509068814606e-06,
+      "loss": 0.7395,
+      "step": 5086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677759380603066e-06,
+      "loss": 0.7212,
+      "step": 5087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674010439570745e-06,
+      "loss": 0.7555,
+      "step": 5088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670262246365705e-06,
+      "loss": 0.7045,
+      "step": 5089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666514801635899e-06,
+      "loss": 0.7558,
+      "step": 5090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662768106029142e-06,
+      "loss": 0.7229,
+      "step": 5091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659022160193113e-06,
+      "loss": 0.7354,
+      "step": 5092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655276964775371e-06,
+      "loss": 0.7058,
+      "step": 5093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651532520423345e-06,
+      "loss": 0.7451,
+      "step": 5094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647788827784333e-06,
+      "loss": 0.7786,
+      "step": 5095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644045887505494e-06,
+      "loss": 0.6982,
+      "step": 5096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640303700233874e-06,
+      "loss": 0.7269,
+      "step": 5097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636562266616369e-06,
+      "loss": 0.689,
+      "step": 5098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632821587299759e-06,
+      "loss": 0.7009,
+      "step": 5099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6290816629306925e-06,
+      "loss": 0.6871,
+      "step": 5100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625342494155686e-06,
+      "loss": 0.7246,
+      "step": 5101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621604081621114e-06,
+      "loss": 0.7522,
+      "step": 5102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61786642597324e-06,
+      "loss": 0.7694,
+      "step": 5103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614129527858188e-06,
+      "loss": 0.7606,
+      "step": 5104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610393387921936e-06,
+      "loss": 0.7599,
+      "step": 5105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606658006810362e-06,
+      "loss": 0.7266,
+      "step": 5106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602923385169185e-06,
+      "loss": 0.7064,
+      "step": 5107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599189523644011e-06,
+      "loss": 0.7437,
+      "step": 5108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595456422880299e-06,
+      "loss": 0.7145,
+      "step": 5109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.591724083523391e-06,
+      "loss": 0.7351,
+      "step": 5110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5879925062184805e-06,
+      "loss": 0.6954,
+      "step": 5111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584261691610654e-06,
+      "loss": 0.7405,
+      "step": 5112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580531640344842e-06,
+      "loss": 0.6977,
+      "step": 5113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576802353065855e-06,
+      "loss": 0.6997,
+      "step": 5114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573073830418376e-06,
+      "loss": 0.7625,
+      "step": 5115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569346073046937e-06,
+      "loss": 0.7065,
+      "step": 5116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565619081595957e-06,
+      "loss": 0.7016,
+      "step": 5117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561892856709714e-06,
+      "loss": 0.7449,
+      "step": 5118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55816739903236e-06,
+      "loss": 0.69,
+      "step": 5119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554442709207899e-06,
+      "loss": 0.7417,
+      "step": 5120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550718787880222e-06,
+      "loss": 0.6767,
+      "step": 5121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5469956356930686e-06,
+      "loss": 0.7003,
+      "step": 5122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543273253290062e-06,
+      "loss": 0.7417,
+      "step": 5123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53955164131468e-06,
+      "loss": 0.7255,
+      "step": 5124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535830800410276e-06,
+      "loss": 0.7518,
+      "step": 5125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5321107312200685e-06,
+      "loss": 0.7102,
+      "step": 5126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52839143438713e-06,
+      "loss": 0.6939,
+      "step": 5127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524672910554423e-06,
+      "loss": 0.7544,
+      "step": 5128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52095516036475e-06,
+      "loss": 0.7082,
+      "step": 5129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5172381844608e-06,
+      "loss": 0.2637,
+      "step": 5130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513521983485119e-06,
+      "loss": 0.7638,
+      "step": 5131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509806558080126e-06,
+      "loss": 0.7841,
+      "step": 5132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506091908888094e-06,
+      "loss": 0.7662,
+      "step": 5133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502378036551176e-06,
+      "loss": 0.7573,
+      "step": 5134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498664941711373e-06,
+      "loss": 0.7667,
+      "step": 5135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494952625010571e-06,
+      "loss": 0.6958,
+      "step": 5136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4912410870905085e-06,
+      "loss": 0.7332,
+      "step": 5137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487530328592796e-06,
+      "loss": 0.7021,
+      "step": 5138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483820350158908e-06,
+      "loss": 0.7738,
+      "step": 5139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4801111524301785e-06,
+      "loss": 0.7628,
+      "step": 5140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476402736047817e-06,
+      "loss": 0.7025,
+      "step": 5141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472695101652882e-06,
+      "loss": 0.8112,
+      "step": 5142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468988249886314e-06,
+      "loss": 0.6993,
+      "step": 5143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465282181388908e-06,
+      "loss": 0.8066,
+      "step": 5144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4615768968013325e-06,
+      "loss": 0.7538,
+      "step": 5145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457872396764106e-06,
+      "loss": 0.7489,
+      "step": 5146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454168681917627e-06,
+      "loss": 0.7988,
+      "step": 5147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4504657529021435e-06,
+      "loss": 0.6861,
+      "step": 5148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446763610357779e-06,
+      "loss": 0.725,
+      "step": 5149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443062254924518e-06,
+      "loss": 0.6669,
+      "step": 5150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4393616872422095e-06,
+      "loss": 0.7363,
+      "step": 5151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.435661907950567e-06,
+      "loss": 0.7021,
+      "step": 5152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431962917689158e-06,
+      "loss": 0.6486,
+      "step": 5153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428264717097429e-06,
+      "loss": 0.7398,
+      "step": 5154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424567306814674e-06,
+      "loss": 0.6558,
+      "step": 5155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420870687480065e-06,
+      "loss": 0.6722,
+      "step": 5156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41717485973263e-06,
+      "loss": 0.6519,
+      "step": 5157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413479824211265e-06,
+      "loss": 0.725,
+      "step": 5158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409785581554716e-06,
+      "loss": 0.7114,
+      "step": 5159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4060921324016084e-06,
+      "loss": 0.7307,
+      "step": 5160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402399477390423e-06,
+      "loss": 0.7104,
+      "step": 5161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398707617159493e-06,
+      "loss": 0.7097,
+      "step": 5162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395016552347043e-06,
+      "loss": 0.7111,
+      "step": 5163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3913262835911275e-06,
+      "loss": 0.2653,
+      "step": 5164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387636811529686e-06,
+      "loss": 0.6358,
+      "step": 5165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383948136800505e-06,
+      "loss": 0.7599,
+      "step": 5166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380260260041242e-06,
+      "loss": 0.748,
+      "step": 5167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376573181889417e-06,
+      "loss": 0.724,
+      "step": 5168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3728869029824125e-06,
+      "loss": 0.7423,
+      "step": 5169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369201423957462e-06,
+      "loss": 0.2668,
+      "step": 5170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365516745451673e-06,
+      "loss": 0.7221,
+      "step": 5171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361832868102013e-06,
+      "loss": 0.7123,
+      "step": 5172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358149792545303e-06,
+      "loss": 0.7242,
+      "step": 5173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354467519418232e-06,
+      "loss": 0.7561,
+      "step": 5174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350786049357351e-06,
+      "loss": 0.73,
+      "step": 5175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347105382999074e-06,
+      "loss": 0.7478,
+      "step": 5176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343425520979661e-06,
+      "loss": 0.6907,
+      "step": 5177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339746463935257e-06,
+      "loss": 0.7505,
+      "step": 5178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336068212501844e-06,
+      "loss": 0.7679,
+      "step": 5179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332390767315281e-06,
+      "loss": 0.6829,
+      "step": 5180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328714129011282e-06,
+      "loss": 0.6256,
+      "step": 5181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325038298225426e-06,
+      "loss": 0.6272,
+      "step": 5182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32136327559314e-06,
+      "loss": 0.7105,
+      "step": 5183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317689061749723e-06,
+      "loss": 0.7138,
+      "step": 5184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314015657330337e-06,
+      "loss": 0.731,
+      "step": 5185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310343062969988e-06,
+      "loss": 0.7337,
+      "step": 5186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306671279303557e-06,
+      "loss": 0.7575,
+      "step": 5187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303000306965778e-06,
+      "loss": 0.657,
+      "step": 5188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299330146591252e-06,
+      "loss": 0.7338,
+      "step": 5189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295660798814425e-06,
+      "loss": 0.7009,
+      "step": 5190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291992264269621e-06,
+      "loss": 0.7471,
+      "step": 5191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288324543591005e-06,
+      "loss": 0.6958,
+      "step": 5192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284657637412613e-06,
+      "loss": 0.7077,
+      "step": 5193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280991546368341e-06,
+      "loss": 0.7606,
+      "step": 5194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277326271091939e-06,
+      "loss": 0.746,
+      "step": 5195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273661812217022e-06,
+      "loss": 0.7447,
+      "step": 5196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2699981703770485e-06,
+      "loss": 0.6922,
+      "step": 5197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266335346205358e-06,
+      "loss": 0.7493,
+      "step": 5198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262673340335129e-06,
+      "loss": 0.7097,
+      "step": 5199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25901215339941e-06,
+      "loss": 0.7157,
+      "step": 5200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2553517860311055e-06,
+      "loss": 0.757,
+      "step": 5201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251692238862981e-06,
+      "loss": 0.7368,
+      "step": 5202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248033512527649e-06,
+      "loss": 0.7312,
+      "step": 5203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244375607657597e-06,
+      "loss": 0.7594,
+      "step": 5204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240718524885153e-06,
+      "loss": 0.7148,
+      "step": 5205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23706226484251e-06,
+      "loss": 0.7623,
+      "step": 5206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2334068281617345e-06,
+      "loss": 0.6406,
+      "step": 5207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2297522154747225e-06,
+      "loss": 0.7647,
+      "step": 5208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226098427413251e-06,
+      "loss": 0.7018,
+      "step": 5209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222445464608934e-06,
+      "loss": 0.7244,
+      "step": 5210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218793327693266e-06,
+      "loss": 0.2813,
+      "step": 5211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21514201729757e-06,
+      "loss": 0.6492,
+      "step": 5212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2114915340530615e-06,
+      "loss": 0.7542,
+      "step": 5213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20784187859078e-06,
+      "loss": 0.6601,
+      "step": 5214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204193051541646e-06,
+      "loss": 0.6837,
+      "step": 5215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200545053536417e-06,
+      "loss": 0.7858,
+      "step": 5216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1968978852057265e-06,
+      "loss": 0.7147,
+      "step": 5217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193251547180041e-06,
+      "loss": 0.635,
+      "step": 5218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.189606040089711e-06,
+      "loss": 0.6872,
+      "step": 5219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185961364564929e-06,
+      "loss": 0.7471,
+      "step": 5220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182317521235737e-06,
+      "loss": 0.7509,
+      "step": 5221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178674510732049e-06,
+      "loss": 0.6453,
+      "step": 5222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175032333683617e-06,
+      "loss": 0.7073,
+      "step": 5223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171390990720064e-06,
+      "loss": 0.727,
+      "step": 5224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167750482470866e-06,
+      "loss": 0.7907,
+      "step": 5225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16411080956535e-06,
+      "loss": 0.733,
+      "step": 5226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160471972632699e-06,
+      "loss": 0.7425,
+      "step": 5227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.156833972301957e-06,
+      "loss": 0.6697,
+      "step": 5228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153196809202015e-06,
+      "loss": 0.7763,
+      "step": 5229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149560483961624e-06,
+      "loss": 0.7407,
+      "step": 5230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1459249972093926e-06,
+      "loss": 0.7448,
+      "step": 5231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142290349573782e-06,
+      "loss": 0.7706,
+      "step": 5232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138656541683113e-06,
+      "loss": 0.6499,
+      "step": 5233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135023574165545e-06,
+      "loss": 0.689,
+      "step": 5234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131391447649114e-06,
+      "loss": 0.7434,
+      "step": 5235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127760162761694e-06,
+      "loss": 0.7706,
+      "step": 5236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124129720131019e-06,
+      "loss": 0.7796,
+      "step": 5237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120500120384684e-06,
+      "loss": 0.7341,
+      "step": 5238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1168713641501315e-06,
+      "loss": 0.7559,
+      "step": 5239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113243452054652e-06,
+      "loss": 0.6668,
+      "step": 5240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109616384725404e-06,
+      "loss": 0.783,
+      "step": 5241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1059901627893945e-06,
+      "loss": 0.7594,
+      "step": 5242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102364786873477e-06,
+      "loss": 0.2664,
+      "step": 5243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098740257604366e-06,
+      "loss": 0.6744,
+      "step": 5244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095116575608631e-06,
+      "loss": 0.7084,
+      "step": 5245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091493741512694e-06,
+      "loss": 0.6912,
+      "step": 5246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087871755942824e-06,
+      "loss": 0.687,
+      "step": 5247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084250619525154e-06,
+      "loss": 0.6614,
+      "step": 5248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080630332885656e-06,
+      "loss": 0.7347,
+      "step": 5249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077010896650169e-06,
+      "loss": 0.2669,
+      "step": 5250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07339231144438e-06,
+      "loss": 0.7346,
+      "step": 5251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069774577893829e-06,
+      "loss": 0.6975,
+      "step": 5252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066157696623903e-06,
+      "loss": 0.78,
+      "step": 5253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.06254166825985e-06,
+      "loss": 0.6852,
+      "step": 5254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058926493426773e-06,
+      "loss": 0.712,
+      "step": 5255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055312172749612e-06,
+      "loss": 0.6231,
+      "step": 5256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051698706853172e-06,
+      "loss": 0.7978,
+      "step": 5257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.04808609636211e-06,
+      "loss": 0.6975,
+      "step": 5258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044474341900935e-06,
+      "loss": 0.265,
+      "step": 5259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040863444093997e-06,
+      "loss": 0.7089,
+      "step": 5260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037253403565517e-06,
+      "loss": 0.7071,
+      "step": 5261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0336442209395445e-06,
+      "loss": 0.7087,
+      "step": 5262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030035896840007e-06,
+      "loss": 0.7295,
+      "step": 5263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0264284318906596e-06,
+      "loss": 0.6766,
+      "step": 5264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022821826715123e-06,
+      "loss": 0.7114,
+      "step": 5265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0192160819368705e-06,
+      "loss": 0.7074,
+      "step": 5266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015611198179214e-06,
+      "loss": 0.7388,
+      "step": 5267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012007176065325e-06,
+      "loss": 0.7034,
+      "step": 5268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008404016218232e-06,
+      "loss": 0.7016,
+      "step": 5269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004801719260805e-06,
+      "loss": 0.769,
+      "step": 5270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001200285815762e-06,
+      "loss": 0.6865,
+      "step": 5271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997599716505687e-06,
+      "loss": 0.7533,
+      "step": 5272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994000011952995e-06,
+      "loss": 0.7873,
+      "step": 5273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990401172779966e-06,
+      "loss": 0.627,
+      "step": 5274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9868031996087275e-06,
+      "loss": 0.7363,
+      "step": 5275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983206093061252e-06,
+      "loss": 0.6681,
+      "step": 5276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979609853759374e-06,
+      "loss": 0.7166,
+      "step": 5277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976014482324759e-06,
+      "loss": 0.7358,
+      "step": 5278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972419979378944e-06,
+      "loss": 0.7119,
+      "step": 5279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.968826345543295e-06,
+      "loss": 0.6973,
+      "step": 5280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965233581439043e-06,
+      "loss": 0.6699,
+      "step": 5281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961641687687263e-06,
+      "loss": 0.7991,
+      "step": 5282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958050664908886e-06,
+      "loss": 0.6843,
+      "step": 5283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954460513724676e-06,
+      "loss": 0.7282,
+      "step": 5284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950871234755268e-06,
+      "loss": 0.8058,
+      "step": 5285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947282828621125e-06,
+      "loss": 0.2366,
+      "step": 5286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943695295942574e-06,
+      "loss": 0.6913,
+      "step": 5287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9401086373397875e-06,
+      "loss": 0.7468,
+      "step": 5288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.936522853432785e-06,
+      "loss": 0.2865,
+      "step": 5289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9329379448414385e-06,
+      "loss": 0.7261,
+      "step": 5290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929353912185458e-06,
+      "loss": 0.7034,
+      "step": 5291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925770756084419e-06,
+      "loss": 0.7099,
+      "step": 5292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922188477157727e-06,
+      "loss": 0.7935,
+      "step": 5293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918607076024652e-06,
+      "loss": 0.7166,
+      "step": 5294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915026553304302e-06,
+      "loss": 0.6749,
+      "step": 5295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911446909615641e-06,
+      "loss": 0.6556,
+      "step": 5296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90786814557747e-06,
+      "loss": 0.7169,
+      "step": 5297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904290261808453e-06,
+      "loss": 0.7336,
+      "step": 5298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9007132589270835e-06,
+      "loss": 0.7118,
+      "step": 5299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897137137551717e-06,
+      "loss": 0.7799,
+      "step": 5300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893561898300553e-06,
+      "loss": 0.6923,
+      "step": 5301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889987541791636e-06,
+      "loss": 0.7438,
+      "step": 5302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886414068642868e-06,
+      "loss": 0.7703,
+      "step": 5303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8828414794719755e-06,
+      "loss": 0.7554,
+      "step": 5304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.87926977489656e-06,
+      "loss": 0.6932,
+      "step": 5305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875698955534046e-06,
+      "loss": 0.7412,
+      "step": 5306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.87212902200172e-06,
+      "loss": 0.6937,
+      "step": 5307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8685599749167126e-06,
+      "loss": 0.751,
+      "step": 5308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864991814896001e-06,
+      "loss": 0.6839,
+      "step": 5309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861424542556401e-06,
+      "loss": 0.7649,
+      "step": 5310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857858158514586e-06,
+      "loss": 0.7478,
+      "step": 5311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854292663387075e-06,
+      "loss": 0.7169,
+      "step": 5312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850728057790217e-06,
+      "loss": 0.7106,
+      "step": 5313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8471643423402374e-06,
+      "loss": 0.6839,
+      "step": 5314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843601517653177e-06,
+      "loss": 0.7877,
+      "step": 5315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840039584344945e-06,
+      "loss": 0.7286,
+      "step": 5316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836478543031279e-06,
+      "loss": 0.6839,
+      "step": 5317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832918394327778e-06,
+      "loss": 0.6984,
+      "step": 5318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829359138849869e-06,
+      "loss": 0.6898,
+      "step": 5319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82580077721285e-06,
+      "loss": 0.6808,
+      "step": 5320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822243310031839e-06,
+      "loss": 0.6502,
+      "step": 5321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818686737921813e-06,
+      "loss": 0.7617,
+      "step": 5322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815131061497594e-06,
+      "loss": 0.7323,
+      "step": 5323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811576281373841e-06,
+      "loss": 0.706,
+      "step": 5324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808022398165065e-06,
+      "loss": 0.6851,
+      "step": 5325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8044694124856225e-06,
+      "loss": 0.6857,
+      "step": 5326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800917324949715e-06,
+      "loss": 0.6948,
+      "step": 5327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797366136171381e-06,
+      "loss": 0.6684,
+      "step": 5328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793815846764515e-06,
+      "loss": 0.7135,
+      "step": 5329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790266457342841e-06,
+      "loss": 0.7096,
+      "step": 5330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786717968519945e-06,
+      "loss": 0.7557,
+      "step": 5331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783170380909244e-06,
+      "loss": 0.7621,
+      "step": 5332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779623695124012e-06,
+      "loss": 0.6625,
+      "step": 5333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776077911777348e-06,
+      "loss": 0.7242,
+      "step": 5334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772533031482213e-06,
+      "loss": 0.6968,
+      "step": 5335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768989054851407e-06,
+      "loss": 0.6597,
+      "step": 5336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765445982497565e-06,
+      "loss": 0.7336,
+      "step": 5337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7619038150331775e-06,
+      "loss": 0.6446,
+      "step": 5338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7583625530705714e-06,
+      "loss": 0.7483,
+      "step": 5339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754822197221924e-06,
+      "loss": 0.6545,
+      "step": 5340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751282748099245e-06,
+      "loss": 0.7283,
+      "step": 5341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747744206314398e-06,
+      "loss": 0.691,
+      "step": 5342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744206572479082e-06,
+      "loss": 0.7431,
+      "step": 5343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740669847204844e-06,
+      "loss": 0.7112,
+      "step": 5344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7371340311030714e-06,
+      "loss": 0.6909,
+      "step": 5345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733599124784997e-06,
+      "loss": 0.6911,
+      "step": 5346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730065128861698e-06,
+      "loss": 0.6879,
+      "step": 5347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726532043944083e-06,
+      "loss": 0.6571,
+      "step": 5348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722999870642919e-06,
+      "loss": 0.7549,
+      "step": 5349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719468609568801e-06,
+      "loss": 0.6817,
+      "step": 5350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715938261332174e-06,
+      "loss": 0.7429,
+      "step": 5351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7124088265433255e-06,
+      "loss": 0.7188,
+      "step": 5352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7088803058123876e-06,
+      "loss": 0.685,
+      "step": 5353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705352699749321e-06,
+      "loss": 0.2715,
+      "step": 5354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701826008963949e-06,
+      "loss": 0.7682,
+      "step": 5355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698300234065915e-06,
+      "loss": 0.726,
+      "step": 5356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694775375664713e-06,
+      "loss": 0.7415,
+      "step": 5357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691251434369694e-06,
+      "loss": 0.7346,
+      "step": 5358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687728410790025e-06,
+      "loss": 0.7367,
+      "step": 5359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684206305534732e-06,
+      "loss": 0.7105,
+      "step": 5360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680685119212668e-06,
+      "loss": 0.6853,
+      "step": 5361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677164852432544e-06,
+      "loss": 0.7494,
+      "step": 5362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673645505802892e-06,
+      "loss": 0.6777,
+      "step": 5363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670127079932111e-06,
+      "loss": 0.7128,
+      "step": 5364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666609575428413e-06,
+      "loss": 0.702,
+      "step": 5365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6630929928998745e-06,
+      "loss": 0.7958,
+      "step": 5366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659577332954392e-06,
+      "loss": 0.6887,
+      "step": 5367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656062596199716e-06,
+      "loss": 0.7211,
+      "step": 5368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652548783243433e-06,
+      "loss": 0.7535,
+      "step": 5369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649035894692974e-06,
+      "loss": 0.7271,
+      "step": 5370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645523931155608e-06,
+      "loss": 0.7588,
+      "step": 5371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6420128932384345e-06,
+      "loss": 0.6807,
+      "step": 5372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638502781548411e-06,
+      "loss": 0.7227,
+      "step": 5373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634993596692315e-06,
+      "loss": 0.7609,
+      "step": 5374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6314853392767815e-06,
+      "loss": 0.7163,
+      "step": 5375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627978009908274e-06,
+      "loss": 0.7137,
+      "step": 5376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624471609193105e-06,
+      "loss": 0.7351,
+      "step": 5377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620966137737414e-06,
+      "loss": 0.6813,
+      "step": 5378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617461596147192e-06,
+      "loss": 0.7716,
+      "step": 5379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613957985028256e-06,
+      "loss": 0.6883,
+      "step": 5380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610455304986276e-06,
+      "loss": 0.7022,
+      "step": 5381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606953556626753e-06,
+      "loss": 0.7014,
+      "step": 5382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603452740555031e-06,
+      "loss": 0.7281,
+      "step": 5383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599952857376293e-06,
+      "loss": 0.794,
+      "step": 5384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59645390769555e-06,
+      "loss": 0.7284,
+      "step": 5385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592955892117671e-06,
+      "loss": 0.7309,
+      "step": 5386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5894588112473426e-06,
+      "loss": 0.7049,
+      "step": 5387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585962665689105e-06,
+      "loss": 0.6767,
+      "step": 5388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582467456047329e-06,
+      "loss": 0.7402,
+      "step": 5389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578973182926233e-06,
+      "loss": 0.6603,
+      "step": 5390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575479846929858e-06,
+      "loss": 0.6703,
+      "step": 5391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571987448662094e-06,
+      "loss": 0.7363,
+      "step": 5392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568495988726674e-06,
+      "loss": 0.7199,
+      "step": 5393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565005467727151e-06,
+      "loss": 0.7199,
+      "step": 5394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56151588626693e-06,
+      "loss": 0.6386,
+      "step": 5395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5580272449492516e-06,
+      "loss": 0.7166,
+      "step": 5396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5545395443771936e-06,
+      "loss": 0.7184,
+      "step": 5397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5510527851536624e-06,
+      "loss": 0.6873,
+      "step": 5398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547566967881417e-06,
+      "loss": 0.7099,
+      "step": 5399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544082093163038e-06,
+      "loss": 0.6506,
+      "step": 5400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540598161600954e-06,
+      "loss": 0.6858,
+      "step": 5401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537115173797427e-06,
+      "loss": 0.695,
+      "step": 5402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533633130354556e-06,
+      "loss": 0.7363,
+      "step": 5403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530152031874279e-06,
+      "loss": 0.7393,
+      "step": 5404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526671878958362e-06,
+      "loss": 0.7423,
+      "step": 5405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523192672208421e-06,
+      "loss": 0.7474,
+      "step": 5406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519714412225894e-06,
+      "loss": 0.6726,
+      "step": 5407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516237099612065e-06,
+      "loss": 0.7485,
+      "step": 5408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512760734968055e-06,
+      "loss": 0.7358,
+      "step": 5409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509285318894817e-06,
+      "loss": 0.7284,
+      "step": 5410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5058108519931374e-06,
+      "loss": 0.7591,
+      "step": 5411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502337334863648e-06,
+      "loss": 0.7028,
+      "step": 5412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4988647681068036e-06,
+      "loss": 0.7519,
+      "step": 5413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495393152322903e-06,
+      "loss": 0.7179,
+      "step": 5414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491922488112082e-06,
+      "loss": 0.7405,
+      "step": 5415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488452776074308e-06,
+      "loss": 0.6826,
+      "step": 5416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484984016809388e-06,
+      "loss": 0.7053,
+      "step": 5417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481516210916957e-06,
+      "loss": 0.6319,
+      "step": 5418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478049358996491e-06,
+      "loss": 0.701,
+      "step": 5419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474583461647293e-06,
+      "loss": 0.702,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47111851946852e-06,
+      "loss": 0.2923,
+      "step": 5421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467654533059141e-06,
+      "loss": 0.6435,
+      "step": 5422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464191503017978e-06,
+      "loss": 0.6446,
+      "step": 5423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460729429943672e-06,
+      "loss": 0.7155,
+      "step": 5424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4572683144347106e-06,
+      "loss": 0.7457,
+      "step": 5425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45380815708941e-06,
+      "loss": 0.7166,
+      "step": 5426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450348958505924e-06,
+      "loss": 0.6867,
+      "step": 5427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446890719282241e-06,
+      "loss": 0.729,
+      "step": 5428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443433440016176e-06,
+      "loss": 0.6633,
+      "step": 5429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43997712130539e-06,
+      "loss": 0.2676,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436521763747364e-06,
+      "loss": 0.6931,
+      "step": 5431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4330673679394255e-06,
+      "loss": 0.6021,
+      "step": 5432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429613934478729e-06,
+      "loss": 0.7297,
+      "step": 5433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42616146396227e-06,
+      "loss": 0.7658,
+      "step": 5434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422709956986863e-06,
+      "loss": 0.719,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4192594141491715e-06,
+      "loss": 0.6537,
+      "step": 5436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41580983604568e-06,
+      "loss": 0.6607,
+      "step": 5437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412361223272715e-06,
+      "loss": 0.726,
+      "step": 5438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408913576426432e-06,
+      "loss": 0.5975,
+      "step": 5439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405466896102823e-06,
+      "loss": 0.683,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402021182897711e-06,
+      "loss": 0.7894,
+      "step": 5441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398576437406747e-06,
+      "loss": 0.6542,
+      "step": 5442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395132660225423e-06,
+      "loss": 0.7099,
+      "step": 5443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391689851949053e-06,
+      "loss": 0.6585,
+      "step": 5444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388248013172797e-06,
+      "loss": 0.6393,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384807144491636e-06,
+      "loss": 0.7456,
+      "step": 5446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381367246500394e-06,
+      "loss": 0.7427,
+      "step": 5447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377928319793714e-06,
+      "loss": 0.2686,
+      "step": 5448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374490364966084e-06,
+      "loss": 0.7281,
+      "step": 5449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371053382611812e-06,
+      "loss": 0.7089,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367617373325046e-06,
+      "loss": 0.6802,
+      "step": 5451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364182337699765e-06,
+      "loss": 0.71,
+      "step": 5452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360748276329777e-06,
+      "loss": 0.7198,
+      "step": 5453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357315189808728e-06,
+      "loss": 0.7152,
+      "step": 5454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3538830787300846e-06,
+      "loss": 0.7352,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3504519436871554e-06,
+      "loss": 0.684,
+      "step": 5456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347021785273068e-06,
+      "loss": 0.6348,
+      "step": 5457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343592604080795e-06,
+      "loss": 0.7395,
+      "step": 5458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3401644007031305e-06,
+      "loss": 0.2586,
+      "step": 5459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33673717573271e-06,
+      "loss": 0.7226,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333310929761982e-06,
+      "loss": 0.7321,
+      "step": 5461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3298856633832435e-06,
+      "loss": 0.5965,
+      "step": 5462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326461377188616e-06,
+      "loss": 0.7062,
+      "step": 5463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3230380717700416e-06,
+      "loss": 0.7253,
+      "step": 5464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3196157477193145e-06,
+      "loss": 0.7298,
+      "step": 5465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316194405628038e-06,
+      "loss": 0.6848,
+      "step": 5466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31277404608766e-06,
+      "loss": 0.6996,
+      "step": 5467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309354669689446e-06,
+      "loss": 0.7293,
+      "step": 5468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305936277024506e-06,
+      "loss": 0.7625,
+      "step": 5469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302518868683762e-06,
+      "loss": 0.6278,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29910244525799e-06,
+      "loss": 0.732,
+      "step": 5471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295687007337771e-06,
+      "loss": 0.7049,
+      "step": 5472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29227255551353e-06,
+      "loss": 0.7093,
+      "step": 5473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288859090375524e-06,
+      "loss": 0.7123,
+      "step": 5474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285446612513823e-06,
+      "loss": 0.7603,
+      "step": 5475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282035122518342e-06,
+      "loss": 0.716,
+      "step": 5476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278624620978821e-06,
+      "loss": 0.7156,
+      "step": 5477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275215108484829e-06,
+      "loss": 0.7914,
+      "step": 5478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271806585625761e-06,
+      "loss": 0.7294,
+      "step": 5479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268399052990846e-06,
+      "loss": 0.7042,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2649925111691324e-06,
+      "loss": 0.7902,
+      "step": 5481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26158696074951e-06,
+      "loss": 0.7331,
+      "step": 5482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258182402320689e-06,
+      "loss": 0.6727,
+      "step": 5483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2547788364712106e-06,
+      "loss": 0.7072,
+      "step": 5484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251376263789449e-06,
+      "loss": 0.7839,
+      "step": 5485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247974684863593e-06,
+      "loss": 0.7939,
+      "step": 5486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244574100281676e-06,
+      "loss": 0.6896,
+      "step": 5487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241174510631546e-06,
+      "loss": 0.677,
+      "step": 5488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237775916500886e-06,
+      "loss": 0.7054,
+      "step": 5489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234378318477208e-06,
+      "loss": 0.7053,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230981717147853e-06,
+      "loss": 0.7501,
+      "step": 5491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227586113099979e-06,
+      "loss": 0.626,
+      "step": 5492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224191506920584e-06,
+      "loss": 0.6852,
+      "step": 5493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220797899196485e-06,
+      "loss": 0.6947,
+      "step": 5494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21740529051433e-06,
+      "loss": 0.732,
+      "step": 5495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214013681460597e-06,
+      "loss": 0.6135,
+      "step": 5496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2106230726215865e-06,
+      "loss": 0.6777,
+      "step": 5497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2072334645834325e-06,
+      "loss": 0.6999,
+      "step": 5498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203844857932083e-06,
+      "loss": 0.692,
+      "step": 5499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20045725325333e-06,
+      "loss": 0.6921,
+      "step": 5500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197070651132776e-06,
+      "loss": 0.7562,
+      "step": 5501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193685052155861e-06,
+      "loss": 0.8037,
+      "step": 5502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190300456907848e-06,
+      "loss": 0.7122,
+      "step": 5503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186916865973831e-06,
+      "loss": 0.7506,
+      "step": 5504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183534279938719e-06,
+      "loss": 0.6698,
+      "step": 5505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1801526993872625e-06,
+      "loss": 0.6884,
+      "step": 5506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17677212490402e-06,
+      "loss": 0.6692,
+      "step": 5507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173392557073394e-06,
+      "loss": 0.6853,
+      "step": 5508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170013996479605e-06,
+      "loss": 0.7175,
+      "step": 5509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166636443706697e-06,
+      "loss": 0.625,
+      "step": 5510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1632598993385454e-06,
+      "loss": 0.7249,
+      "step": 5511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159884363958845e-06,
+      "loss": 0.5379,
+      "step": 5512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156509838151124e-06,
+      "loss": 0.7667,
+      "step": 5513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153136322498726e-06,
+      "loss": 0.644,
+      "step": 5514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149763817584826e-06,
+      "loss": 0.7029,
+      "step": 5515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146392323992428e-06,
+      "loss": 0.6653,
+      "step": 5516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143021842304358e-06,
+      "loss": 0.6736,
+      "step": 5517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139652373103259e-06,
+      "loss": 0.6721,
+      "step": 5518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1362839169716105e-06,
+      "loss": 0.7816,
+      "step": 5519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132916474491713e-06,
+      "loss": 0.6811,
+      "step": 5520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129550046245689e-06,
+      "loss": 0.7554,
+      "step": 5521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126184632815494e-06,
+      "loss": 0.7249,
+      "step": 5522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122820234782893e-06,
+      "loss": 0.7054,
+      "step": 5523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119456852729492e-06,
+      "loss": 0.6497,
+      "step": 5524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116094487236707e-06,
+      "loss": 0.7427,
+      "step": 5525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112733138885788e-06,
+      "loss": 0.6379,
+      "step": 5526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109372808257807e-06,
+      "loss": 0.7097,
+      "step": 5527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106013495933662e-06,
+      "loss": 0.7357,
+      "step": 5528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1026552024940655e-06,
+      "loss": 0.6349,
+      "step": 5529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099297928519569e-06,
+      "loss": 0.6512,
+      "step": 5530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09594167459053e-06,
+      "loss": 0.7234,
+      "step": 5531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092586441287144e-06,
+      "loss": 0.7318,
+      "step": 5532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089232229189423e-06,
+      "loss": 0.6567,
+      "step": 5533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08587903887721e-06,
+      "loss": 0.6748,
+      "step": 5534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082526870930164e-06,
+      "loss": 0.2798,
+      "step": 5535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0791757259277634e-06,
+      "loss": 0.6503,
+      "step": 5536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075825604449324e-06,
+      "loss": 0.7089,
+      "step": 5537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072476507073968e-06,
+      "loss": 0.7281,
+      "step": 5538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0691284343806525e-06,
+      "loss": 0.746,
+      "step": 5539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065781386948154e-06,
+      "loss": 0.7115,
+      "step": 5540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062435365355076e-06,
+      "loss": 0.7602,
+      "step": 5541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059090370179832e-06,
+      "loss": 0.7443,
+      "step": 5542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0557464020006695e-06,
+      "loss": 0.6994,
+      "step": 5543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052403461395659e-06,
+      "loss": 0.7221,
+      "step": 5544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049061548942685e-06,
+      "loss": 0.7094,
+      "step": 5545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045720665219458e-06,
+      "loss": 0.7073,
+      "step": 5546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042380810803514e-06,
+      "loss": 0.7121,
+      "step": 5547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039041986272213e-06,
+      "loss": 0.7547,
+      "step": 5548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035704192202724e-06,
+      "loss": 0.7518,
+      "step": 5549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032367429172054e-06,
+      "loss": 0.7369,
+      "step": 5550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029031697757016e-06,
+      "loss": 0.6955,
+      "step": 5551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025696998534259e-06,
+      "loss": 0.7987,
+      "step": 5552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022363332080245e-06,
+      "loss": 0.7286,
+      "step": 5553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0190306989712615e-06,
+      "loss": 0.7211,
+      "step": 5554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015699099783416e-06,
+      "loss": 0.6775,
+      "step": 5555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0123685350926356e-06,
+      "loss": 0.673,
+      "step": 5556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009039005474672e-06,
+      "loss": 0.7188,
+      "step": 5557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005710511505091e-06,
+      "loss": 0.7016,
+      "step": 5558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002383053759288e-06,
+      "loss": 0.7423,
+      "step": 5559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9990566328124755e-06,
+      "loss": 0.7136,
+      "step": 5560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995731249239689e-06,
+      "loss": 0.678,
+      "step": 5561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992406903615779e-06,
+      "loss": 0.789,
+      "step": 5562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989083596515423e-06,
+      "loss": 0.7296,
+      "step": 5563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985761328513112e-06,
+      "loss": 0.7486,
+      "step": 5564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982440100183158e-06,
+      "loss": 0.2682,
+      "step": 5565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979119912099712e-06,
+      "loss": 0.6982,
+      "step": 5566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975800764836716e-06,
+      "loss": 0.7045,
+      "step": 5567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972482658967953e-06,
+      "loss": 0.7278,
+      "step": 5568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969165595067014e-06,
+      "loss": 0.6772,
+      "step": 5569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.965849573707321e-06,
+      "loss": 0.7299,
+      "step": 5570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962534595462099e-06,
+      "loss": 0.761,
+      "step": 5571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959220660904416e-06,
+      "loss": 0.6871,
+      "step": 5572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955907770607138e-06,
+      "loss": 0.6828,
+      "step": 5573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.952595925142966e-06,
+      "loss": 0.6683,
+      "step": 5574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949285125084406e-06,
+      "loss": 0.6633,
+      "step": 5575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945975371003799e-06,
+      "loss": 0.6538,
+      "step": 5576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942666663473286e-06,
+      "loss": 0.7173,
+      "step": 5577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939359003064848e-06,
+      "loss": 0.6457,
+      "step": 5578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936052390350279e-06,
+      "loss": 0.7062,
+      "step": 5579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9327468259011765e-06,
+      "loss": 0.7181,
+      "step": 5580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9294423102889775e-06,
+      "loss": 0.6659,
+      "step": 5581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926138844084922e-06,
+      "loss": 0.7645,
+      "step": 5582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922836427860078e-06,
+      "loss": 0.7326,
+      "step": 5583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91953506218533e-06,
+      "loss": 0.7218,
+      "step": 5584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916234747631382e-06,
+      "loss": 0.7414,
+      "step": 5585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912935484768749e-06,
+      "loss": 0.7462,
+      "step": 5586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909637274167777e-06,
+      "loss": 0.7625,
+      "step": 5587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906340116398613e-06,
+      "loss": 0.6494,
+      "step": 5588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903044012031238e-06,
+      "loss": 0.7444,
+      "step": 5589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8997489616354434e-06,
+      "loss": 0.6472,
+      "step": 5590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896454965780839e-06,
+      "loss": 0.6836,
+      "step": 5591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893162025036857e-06,
+      "loss": 0.7141,
+      "step": 5592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889870139972738e-06,
+      "loss": 0.7098,
+      "step": 5593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886579311157549e-06,
+      "loss": 0.6344,
+      "step": 5594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883289539160165e-06,
+      "loss": 0.7693,
+      "step": 5595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880000824549287e-06,
+      "loss": 0.8189,
+      "step": 5596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876713167893431e-06,
+      "loss": 0.691,
+      "step": 5597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873426569760932e-06,
+      "loss": 0.7056,
+      "step": 5598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870141030719933e-06,
+      "loss": 0.7437,
+      "step": 5599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866856551338403e-06,
+      "loss": 0.6841,
+      "step": 5600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863573132184128e-06,
+      "loss": 0.7057,
+      "step": 5601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860290773824703e-06,
+      "loss": 0.7742,
+      "step": 5602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857009476827546e-06,
+      "loss": 0.7094,
+      "step": 5603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853729241759889e-06,
+      "loss": 0.731,
+      "step": 5604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8504500691887865e-06,
+      "loss": 0.6752,
+      "step": 5605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8471719596810965e-06,
+      "loss": 0.7279,
+      "step": 5606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8438949138035075e-06,
+      "loss": 0.6946,
+      "step": 5607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840618932122512e-06,
+      "loss": 0.7852,
+      "step": 5608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837344015204425e-06,
+      "loss": 0.8117,
+      "step": 5609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8340701636153785e-06,
+      "loss": 0.7479,
+      "step": 5610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83079737792132e-06,
+      "loss": 0.7141,
+      "step": 5611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8275256586880075e-06,
+      "loss": 0.617,
+      "step": 5612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824255006481017e-06,
+      "loss": 0.709,
+      "step": 5613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820985421865748e-06,
+      "loss": 0.788,
+      "step": 5614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817716905407397e-06,
+      "loss": 0.7548,
+      "step": 5615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8144494576710013e-06,
+      "loss": 0.6913,
+      "step": 5616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81118307922139e-06,
+      "loss": 0.7093,
+      "step": 5617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807917770623223e-06,
+      "loss": 0.7206,
+      "step": 5618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804653532440963e-06,
+      "loss": 0.7691,
+      "step": 5619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801390365238901e-06,
+      "loss": 0.685,
+      "step": 5620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7981282695811252e-06,
+      "loss": 0.6745,
+      "step": 5621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7948672460315616e-06,
+      "loss": 0.7123,
+      "step": 5622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7916072951539306e-06,
+      "loss": 0.6178,
+      "step": 5623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788348417511777e-06,
+      "loss": 0.2898,
+      "step": 5624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7850906136684616e-06,
+      "loss": 0.6815,
+      "step": 5625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.781833884187148e-06,
+      "loss": 0.71,
+      "step": 5626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778578229630828e-06,
+      "loss": 0.7098,
+      "step": 5627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7753236505623012e-06,
+      "loss": 0.6735,
+      "step": 5628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7720701475441836e-06,
+      "loss": 0.7485,
+      "step": 5629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7688177211388965e-06,
+      "loss": 0.7269,
+      "step": 5630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.765566371908691e-06,
+      "loss": 0.745,
+      "step": 5631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762316100415614e-06,
+      "loss": 0.7377,
+      "step": 5632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7590669072215403e-06,
+      "loss": 0.7894,
+      "step": 5633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7558187928881506e-06,
+      "loss": 0.7483,
+      "step": 5634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7525717579769426e-06,
+      "loss": 0.7156,
+      "step": 5635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7493258030492308e-06,
+      "loss": 0.7194,
+      "step": 5636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7460809286661316e-06,
+      "loss": 0.6985,
+      "step": 5637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7428371353885862e-06,
+      "loss": 0.7321,
+      "step": 5638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7395944237773394e-06,
+      "loss": 0.6768,
+      "step": 5639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736352794392957e-06,
+      "loss": 0.746,
+      "step": 5640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733112247795815e-06,
+      "loss": 0.6965,
+      "step": 5641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7298727845461038e-06,
+      "loss": 0.7215,
+      "step": 5642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726634405203817e-06,
+      "loss": 0.6805,
+      "step": 5643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7233971103287793e-06,
+      "loss": 0.3013,
+      "step": 5644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720160900480605e-06,
+      "loss": 0.7289,
+      "step": 5645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7169257762187406e-06,
+      "loss": 0.7125,
+      "step": 5646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713691738102434e-06,
+      "loss": 0.7693,
+      "step": 5647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7104587866907504e-06,
+      "loss": 0.7018,
+      "step": 5648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7072269225425673e-06,
+      "loss": 0.7148,
+      "step": 5649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7039961462165662e-06,
+      "loss": 0.684,
+      "step": 5650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7007664582712546e-06,
+      "loss": 0.7545,
+      "step": 5651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6975378592649337e-06,
+      "loss": 0.6537,
+      "step": 5652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694310349755733e-06,
+      "loss": 0.7085,
+      "step": 5653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6910839303015857e-06,
+      "loss": 0.2643,
+      "step": 5654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6878586014602412e-06,
+      "loss": 0.6917,
+      "step": 5655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6846343637892514e-06,
+      "loss": 0.7404,
+      "step": 5656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6814112178459915e-06,
+      "loss": 0.7602,
+      "step": 5657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6781891641876354e-06,
+      "loss": 0.6329,
+      "step": 5658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6749682033711774e-06,
+      "loss": 0.7306,
+      "step": 5659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6717483359534224e-06,
+      "loss": 0.6527,
+      "step": 5660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668529562490981e-06,
+      "loss": 0.7424,
+      "step": 5661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6653118835402832e-06,
+      "loss": 0.2476,
+      "step": 5662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6620952996575564e-06,
+      "loss": 0.6818,
+      "step": 5663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6588798113988534e-06,
+      "loss": 0.7181,
+      "step": 5664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6556654193200247e-06,
+      "loss": 0.6814,
+      "step": 5665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6524521239767406e-06,
+      "loss": 0.6847,
+      "step": 5666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6492399259244793e-06,
+      "loss": 0.7049,
+      "step": 5667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6460288257185306e-06,
+      "loss": 0.8046,
+      "step": 5668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642818823913986e-06,
+      "loss": 0.7385,
+      "step": 5669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6396099210657574e-06,
+      "loss": 0.6957,
+      "step": 5670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636402117728567e-06,
+      "loss": 0.6522,
+      "step": 5671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6331954144569314e-06,
+      "loss": 0.224,
+      "step": 5672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6299898118052035e-06,
+      "loss": 0.2667,
+      "step": 5673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6267853103275197e-06,
+      "loss": 0.7561,
+      "step": 5674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6235819105778456e-06,
+      "loss": 0.6625,
+      "step": 5675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62037961310994e-06,
+      "loss": 0.6418,
+      "step": 5676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617178418477386e-06,
+      "loss": 0.6727,
+      "step": 5677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6139783272335595e-06,
+      "loss": 0.6991,
+      "step": 5678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6107793399316692e-06,
+      "loss": 0.7293,
+      "step": 5679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607581457124708e-06,
+      "loss": 0.3024,
+      "step": 5680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6043846793654945e-06,
+      "loss": 0.641,
+      "step": 5681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6011890072066523e-06,
+      "loss": 0.7084,
+      "step": 5682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5979944412006064e-06,
+      "loss": 0.7182,
+      "step": 5683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5948009818995997e-06,
+      "loss": 0.688,
+      "step": 5684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591608629855682e-06,
+      "loss": 0.7586,
+      "step": 5685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5884173856207118e-06,
+      "loss": 0.7092,
+      "step": 5686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.585227249746348e-06,
+      "loss": 0.6857,
+      "step": 5687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.582038222784073e-06,
+      "loss": 0.7684,
+      "step": 5688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578850305285162e-06,
+      "loss": 0.7006,
+      "step": 5689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.575663497800708e-06,
+      "loss": 0.6701,
+      "step": 5690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5724778008816097e-06,
+      "loss": 0.2601,
+      "step": 5691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5692932150785775e-06,
+      "loss": 0.7459,
+      "step": 5692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5661097409421185e-06,
+      "loss": 0.6556,
+      "step": 5693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.562927379022559e-06,
+      "loss": 0.6671,
+      "step": 5694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559746129870033e-06,
+      "loss": 0.7002,
+      "step": 5695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556565994034471e-06,
+      "loss": 0.794,
+      "step": 5696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5533869720656212e-06,
+      "loss": 0.7079,
+      "step": 5697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550209064513037e-06,
+      "loss": 0.7332,
+      "step": 5698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547032271926082e-06,
+      "loss": 0.7308,
+      "step": 5699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543856594853915e-06,
+      "loss": 0.7036,
+      "step": 5700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5406820338455195e-06,
+      "loss": 0.7096,
+      "step": 5701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5375085894496696e-06,
+      "loss": 0.6823,
+      "step": 5702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534336262214957e-06,
+      "loss": 0.71,
+      "step": 5703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5311650526897767e-06,
+      "loss": 0.7425,
+      "step": 5704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527994961422332e-06,
+      "loss": 0.6843,
+      "step": 5705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524825988960634e-06,
+      "loss": 0.7047,
+      "step": 5706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5216581358524916e-06,
+      "loss": 0.6839,
+      "step": 5707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518491402645534e-06,
+      "loss": 0.66,
+      "step": 5708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515325789887184e-06,
+      "loss": 0.6432,
+      "step": 5709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5121612981246778e-06,
+      "loss": 0.7159,
+      "step": 5710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508997927905057e-06,
+      "loss": 0.7543,
+      "step": 5711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5058356797751714e-06,
+      "loss": 0.6454,
+      "step": 5712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5026745542816674e-06,
+      "loss": 0.6994,
+      "step": 5713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499514551971013e-06,
+      "loss": 0.6595,
+      "step": 5714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4963556733894633e-06,
+      "loss": 0.7357,
+      "step": 5715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49319791908309e-06,
+      "loss": 0.7716,
+      "step": 5716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4900412895977797e-06,
+      "loss": 0.6569,
+      "step": 5717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4868857854792047e-06,
+      "loss": 0.6963,
+      "step": 5718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4837314072728567e-06,
+      "loss": 0.6989,
+      "step": 5719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4805781555240235e-06,
+      "loss": 0.7738,
+      "step": 5720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4774260307778097e-06,
+      "loss": 0.6721,
+      "step": 5721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4742750335791074e-06,
+      "loss": 0.7397,
+      "step": 5722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4711251644726396e-06,
+      "loss": 0.7483,
+      "step": 5723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4679764240029067e-06,
+      "loss": 0.6857,
+      "step": 5724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.464828812714236e-06,
+      "loss": 0.7359,
+      "step": 5725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461682331150742e-06,
+      "loss": 0.6089,
+      "step": 5726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4585369798563605e-06,
+      "loss": 0.7022,
+      "step": 5727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4553927593748117e-06,
+      "loss": 0.7575,
+      "step": 5728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452249670249643e-06,
+      "loss": 0.7083,
+      "step": 5729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449107713024197e-06,
+      "loss": 0.7093,
+      "step": 5730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4459668882416097e-06,
+      "loss": 0.6667,
+      "step": 5731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.44282719644484e-06,
+      "loss": 0.7219,
+      "step": 5732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4396886381766326e-06,
+      "loss": 0.7044,
+      "step": 5733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.436551213979551e-06,
+      "loss": 0.7619,
+      "step": 5734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4334149243959557e-06,
+      "loss": 0.6927,
+      "step": 5735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4302797699680157e-06,
+      "loss": 0.6934,
+      "step": 5736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4271457512376938e-06,
+      "loss": 0.6684,
+      "step": 5737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42401286874677e-06,
+      "loss": 0.6725,
+      "step": 5738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4208811230368133e-06,
+      "loss": 0.6751,
+      "step": 5739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4177505146492096e-06,
+      "loss": 0.6985,
+      "step": 5740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4146210441251402e-06,
+      "loss": 0.6921,
+      "step": 5741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411492712005592e-06,
+      "loss": 0.7276,
+      "step": 5742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40836551883136e-06,
+      "loss": 0.7371,
+      "step": 5743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40523946514303e-06,
+      "loss": 0.7077,
+      "step": 5744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4021145514810037e-06,
+      "loss": 0.6716,
+      "step": 5745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3989907783854747e-06,
+      "loss": 0.6723,
+      "step": 5746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395868146396448e-06,
+      "loss": 0.5945,
+      "step": 5747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.392746656053728e-06,
+      "loss": 0.7242,
+      "step": 5748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389626307896926e-06,
+      "loss": 0.672,
+      "step": 5749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386507102465445e-06,
+      "loss": 0.6841,
+      "step": 5750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3833890402985005e-06,
+      "loss": 0.7147,
+      "step": 5751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3802721219351096e-06,
+      "loss": 0.6545,
+      "step": 5752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3771563479140847e-06,
+      "loss": 0.7367,
+      "step": 5753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3740417187740472e-06,
+      "loss": 0.742,
+      "step": 5754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370928235053418e-06,
+      "loss": 0.6386,
+      "step": 5755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.367815897290424e-06,
+      "loss": 0.6457,
+      "step": 5756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364704706023083e-06,
+      "loss": 0.645,
+      "step": 5757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3615946617892303e-06,
+      "loss": 0.7251,
+      "step": 5758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3584857651264858e-06,
+      "loss": 0.6741,
+      "step": 5759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355378016572285e-06,
+      "loss": 0.7618,
+      "step": 5760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352271416663858e-06,
+      "loss": 0.6997,
+      "step": 5761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.349165965938239e-06,
+      "loss": 0.7196,
+      "step": 5762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3460616649322664e-06,
+      "loss": 0.7064,
+      "step": 5763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3429585141825683e-06,
+      "loss": 0.7558,
+      "step": 5764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.339856514225589e-06,
+      "loss": 0.7205,
+      "step": 5765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3367556655975597e-06,
+      "loss": 0.6973,
+      "step": 5766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3336559688345226e-06,
+      "loss": 0.7207,
+      "step": 5767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3305574244723183e-06,
+      "loss": 0.7098,
+      "step": 5768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327460033046589e-06,
+      "loss": 0.6458,
+      "step": 5769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324363795092771e-06,
+      "loss": 0.6925,
+      "step": 5770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321268711146114e-06,
+      "loss": 0.7632,
+      "step": 5771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3181747817416508e-06,
+      "loss": 0.6923,
+      "step": 5772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31508200741423e-06,
+      "loss": 0.7416,
+      "step": 5773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311990388698494e-06,
+      "loss": 0.6515,
+      "step": 5774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308899926128887e-06,
+      "loss": 0.6511,
+      "step": 5775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3058106202396546e-06,
+      "loss": 0.2608,
+      "step": 5776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3027224715648342e-06,
+      "loss": 0.7396,
+      "step": 5777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2996354806382737e-06,
+      "loss": 0.6145,
+      "step": 5778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2965496479936153e-06,
+      "loss": 0.7164,
+      "step": 5779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2934649741643067e-06,
+      "loss": 0.6838,
+      "step": 5780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2903814596835828e-06,
+      "loss": 0.6867,
+      "step": 5781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.287299105084495e-06,
+      "loss": 0.6255,
+      "step": 5782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284217910899876e-06,
+      "loss": 0.7478,
+      "step": 5783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2811378776623716e-06,
+      "loss": 0.7177,
+      "step": 5784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2780590059044235e-06,
+      "loss": 0.7558,
+      "step": 5785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2749812961582693e-06,
+      "loss": 0.595,
+      "step": 5786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271904748955954e-06,
+      "loss": 0.2609,
+      "step": 5787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268829364829308e-06,
+      "loss": 0.6904,
+      "step": 5788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2657551443099743e-06,
+      "loss": 0.7106,
+      "step": 5789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262682087929382e-06,
+      "loss": 0.6902,
+      "step": 5790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259610196218771e-06,
+      "loss": 0.7335,
+      "step": 5791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2565394697091724e-06,
+      "loss": 0.7343,
+      "step": 5792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2534699089314215e-06,
+      "loss": 0.7037,
+      "step": 5793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250401514416144e-06,
+      "loss": 0.6773,
+      "step": 5794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247334286693773e-06,
+      "loss": 0.7523,
+      "step": 5795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24426822629453e-06,
+      "loss": 0.6859,
+      "step": 5796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2412033337484427e-06,
+      "loss": 0.7721,
+      "step": 5797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238139609585337e-06,
+      "loss": 0.7188,
+      "step": 5798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235077054334831e-06,
+      "loss": 0.6821,
+      "step": 5799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2320156685263495e-06,
+      "loss": 0.6286,
+      "step": 5800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228955452689101e-06,
+      "loss": 0.62,
+      "step": 5801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225896407352109e-06,
+      "loss": 0.7075,
+      "step": 5802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2228385330441793e-06,
+      "loss": 0.7504,
+      "step": 5803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2197818302939243e-06,
+      "loss": 0.6579,
+      "step": 5804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2167262996297523e-06,
+      "loss": 0.2289,
+      "step": 5805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21367194157987e-06,
+      "loss": 0.6911,
+      "step": 5806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2106187566722757e-06,
+      "loss": 0.6836,
+      "step": 5807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2075667454347735e-06,
+      "loss": 0.7432,
+      "step": 5808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2045159083949538e-06,
+      "loss": 0.7339,
+      "step": 5809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201466246080215e-06,
+      "loss": 0.6503,
+      "step": 5810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1984177590177444e-06,
+      "loss": 0.6902,
+      "step": 5811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1953704477345326e-06,
+      "loss": 0.2668,
+      "step": 5812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1923243127573656e-06,
+      "loss": 0.7023,
+      "step": 5813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1892793546128177e-06,
+      "loss": 0.7989,
+      "step": 5814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186235573827272e-06,
+      "loss": 0.697,
+      "step": 5815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.183192970926897e-06,
+      "loss": 0.756,
+      "step": 5816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180151546437665e-06,
+      "loss": 0.7431,
+      "step": 5817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1771113008853438e-06,
+      "loss": 0.2586,
+      "step": 5818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1740722347954987e-06,
+      "loss": 0.7407,
+      "step": 5819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1710343486934814e-06,
+      "loss": 0.6721,
+      "step": 5820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167997643104451e-06,
+      "loss": 0.6618,
+      "step": 5821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1649621185533596e-06,
+      "loss": 0.753,
+      "step": 5822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161927775564947e-06,
+      "loss": 0.7022,
+      "step": 5823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1588946146637656e-06,
+      "loss": 0.683,
+      "step": 5824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155862636374145e-06,
+      "loss": 0.7613,
+      "step": 5825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1528318412202242e-06,
+      "loss": 0.728,
+      "step": 5826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149802229725928e-06,
+      "loss": 0.6995,
+      "step": 5827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146773802414984e-06,
+      "loss": 0.6576,
+      "step": 5828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143746559810905e-06,
+      "loss": 0.6934,
+      "step": 5829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140720502437017e-06,
+      "loss": 0.7215,
+      "step": 5830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137695630816421e-06,
+      "loss": 0.7384,
+      "step": 5831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1346719454720252e-06,
+      "loss": 0.7397,
+      "step": 5832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1316494469265336e-06,
+      "loss": 0.7015,
+      "step": 5833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1286281357024327e-06,
+      "loss": 0.6675,
+      "step": 5834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1256080123220157e-06,
+      "loss": 0.7177,
+      "step": 5835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1225890773073684e-06,
+      "loss": 0.6789,
+      "step": 5836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119571331180371e-06,
+      "loss": 0.6766,
+      "step": 5837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11655477446269e-06,
+      "loss": 0.7491,
+      "step": 5838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1135394076758017e-06,
+      "loss": 0.722,
+      "step": 5839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1105252313409594e-06,
+      "loss": 0.6922,
+      "step": 5840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1075122459792228e-06,
+      "loss": 0.6196,
+      "step": 5841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104500452111443e-06,
+      "loss": 0.6141,
+      "step": 5842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101489850258268e-06,
+      "loss": 0.7213,
+      "step": 5843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0984804409401272e-06,
+      "loss": 0.6834,
+      "step": 5844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0954722246772574e-06,
+      "loss": 0.7737,
+      "step": 5845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0924652019896884e-06,
+      "loss": 0.7359,
+      "step": 5846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089459373397231e-06,
+      "loss": 0.7156,
+      "step": 5847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086454739419503e-06,
+      "loss": 0.7126,
+      "step": 5848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0834513005759114e-06,
+      "loss": 0.6845,
+      "step": 5849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080449057385658e-06,
+      "loss": 0.2375,
+      "step": 5850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0774480103677307e-06,
+      "loss": 0.6603,
+      "step": 5851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0744481600409227e-06,
+      "loss": 0.6887,
+      "step": 5852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0714495069238048e-06,
+      "loss": 0.2383,
+      "step": 5853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068452051534756e-06,
+      "loss": 0.6627,
+      "step": 5854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0654557943919396e-06,
+      "loss": 0.6456,
+      "step": 5855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062460736013314e-06,
+      "loss": 0.7462,
+      "step": 5856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059466876916636e-06,
+      "loss": 0.6906,
+      "step": 5857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0564742176194405e-06,
+      "loss": 0.7373,
+      "step": 5858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0534827586390724e-06,
+      "loss": 0.7585,
+      "step": 5859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050492500492653e-06,
+      "loss": 0.7169,
+      "step": 5860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047503443697107e-06,
+      "loss": 0.5846,
+      "step": 5861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044515588769149e-06,
+      "loss": 0.668,
+      "step": 5862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0415289362252866e-06,
+      "loss": 0.6535,
+      "step": 5863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.038543486581813e-06,
+      "loss": 0.736,
+      "step": 5864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0355592403548252e-06,
+      "loss": 0.6844,
+      "step": 5865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0325761980601963e-06,
+      "loss": 0.6687,
+      "step": 5866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0295943602136036e-06,
+      "loss": 0.6381,
+      "step": 5867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0266137273305198e-06,
+      "loss": 0.7326,
+      "step": 5868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0236342999261947e-06,
+      "loss": 0.6664,
+      "step": 5869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0206560785156824e-06,
+      "loss": 0.747,
+      "step": 5870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017679063613819e-06,
+      "loss": 0.7491,
+      "step": 5871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014703255735241e-06,
+      "loss": 0.7189,
+      "step": 5872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0117286553943626e-06,
+      "loss": 0.7065,
+      "step": 5873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008755263105412e-06,
+      "loss": 0.6621,
+      "step": 5874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005783079382385e-06,
+      "loss": 0.6962,
+      "step": 5875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0028121047390855e-06,
+      "loss": 0.7078,
+      "step": 5876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999842339689094e-06,
+      "loss": 0.6903,
+      "step": 5877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996873784745793e-06,
+      "loss": 0.6953,
+      "step": 5878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993906440422353e-06,
+      "loss": 0.7198,
+      "step": 5879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990940307231733e-06,
+      "loss": 0.7005,
+      "step": 5880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987975385686688e-06,
+      "loss": 0.7775,
+      "step": 5881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9850116762997516e-06,
+      "loss": 0.6776,
+      "step": 5882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982049179583264e-06,
+      "loss": 0.6752,
+      "step": 5883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9790878960493406e-06,
+      "loss": 0.7084,
+      "step": 5884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.976127826209897e-06,
+      "loss": 0.7328,
+      "step": 5885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9731689705766364e-06,
+      "loss": 0.6919,
+      "step": 5886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9702113296610544e-06,
+      "loss": 0.7303,
+      "step": 5887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.967254903974428e-06,
+      "loss": 0.6712,
+      "step": 5888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9642996940278357e-06,
+      "loss": 0.7532,
+      "step": 5889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9613457003321344e-06,
+      "loss": 0.6679,
+      "step": 5890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958392923397979e-06,
+      "loss": 0.7268,
+      "step": 5891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9554413637358128e-06,
+      "loss": 0.7514,
+      "step": 5892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9524910218558667e-06,
+      "loss": 0.6838,
+      "step": 5893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9495418982681656e-06,
+      "loss": 0.7337,
+      "step": 5894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9465939934825127e-06,
+      "loss": 0.7386,
+      "step": 5895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.943647308008515e-06,
+      "loss": 0.6508,
+      "step": 5896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9407018423555545e-06,
+      "loss": 0.6818,
+      "step": 5897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9377575970328144e-06,
+      "loss": 0.7204,
+      "step": 5898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9348145725492595e-06,
+      "loss": 0.2454,
+      "step": 5899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9318727694136507e-06,
+      "loss": 0.7146,
+      "step": 5900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.691,
+      "step": 5901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9259928292202224e-06,
+      "loss": 0.2237,
+      "step": 5902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.923054693178866e-06,
+      "loss": 0.6888,
+      "step": 5903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.920117780518362e-06,
+      "loss": 0.724,
+      "step": 5904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917182091746413e-06,
+      "loss": 0.7205,
+      "step": 5905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9142476273705067e-06,
+      "loss": 0.7319,
+      "step": 5906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911314387897922e-06,
+      "loss": 0.7195,
+      "step": 5907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9083823738357186e-06,
+      "loss": 0.6575,
+      "step": 5908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905451585690756e-06,
+      "loss": 0.703,
+      "step": 5909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9025220239696673e-06,
+      "loss": 0.6372,
+      "step": 5910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8995936891788845e-06,
+      "loss": 0.7457,
+      "step": 5911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8966665818246263e-06,
+      "loss": 0.7096,
+      "step": 5912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893740702412896e-06,
+      "loss": 0.6883,
+      "step": 5913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890816051449489e-06,
+      "loss": 0.6987,
+      "step": 5914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887892629439979e-06,
+      "loss": 0.7003,
+      "step": 5915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.88497043688974e-06,
+      "loss": 0.7768,
+      "step": 5916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882049474303922e-06,
+      "loss": 0.6782,
+      "step": 5917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8791297421874685e-06,
+      "loss": 0.6691,
+      "step": 5918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8762112410451105e-06,
+      "loss": 0.6096,
+      "step": 5919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873293971381368e-06,
+      "loss": 0.6849,
+      "step": 5920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8703779337005377e-06,
+      "loss": 0.7625,
+      "step": 5921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8674631285067177e-06,
+      "loss": 0.7648,
+      "step": 5922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8645495563037795e-06,
+      "loss": 0.7421,
+      "step": 5923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.861637217595391e-06,
+      "loss": 0.7241,
+      "step": 5924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8587261128850043e-06,
+      "loss": 0.6969,
+      "step": 5925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8558162426758574e-06,
+      "loss": 0.6154,
+      "step": 5926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852907607470977e-06,
+      "loss": 0.6618,
+      "step": 5927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8500002077731715e-06,
+      "loss": 0.6765,
+      "step": 5928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8470940440850415e-06,
+      "loss": 0.6997,
+      "step": 5929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844189116908963e-06,
+      "loss": 0.6844,
+      "step": 5930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841285426747119e-06,
+      "loss": 0.6751,
+      "step": 5931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838382974101458e-06,
+      "loss": 0.6946,
+      "step": 5932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8354817594737263e-06,
+      "loss": 0.7686,
+      "step": 5933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832581783365448e-06,
+      "loss": 0.6493,
+      "step": 5934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8296830462779423e-06,
+      "loss": 0.6652,
+      "step": 5935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826785548712302e-06,
+      "loss": 0.6225,
+      "step": 5936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8238892911694214e-06,
+      "loss": 0.7665,
+      "step": 5937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820994274149973e-06,
+      "loss": 0.7294,
+      "step": 5938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8181004981544057e-06,
+      "loss": 0.7162,
+      "step": 5939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815207963682971e-06,
+      "loss": 0.7185,
+      "step": 5940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8123166712356896e-06,
+      "loss": 0.6689,
+      "step": 5941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809426621312379e-06,
+      "loss": 0.6945,
+      "step": 5942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8065378144126354e-06,
+      "loss": 0.7297,
+      "step": 5943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8036502510358477e-06,
+      "loss": 0.7265,
+      "step": 5944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8007639316811776e-06,
+      "loss": 0.6532,
+      "step": 5945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7978788568475847e-06,
+      "loss": 0.699,
+      "step": 5946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7949950270338023e-06,
+      "loss": 0.6543,
+      "step": 5947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792112442738355e-06,
+      "loss": 0.7232,
+      "step": 5948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.789231104459553e-06,
+      "loss": 0.7141,
+      "step": 5949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7863510126954885e-06,
+      "loss": 0.6706,
+      "step": 5950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7834721679440403e-06,
+      "loss": 0.6547,
+      "step": 5951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7805945707028647e-06,
+      "loss": 0.7515,
+      "step": 5952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7777182214694143e-06,
+      "loss": 0.7737,
+      "step": 5953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7748431207409122e-06,
+      "loss": 0.7288,
+      "step": 5954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.771969269014375e-06,
+      "loss": 0.6395,
+      "step": 5955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7690966667866038e-06,
+      "loss": 0.2661,
+      "step": 5956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7662253145541806e-06,
+      "loss": 0.6554,
+      "step": 5957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7633552128134687e-06,
+      "loss": 0.7544,
+      "step": 5958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7604863620606204e-06,
+      "loss": 0.6859,
+      "step": 5959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757618762791572e-06,
+      "loss": 0.6854,
+      "step": 5960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7547524155020345e-06,
+      "loss": 0.7154,
+      "step": 5961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751887320687513e-06,
+      "loss": 0.7441,
+      "step": 5962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7490234788432926e-06,
+      "loss": 0.6669,
+      "step": 5963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7461608904644433e-06,
+      "loss": 0.7192,
+      "step": 5964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743299556045811e-06,
+      "loss": 0.7223,
+      "step": 5965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7404394760820365e-06,
+      "loss": 0.6693,
+      "step": 5966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7375806510675306e-06,
+      "loss": 0.7594,
+      "step": 5967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7347230814964977e-06,
+      "loss": 0.6262,
+      "step": 5968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731866767862922e-06,
+      "loss": 0.7157,
+      "step": 5969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729011710660571e-06,
+      "loss": 0.7394,
+      "step": 5970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7261579103829904e-06,
+      "loss": 0.7213,
+      "step": 5971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7233053675235154e-06,
+      "loss": 0.7088,
+      "step": 5972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720454082575262e-06,
+      "loss": 0.6474,
+      "step": 5973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.71760405603112e-06,
+      "loss": 0.7057,
+      "step": 5974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7147552883837803e-06,
+      "loss": 0.6672,
+      "step": 5975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7119077801256964e-06,
+      "loss": 0.6612,
+      "step": 5976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709061531749119e-06,
+      "loss": 0.6674,
+      "step": 5977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7062165437460687e-06,
+      "loss": 0.7018,
+      "step": 5978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7033728166083605e-06,
+      "loss": 0.6699,
+      "step": 5979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.700530350827575e-06,
+      "loss": 0.7101,
+      "step": 5980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6976891468950985e-06,
+      "loss": 0.6585,
+      "step": 5981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6948492053020758e-06,
+      "loss": 0.7037,
+      "step": 5982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692010526539447e-06,
+      "loss": 0.6499,
+      "step": 5983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6891731110979314e-06,
+      "loss": 0.7215,
+      "step": 5984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6863369594680254e-06,
+      "loss": 0.6417,
+      "step": 5985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683502072140011e-06,
+      "loss": 0.6542,
+      "step": 5986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6806684496039525e-06,
+      "loss": 0.6537,
+      "step": 5987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677836092349696e-06,
+      "loss": 0.7506,
+      "step": 5988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6750050008668604e-06,
+      "loss": 0.7354,
+      "step": 5989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67217517564486e-06,
+      "loss": 0.681,
+      "step": 5990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6693466171728744e-06,
+      "loss": 0.7224,
+      "step": 5991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6665193259398757e-06,
+      "loss": 0.7146,
+      "step": 5992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6636933024346156e-06,
+      "loss": 0.7551,
+      "step": 5993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.660868547145622e-06,
+      "loss": 0.6784,
+      "step": 5994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6580450605612095e-06,
+      "loss": 0.7464,
+      "step": 5995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6552228431694637e-06,
+      "loss": 0.6672,
+      "step": 5996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652401895458265e-06,
+      "loss": 0.6653,
+      "step": 5997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6495822179152587e-06,
+      "loss": 0.6214,
+      "step": 5998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6467638110278817e-06,
+      "loss": 0.7204,
+      "step": 5999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6439466752833476e-06,
+      "loss": 0.6713,
+      "step": 6000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641130811168653e-06,
+      "loss": 0.7034,
+      "step": 6001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.638316219170567e-06,
+      "loss": 0.7051,
+      "step": 6002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6355028997756493e-06,
+      "loss": 0.7166,
+      "step": 6003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632690853470229e-06,
+      "loss": 0.7526,
+      "step": 6004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6298800807404213e-06,
+      "loss": 0.6436,
+      "step": 6005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627070582072122e-06,
+      "loss": 0.6511,
+      "step": 6006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624262357951004e-06,
+      "loss": 0.7171,
+      "step": 6007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6214554088625254e-06,
+      "loss": 0.5874,
+      "step": 6008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6186497352919116e-06,
+      "loss": 0.6896,
+      "step": 6009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615845337724181e-06,
+      "loss": 0.2413,
+      "step": 6010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6130422166441203e-06,
+      "loss": 0.7121,
+      "step": 6011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6102403725363035e-06,
+      "loss": 0.7312,
+      "step": 6012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6074398058850813e-06,
+      "loss": 0.7471,
+      "step": 6013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604640517174587e-06,
+      "loss": 0.6678,
+      "step": 6014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601842506888723e-06,
+      "loss": 0.6386,
+      "step": 6015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5990457755111833e-06,
+      "loss": 0.2325,
+      "step": 6016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5962503235254267e-06,
+      "loss": 0.7359,
+      "step": 6017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593456151414705e-06,
+      "loss": 0.6874,
+      "step": 6018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5906632596620395e-06,
+      "loss": 0.763,
+      "step": 6019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5878716487502365e-06,
+      "loss": 0.6225,
+      "step": 6020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5850813191618775e-06,
+      "loss": 0.6574,
+      "step": 6021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5822922713793173e-06,
+      "loss": 0.7356,
+      "step": 6022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579504505884701e-06,
+      "loss": 0.748,
+      "step": 6023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57671802315994e-06,
+      "loss": 0.6505,
+      "step": 6024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5739328236867313e-06,
+      "loss": 0.6076,
+      "step": 6025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5711489079465478e-06,
+      "loss": 0.7698,
+      "step": 6026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5683662764206454e-06,
+      "loss": 0.6986,
+      "step": 6027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5655849295900448e-06,
+      "loss": 0.6866,
+      "step": 6028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5628048679355587e-06,
+      "loss": 0.7197,
+      "step": 6029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5600260919377717e-06,
+      "loss": 0.6941,
+      "step": 6030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557248602077047e-06,
+      "loss": 0.6548,
+      "step": 6031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5544723988335265e-06,
+      "loss": 0.7783,
+      "step": 6032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551697482687122e-06,
+      "loss": 0.6795,
+      "step": 6033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5489238541175375e-06,
+      "loss": 0.7816,
+      "step": 6034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546151513604238e-06,
+      "loss": 0.7364,
+      "step": 6035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543380461626478e-06,
+      "loss": 0.7065,
+      "step": 6036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540610698663284e-06,
+      "loss": 0.7452,
+      "step": 6037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5378422251934655e-06,
+      "loss": 0.6949,
+      "step": 6038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5350750416955962e-06,
+      "loss": 0.6531,
+      "step": 6039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5323091486480412e-06,
+      "loss": 0.7708,
+      "step": 6040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5295445465289315e-06,
+      "loss": 0.7886,
+      "step": 6041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5267812358161826e-06,
+      "loss": 0.7277,
+      "step": 6042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524019216987482e-06,
+      "loss": 0.5782,
+      "step": 6043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5212584905202988e-06,
+      "loss": 0.736,
+      "step": 6044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5184990568918754e-06,
+      "loss": 0.696,
+      "step": 6045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5157409165792266e-06,
+      "loss": 0.7098,
+      "step": 6046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5129840700591546e-06,
+      "loss": 0.783,
+      "step": 6047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.510228517808224e-06,
+      "loss": 0.7208,
+      "step": 6048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5074742603027867e-06,
+      "loss": 0.6683,
+      "step": 6049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5047212980189662e-06,
+      "loss": 0.6614,
+      "step": 6050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501969631432667e-06,
+      "loss": 0.7222,
+      "step": 6051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499219261019559e-06,
+      "loss": 0.7253,
+      "step": 6052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4964701872550976e-06,
+      "loss": 0.2851,
+      "step": 6053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493722410614514e-06,
+      "loss": 0.798,
+      "step": 6054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4909759315728064e-06,
+      "loss": 0.6196,
+      "step": 6055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4882307506047575e-06,
+      "loss": 0.7876,
+      "step": 6056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485486868184922e-06,
+      "loss": 0.6645,
+      "step": 6057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4827442847876336e-06,
+      "loss": 0.7407,
+      "step": 6058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4800030008869945e-06,
+      "loss": 0.7105,
+      "step": 6059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4772630169568902e-06,
+      "loss": 0.5943,
+      "step": 6060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474524333470972e-06,
+      "loss": 0.7151,
+      "step": 6061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.471786950902676e-06,
+      "loss": 0.738,
+      "step": 6062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469050869725208e-06,
+      "loss": 0.6578,
+      "step": 6063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4663160904115523e-06,
+      "loss": 0.7269,
+      "step": 6064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4635826134344664e-06,
+      "loss": 0.6842,
+      "step": 6065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4608504392664778e-06,
+      "loss": 0.7324,
+      "step": 6066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4581195683798986e-06,
+      "loss": 0.7887,
+      "step": 6067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4553900012468058e-06,
+      "loss": 0.6531,
+      "step": 6068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4526617383390573e-06,
+      "loss": 0.6849,
+      "step": 6069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449934780128285e-06,
+      "loss": 0.7544,
+      "step": 6070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4472091270858945e-06,
+      "loss": 0.7269,
+      "step": 6071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4444847796830618e-06,
+      "loss": 0.6708,
+      "step": 6072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4417617383907464e-06,
+      "loss": 0.6888,
+      "step": 6073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439040003679669e-06,
+      "loss": 0.6968,
+      "step": 6074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436319576020334e-06,
+      "loss": 0.6956,
+      "step": 6075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433600455883024e-06,
+      "loss": 0.6821,
+      "step": 6076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.430882643737782e-06,
+      "loss": 0.6688,
+      "step": 6077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4281661400544366e-06,
+      "loss": 0.7471,
+      "step": 6078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425450945302582e-06,
+      "loss": 0.7929,
+      "step": 6079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422737059951593e-06,
+      "loss": 0.6926,
+      "step": 6080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420024484470609e-06,
+      "loss": 0.7194,
+      "step": 6081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4173132193285576e-06,
+      "loss": 0.6026,
+      "step": 6082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414603264994124e-06,
+      "loss": 0.6869,
+      "step": 6083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411894621935781e-06,
+      "loss": 0.7232,
+      "step": 6084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409187290621758e-06,
+      "loss": 0.6929,
+      "step": 6085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406481271520077e-06,
+      "loss": 0.6686,
+      "step": 6086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4037765650985122e-06,
+      "loss": 0.7311,
+      "step": 6087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4010731718246327e-06,
+      "loss": 0.6497,
+      "step": 6088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398371092165769e-06,
+      "loss": 0.6944,
+      "step": 6089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395670326589019e-06,
+      "loss": 0.6871,
+      "step": 6090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392970875561267e-06,
+      "loss": 0.2604,
+      "step": 6091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3902727395491566e-06,
+      "loss": 0.7453,
+      "step": 6092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3875759190191127e-06,
+      "loss": 0.7239,
+      "step": 6093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3848804144373326e-06,
+      "loss": 0.723,
+      "step": 6094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.382186226269786e-06,
+      "loss": 0.7391,
+      "step": 6095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3794933549822076e-06,
+      "loss": 0.7125,
+      "step": 6096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376801801040114e-06,
+      "loss": 0.7133,
+      "step": 6097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3741115649087866e-06,
+      "loss": 0.6916,
+      "step": 6098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3714226470532843e-06,
+      "loss": 0.7174,
+      "step": 6099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3687350479384363e-06,
+      "loss": 0.7622,
+      "step": 6100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3660487680288446e-06,
+      "loss": 0.677,
+      "step": 6101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3633638077888843e-06,
+      "loss": 0.6882,
+      "step": 6102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360680167682694e-06,
+      "loss": 0.6643,
+      "step": 6103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357997848174198e-06,
+      "loss": 0.7401,
+      "step": 6104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3553168497270784e-06,
+      "loss": 0.6886,
+      "step": 6105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352637172804798e-06,
+      "loss": 0.2454,
+      "step": 6106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349958817870589e-06,
+      "loss": 0.2783,
+      "step": 6107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3472817853874564e-06,
+      "loss": 0.6968,
+      "step": 6108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.34460607581817e-06,
+      "loss": 0.6599,
+      "step": 6109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3419316896252777e-06,
+      "loss": 0.6417,
+      "step": 6110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3392586272711002e-06,
+      "loss": 0.7733,
+      "step": 6111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33658688921772e-06,
+      "loss": 0.6593,
+      "step": 6112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3339164759269996e-06,
+      "loss": 0.6453,
+      "step": 6113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331247387860568e-06,
+      "loss": 0.7172,
+      "step": 6114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328579625479831e-06,
+      "loss": 0.7112,
+      "step": 6115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325913189245953e-06,
+      "loss": 0.6242,
+      "step": 6116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323248079619884e-06,
+      "loss": 0.7377,
+      "step": 6117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3205842970623314e-06,
+      "loss": 0.7399,
+      "step": 6118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3179218420337823e-06,
+      "loss": 0.7053,
+      "step": 6119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3152607149944905e-06,
+      "loss": 0.7143,
+      "step": 6120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312600916404485e-06,
+      "loss": 0.6679,
+      "step": 6121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3099424467235543e-06,
+      "loss": 0.7383,
+      "step": 6122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307285306411269e-06,
+      "loss": 0.6908,
+      "step": 6123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3046294959269644e-06,
+      "loss": 0.7371,
+      "step": 6124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3019750157297403e-06,
+      "loss": 0.7072,
+      "step": 6125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2993218662784846e-06,
+      "loss": 0.7137,
+      "step": 6126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2966700480318317e-06,
+      "loss": 0.6978,
+      "step": 6127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2940195614482063e-06,
+      "loss": 0.7326,
+      "step": 6128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291370406985787e-06,
+      "loss": 0.7166,
+      "step": 6129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2887225851025353e-06,
+      "loss": 0.7371,
+      "step": 6130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2860760962561666e-06,
+      "loss": 0.6732,
+      "step": 6131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283430940904188e-06,
+      "loss": 0.7488,
+      "step": 6132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2807871195038535e-06,
+      "loss": 0.6656,
+      "step": 6133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.278144632512201e-06,
+      "loss": 0.5657,
+      "step": 6134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2755034803860353e-06,
+      "loss": 0.696,
+      "step": 6135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2728636635819225e-06,
+      "loss": 0.7103,
+      "step": 6136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270225182556207e-06,
+      "loss": 0.6598,
+      "step": 6137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2675880377649993e-06,
+      "loss": 0.7312,
+      "step": 6138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2649522296641803e-06,
+      "loss": 0.78,
+      "step": 6139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2623177587093937e-06,
+      "loss": 0.691,
+      "step": 6140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2596846253560623e-06,
+      "loss": 0.6701,
+      "step": 6141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2570528300593653e-06,
+      "loss": 0.6828,
+      "step": 6142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2544223732742597e-06,
+      "loss": 0.7191,
+      "step": 6143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25179325545547e-06,
+      "loss": 0.7348,
+      "step": 6144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249165477057487e-06,
+      "loss": 0.7422,
+      "step": 6145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246539038534573e-06,
+      "loss": 0.7203,
+      "step": 6146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243913940340753e-06,
+      "loss": 0.7413,
+      "step": 6147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2412901829298272e-06,
+      "loss": 0.6449,
+      "step": 6148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2386677667553548e-06,
+      "loss": 0.6942,
+      "step": 6149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2360466922706734e-06,
+      "loss": 0.7395,
+      "step": 6150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233426959928884e-06,
+      "loss": 0.734,
+      "step": 6151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2308085701828573e-06,
+      "loss": 0.7055,
+      "step": 6152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2281915234852258e-06,
+      "loss": 0.7142,
+      "step": 6153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2255758202884e-06,
+      "loss": 0.7741,
+      "step": 6154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222961461044546e-06,
+      "loss": 0.2596,
+      "step": 6155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220348446205608e-06,
+      "loss": 0.2746,
+      "step": 6156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217736776223294e-06,
+      "loss": 0.7283,
+      "step": 6157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2151264515490778e-06,
+      "loss": 0.722,
+      "step": 6158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212517472634207e-06,
+      "loss": 0.7453,
+      "step": 6159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209909839929686e-06,
+      "loss": 0.6941,
+      "step": 6160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2073035538862964e-06,
+      "loss": 0.7101,
+      "step": 6161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204698614954579e-06,
+      "loss": 0.7605,
+      "step": 6162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202095023584848e-06,
+      "loss": 0.7704,
+      "step": 6163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1994927802271814e-06,
+      "loss": 0.6681,
+      "step": 6164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1968918853314294e-06,
+      "loss": 0.6721,
+      "step": 6165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194292339347198e-06,
+      "loss": 0.7398,
+      "step": 6166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1916941427238725e-06,
+      "loss": 0.6422,
+      "step": 6167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189097295910595e-06,
+      "loss": 0.7131,
+      "step": 6168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18650179935628e-06,
+      "loss": 0.5974,
+      "step": 6169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183907653509606e-06,
+      "loss": 0.2629,
+      "step": 6170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1813148588190212e-06,
+      "loss": 0.6708,
+      "step": 6171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178723415732741e-06,
+      "loss": 0.6677,
+      "step": 6172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176133324698737e-06,
+      "loss": 0.735,
+      "step": 6173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1735445861647595e-06,
+      "loss": 0.2613,
+      "step": 6174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1709572005783175e-06,
+      "loss": 0.627,
+      "step": 6175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168371168386688e-06,
+      "loss": 0.6856,
+      "step": 6176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1657864900369154e-06,
+      "loss": 0.7295,
+      "step": 6177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1632031659758123e-06,
+      "loss": 0.6854,
+      "step": 6178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160621196649949e-06,
+      "loss": 0.7333,
+      "step": 6179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1580405825056685e-06,
+      "loss": 0.7841,
+      "step": 6180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15546132398908e-06,
+      "loss": 0.7912,
+      "step": 6181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1528834215460483e-06,
+      "loss": 0.6872,
+      "step": 6182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1503068756222232e-06,
+      "loss": 0.7417,
+      "step": 6183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1477316866629995e-06,
+      "loss": 0.682,
+      "step": 6184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451578551135523e-06,
+      "loss": 0.7302,
+      "step": 6185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142585381418809e-06,
+      "loss": 0.7411,
+      "step": 6186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1400142660234757e-06,
+      "loss": 0.7469,
+      "step": 6187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1374445093720085e-06,
+      "loss": 0.6882,
+      "step": 6188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1348761119086503e-06,
+      "loss": 0.6822,
+      "step": 6189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1323090740773853e-06,
+      "loss": 0.7504,
+      "step": 6190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.129743396321977e-06,
+      "loss": 0.7898,
+      "step": 6191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1271790790859537e-06,
+      "loss": 0.7188,
+      "step": 6192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1246161228125996e-06,
+      "loss": 0.6981,
+      "step": 6193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12205452794497e-06,
+      "loss": 0.7324,
+      "step": 6194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119494294925886e-06,
+      "loss": 0.7159,
+      "step": 6195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1169354241979336e-06,
+      "loss": 0.6624,
+      "step": 6196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114377916203454e-06,
+      "loss": 0.6808,
+      "step": 6197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1118217713845667e-06,
+      "loss": 0.6496,
+      "step": 6198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109266990183143e-06,
+      "loss": 0.7237,
+      "step": 6199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106713573040827e-06,
+      "loss": 0.731,
+      "step": 6200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1041615203990227e-06,
+      "loss": 0.7391,
+      "step": 6201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1016108326989037e-06,
+      "loss": 0.7158,
+      "step": 6202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099061510381397e-06,
+      "loss": 0.6803,
+      "step": 6203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096513553887204e-06,
+      "loss": 0.6876,
+      "step": 6204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093966963656787e-06,
+      "loss": 0.6242,
+      "step": 6205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.091421740130368e-06,
+      "loss": 0.6752,
+      "step": 6206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0888778837479385e-06,
+      "loss": 0.7345,
+      "step": 6207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086335394949249e-06,
+      "loss": 0.6863,
+      "step": 6208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0837942741738193e-06,
+      "loss": 0.655,
+      "step": 6209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0812545218609247e-06,
+      "loss": 0.6875,
+      "step": 6210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0787161384496135e-06,
+      "loss": 0.7122,
+      "step": 6211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0761791243786866e-06,
+      "loss": 0.6882,
+      "step": 6212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0736434800867155e-06,
+      "loss": 0.6634,
+      "step": 6213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0711092060120343e-06,
+      "loss": 0.668,
+      "step": 6214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0685763025927386e-06,
+      "loss": 0.7097,
+      "step": 6215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.066044770266692e-06,
+      "loss": 0.6483,
+      "step": 6216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0635146094715087e-06,
+      "loss": 0.7637,
+      "step": 6217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06098582064458e-06,
+      "loss": 0.7045,
+      "step": 6218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0584584042230493e-06,
+      "loss": 0.7457,
+      "step": 6219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0559323606438286e-06,
+      "loss": 0.6786,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.053407690343593e-06,
+      "loss": 0.6974,
+      "step": 6221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0508843937587795e-06,
+      "loss": 0.6997,
+      "step": 6222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048362471325582e-06,
+      "loss": 0.7427,
+      "step": 6223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0458419234799667e-06,
+      "loss": 0.7502,
+      "step": 6224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0433227506576513e-06,
+      "loss": 0.247,
+      "step": 6225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040804953294122e-06,
+      "loss": 0.7434,
+      "step": 6226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0382885318246336e-06,
+      "loss": 0.7574,
+      "step": 6227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0357734866841894e-06,
+      "loss": 0.6774,
+      "step": 6228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033259818307566e-06,
+      "loss": 0.6451,
+      "step": 6229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030747527129292e-06,
+      "loss": 0.5899,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0282366135836697e-06,
+      "loss": 0.6713,
+      "step": 6231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0257270781047468e-06,
+      "loss": 0.7035,
+      "step": 6232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0232189211263564e-06,
+      "loss": 0.7245,
+      "step": 6233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.02071214308207e-06,
+      "loss": 0.7322,
+      "step": 6234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0182067444052357e-06,
+      "loss": 0.6989,
+      "step": 6235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015702725528953e-06,
+      "loss": 0.7096,
+      "step": 6236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0132000868860936e-06,
+      "loss": 0.6669,
+      "step": 6237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0106988289092776e-06,
+      "loss": 0.7366,
+      "step": 6238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0081989520309e-06,
+      "loss": 0.7474,
+      "step": 6239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005700456683112e-06,
+      "loss": 0.6983,
+      "step": 6240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003203343297818e-06,
+      "loss": 0.7427,
+      "step": 6241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000707612306697e-06,
+      "loss": 0.7065,
+      "step": 6242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9982132641411754e-06,
+      "loss": 0.7481,
+      "step": 6243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995720299232452e-06,
+      "loss": 0.7161,
+      "step": 6244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9932287180114807e-06,
+      "loss": 0.7002,
+      "step": 6245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9907385209089814e-06,
+      "loss": 0.6513,
+      "step": 6246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9882497083554233e-06,
+      "loss": 0.7579,
+      "step": 6247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9857622807810496e-06,
+      "loss": 0.7164,
+      "step": 6248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9832762386158532e-06,
+      "loss": 0.6189,
+      "step": 6249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980791582289595e-06,
+      "loss": 0.2855,
+      "step": 6250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9783083122317946e-06,
+      "loss": 0.7428,
+      "step": 6251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97582642887173e-06,
+      "loss": 0.7205,
+      "step": 6252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973345932638443e-06,
+      "loss": 0.2802,
+      "step": 6253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97086682396073e-06,
+      "loss": 0.748,
+      "step": 6254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9683891032671542e-06,
+      "loss": 0.6505,
+      "step": 6255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9659127709860303e-06,
+      "loss": 0.6399,
+      "step": 6256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963437827545441e-06,
+      "loss": 0.7261,
+      "step": 6257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960964273373227e-06,
+      "loss": 0.7068,
+      "step": 6258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958492108896989e-06,
+      "loss": 0.7217,
+      "step": 6259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956021334544083e-06,
+      "loss": 0.7228,
+      "step": 6260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953551950741628e-06,
+      "loss": 0.6381,
+      "step": 6261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9510839579165066e-06,
+      "loss": 0.7376,
+      "step": 6262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9486173564953526e-06,
+      "loss": 0.742,
+      "step": 6263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946152146904565e-06,
+      "loss": 0.6508,
+      "step": 6264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9436883295703022e-06,
+      "loss": 0.659,
+      "step": 6265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9412259049184824e-06,
+      "loss": 0.6956,
+      "step": 6266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9387648733747743e-06,
+      "loss": 0.6979,
+      "step": 6267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936305235364622e-06,
+      "loss": 0.7457,
+      "step": 6268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93384699131321e-06,
+      "loss": 0.7152,
+      "step": 6269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9313901416454962e-06,
+      "loss": 0.6327,
+      "step": 6270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9289346867861924e-06,
+      "loss": 0.6708,
+      "step": 6271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9264806271597693e-06,
+      "loss": 0.6973,
+      "step": 6272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924027963190458e-06,
+      "loss": 0.7103,
+      "step": 6273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921576695302242e-06,
+      "loss": 0.7105,
+      "step": 6274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9191268239188744e-06,
+      "loss": 0.6708,
+      "step": 6275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916678349463855e-06,
+      "loss": 0.6048,
+      "step": 6276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9142312723604504e-06,
+      "loss": 0.7278,
+      "step": 6277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911785593031683e-06,
+      "loss": 0.7147,
+      "step": 6278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909341311900337e-06,
+      "loss": 0.6848,
+      "step": 6279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9068984293889449e-06,
+      "loss": 0.7077,
+      "step": 6280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9044569459198114e-06,
+      "loss": 0.7209,
+      "step": 6281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9020168619149847e-06,
+      "loss": 0.6971,
+      "step": 6282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8995781777962829e-06,
+      "loss": 0.7631,
+      "step": 6283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8971408939852776e-06,
+      "loss": 0.7226,
+      "step": 6284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.894705010903296e-06,
+      "loss": 0.6436,
+      "step": 6285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8922705289714304e-06,
+      "loss": 0.7136,
+      "step": 6286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8898374486105208e-06,
+      "loss": 0.681,
+      "step": 6287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8874057702411707e-06,
+      "loss": 0.7029,
+      "step": 6288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884975494283744e-06,
+      "loss": 0.753,
+      "step": 6289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8825466211583577e-06,
+      "loss": 0.7253,
+      "step": 6290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8801191512848848e-06,
+      "loss": 0.7158,
+      "step": 6291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8776930850829633e-06,
+      "loss": 0.7318,
+      "step": 6292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8752684229719776e-06,
+      "loss": 0.7741,
+      "step": 6293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8728451653710776e-06,
+      "loss": 0.7107,
+      "step": 6294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.87042331269917e-06,
+      "loss": 0.7072,
+      "step": 6295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8680028653749148e-06,
+      "loss": 0.6742,
+      "step": 6296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8655838238167344e-06,
+      "loss": 0.7367,
+      "step": 6297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8631661884427998e-06,
+      "loss": 0.6633,
+      "step": 6298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8607499596710499e-06,
+      "loss": 0.7103,
+      "step": 6299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8583351379191683e-06,
+      "loss": 0.7342,
+      "step": 6300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8559217236046045e-06,
+      "loss": 0.7672,
+      "step": 6301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8535097171445626e-06,
+      "loss": 0.7214,
+      "step": 6302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8510991189560034e-06,
+      "loss": 0.6715,
+      "step": 6303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8486899294556393e-06,
+      "loss": 0.7211,
+      "step": 6304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846282149059948e-06,
+      "loss": 0.6647,
+      "step": 6305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8438757781851534e-06,
+      "loss": 0.6949,
+      "step": 6306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8414708172472461e-06,
+      "loss": 0.697,
+      "step": 6307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8390672666619647e-06,
+      "loss": 0.6699,
+      "step": 6308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.83666512684481e-06,
+      "loss": 0.7266,
+      "step": 6309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8342643982110363e-06,
+      "loss": 0.6947,
+      "step": 6310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8318650811756522e-06,
+      "loss": 0.6482,
+      "step": 6311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8294671761534256e-06,
+      "loss": 0.7485,
+      "step": 6312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8270706835588758e-06,
+      "loss": 0.6465,
+      "step": 6313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824675603806283e-06,
+      "loss": 0.2427,
+      "step": 6314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.822281937309681e-06,
+      "loss": 0.7499,
+      "step": 6315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198896844828617e-06,
+      "loss": 0.6137,
+      "step": 6316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.817498845739365e-06,
+      "loss": 0.6414,
+      "step": 6317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.815109421492497e-06,
+      "loss": 0.2485,
+      "step": 6318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8127214121553083e-06,
+      "loss": 0.7144,
+      "step": 6319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8103348181406144e-06,
+      "loss": 0.666,
+      "step": 6320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8079496398609796e-06,
+      "loss": 0.7218,
+      "step": 6321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8055658777287288e-06,
+      "loss": 0.7231,
+      "step": 6322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80318353215594e-06,
+      "loss": 0.7178,
+      "step": 6323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8008026035544424e-06,
+      "loss": 0.6837,
+      "step": 6324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7984230923358258e-06,
+      "loss": 0.8043,
+      "step": 6325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7960449989114314e-06,
+      "loss": 0.7024,
+      "step": 6326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7936683236923557e-06,
+      "loss": 0.7692,
+      "step": 6327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7912930670894525e-06,
+      "loss": 0.71,
+      "step": 6328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7889192295133317e-06,
+      "loss": 0.6979,
+      "step": 6329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.78654681137435e-06,
+      "loss": 0.667,
+      "step": 6330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7841758130826247e-06,
+      "loss": 0.7391,
+      "step": 6331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7818062350480313e-06,
+      "loss": 0.7121,
+      "step": 6332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7794380776801857e-06,
+      "loss": 0.7255,
+      "step": 6333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7770713413884788e-06,
+      "loss": 0.6597,
+      "step": 6334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7747060265820382e-06,
+      "loss": 0.7531,
+      "step": 6335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7723421336697545e-06,
+      "loss": 0.6469,
+      "step": 6336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7699796630602683e-06,
+      "loss": 0.7343,
+      "step": 6337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767618615161979e-06,
+      "loss": 0.644,
+      "step": 6338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7652589903830319e-06,
+      "loss": 0.7335,
+      "step": 6339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.76290078913134e-06,
+      "loss": 0.7376,
+      "step": 6340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7605440118145556e-06,
+      "loss": 0.6947,
+      "step": 6341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.758188658840093e-06,
+      "loss": 0.5885,
+      "step": 6342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7558347306151225e-06,
+      "loss": 0.7115,
+      "step": 6343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.753482227546557e-06,
+      "loss": 0.7044,
+      "step": 6344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7511311500410743e-06,
+      "loss": 0.675,
+      "step": 6345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.748781498505101e-06,
+      "loss": 0.7206,
+      "step": 6346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7464332733448197e-06,
+      "loss": 0.6461,
+      "step": 6347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74408647496616e-06,
+      "loss": 0.7247,
+      "step": 6348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7417411037748155e-06,
+      "loss": 0.7208,
+      "step": 6349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7393971601762204e-06,
+      "loss": 0.7464,
+      "step": 6350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737054644575572e-06,
+      "loss": 0.6762,
+      "step": 6351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7347135573778184e-06,
+      "loss": 0.6904,
+      "step": 6352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732373898987657e-06,
+      "loss": 0.6971,
+      "step": 6353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7300356698095455e-06,
+      "loss": 0.6834,
+      "step": 6354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7276988702476861e-06,
+      "loss": 0.681,
+      "step": 6355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7253635007060398e-06,
+      "loss": 0.7017,
+      "step": 6356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7230295615883152e-06,
+      "loss": 0.7596,
+      "step": 6357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7206970532979806e-06,
+      "loss": 0.7609,
+      "step": 6358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718365976238251e-06,
+      "loss": 0.7073,
+      "step": 6359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7160363308120986e-06,
+      "loss": 0.7102,
+      "step": 6360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7137081174222424e-06,
+      "loss": 0.6511,
+      "step": 6361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7113813364711607e-06,
+      "loss": 0.7131,
+      "step": 6362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709055988361077e-06,
+      "loss": 0.7514,
+      "step": 6363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706732073493972e-06,
+      "loss": 0.2537,
+      "step": 6364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7044095922715787e-06,
+      "loss": 0.7072,
+      "step": 6365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7020885450953783e-06,
+      "loss": 0.7208,
+      "step": 6366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6997689323666121e-06,
+      "loss": 0.7027,
+      "step": 6367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6974507544862616e-06,
+      "loss": 0.6556,
+      "step": 6368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951340118550718e-06,
+      "loss": 0.6803,
+      "step": 6369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.69281870487353e-06,
+      "loss": 0.7293,
+      "step": 6370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6905048339418816e-06,
+      "loss": 0.7696,
+      "step": 6371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6881923994601223e-06,
+      "loss": 0.6494,
+      "step": 6372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6858814018280022e-06,
+      "loss": 0.2606,
+      "step": 6373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6835718414450143e-06,
+      "loss": 0.7115,
+      "step": 6374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6812637187104142e-06,
+      "loss": 0.6911,
+      "step": 6375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6789570340231987e-06,
+      "loss": 0.683,
+      "step": 6376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.676651787782122e-06,
+      "loss": 0.6878,
+      "step": 6377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674347980385691e-06,
+      "loss": 0.7015,
+      "step": 6378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6720456122321594e-06,
+      "loss": 0.7229,
+      "step": 6379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6697446837195374e-06,
+      "loss": 0.7093,
+      "step": 6380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6674451952455771e-06,
+      "loss": 0.6439,
+      "step": 6381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6651471472077952e-06,
+      "loss": 0.719,
+      "step": 6382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6628505400034412e-06,
+      "loss": 0.6555,
+      "step": 6383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6605553740295377e-06,
+      "loss": 0.7317,
+      "step": 6384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6582616496828385e-06,
+      "loss": 0.6698,
+      "step": 6385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655969367359862e-06,
+      "loss": 0.2495,
+      "step": 6386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6536785274568668e-06,
+      "loss": 0.6849,
+      "step": 6387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651389130369868e-06,
+      "loss": 0.6364,
+      "step": 6388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6491011764946307e-06,
+      "loss": 0.2435,
+      "step": 6389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6468146662266714e-06,
+      "loss": 0.6841,
+      "step": 6390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6445295999612564e-06,
+      "loss": 0.6976,
+      "step": 6391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6422459780933975e-06,
+      "loss": 0.6918,
+      "step": 6392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6399638010178665e-06,
+      "loss": 0.7194,
+      "step": 6393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6376830691291735e-06,
+      "loss": 0.7793,
+      "step": 6394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6354037828215885e-06,
+      "loss": 0.738,
+      "step": 6395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6331259424891289e-06,
+      "loss": 0.7604,
+      "step": 6396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6308495485255626e-06,
+      "loss": 0.6918,
+      "step": 6397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.628574601324402e-06,
+      "loss": 0.6677,
+      "step": 6398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6263011012789198e-06,
+      "loss": 0.6868,
+      "step": 6399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6240290487821254e-06,
+      "loss": 0.6908,
+      "step": 6400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6217584442267898e-06,
+      "loss": 0.6988,
+      "step": 6401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619489288005428e-06,
+      "loss": 0.7036,
+      "step": 6402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6172215805103053e-06,
+      "loss": 0.6983,
+      "step": 6403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6149553221334402e-06,
+      "loss": 0.6322,
+      "step": 6404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6126905132665916e-06,
+      "loss": 0.6707,
+      "step": 6405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6104271543012783e-06,
+      "loss": 0.704,
+      "step": 6406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60816524562876e-06,
+      "loss": 0.6657,
+      "step": 6407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6059047876400514e-06,
+      "loss": 0.6759,
+      "step": 6408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6036457807259143e-06,
+      "loss": 0.6566,
+      "step": 6409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.601388225276862e-06,
+      "loss": 0.713,
+      "step": 6410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59913212168315e-06,
+      "loss": 0.737,
+      "step": 6411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5968774703347901e-06,
+      "loss": 0.6948,
+      "step": 6412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5946242716215433e-06,
+      "loss": 0.7045,
+      "step": 6413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5923725259329104e-06,
+      "loss": 0.6887,
+      "step": 6414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5901222336581513e-06,
+      "loss": 0.7623,
+      "step": 6415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5878733951862702e-06,
+      "loss": 0.6803,
+      "step": 6416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5856260109060229e-06,
+      "loss": 0.6789,
+      "step": 6417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5833800812059052e-06,
+      "loss": 0.7148,
+      "step": 6418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5811356064741745e-06,
+      "loss": 0.727,
+      "step": 6419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5788925870988226e-06,
+      "loss": 0.6791,
+      "step": 6420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5766510234676013e-06,
+      "loss": 0.2676,
+      "step": 6421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5744109159680066e-06,
+      "loss": 0.6918,
+      "step": 6422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5721722649872795e-06,
+      "loss": 0.6473,
+      "step": 6423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5699350709124183e-06,
+      "loss": 0.6424,
+      "step": 6424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.567699334130156e-06,
+      "loss": 0.692,
+      "step": 6425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5654650550269867e-06,
+      "loss": 0.7326,
+      "step": 6426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5632322339891427e-06,
+      "loss": 0.7041,
+      "step": 6427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5610008714026092e-06,
+      "loss": 0.6759,
+      "step": 6428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5587709676531205e-06,
+      "loss": 0.7093,
+      "step": 6429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5565425231261567e-06,
+      "loss": 0.7663,
+      "step": 6430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.554315538206943e-06,
+      "loss": 0.6882,
+      "step": 6431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5520900132804594e-06,
+      "loss": 0.6452,
+      "step": 6432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5498659487314228e-06,
+      "loss": 0.7253,
+      "step": 6433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5476433449443074e-06,
+      "loss": 0.668,
+      "step": 6434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5454222023033315e-06,
+      "loss": 0.7412,
+      "step": 6435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5432025211924605e-06,
+      "loss": 0.6884,
+      "step": 6436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5409843019954097e-06,
+      "loss": 0.6659,
+      "step": 6437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5387675450956341e-06,
+      "loss": 0.7251,
+      "step": 6438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5365522508763452e-06,
+      "loss": 0.6472,
+      "step": 6439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5343384197204924e-06,
+      "loss": 0.7311,
+      "step": 6440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.532126052010786e-06,
+      "loss": 0.5993,
+      "step": 6441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.529915148129668e-06,
+      "loss": 0.7028,
+      "step": 6442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5277057084593372e-06,
+      "loss": 0.7868,
+      "step": 6443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5254977333817323e-06,
+      "loss": 0.7107,
+      "step": 6444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523291223278548e-06,
+      "loss": 0.7008,
+      "step": 6445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5210861785312125e-06,
+      "loss": 0.6651,
+      "step": 6446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5188825995209156e-06,
+      "loss": 0.6778,
+      "step": 6447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5166804866285867e-06,
+      "loss": 0.7216,
+      "step": 6448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5144798402348981e-06,
+      "loss": 0.6998,
+      "step": 6449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5122806607202756e-06,
+      "loss": 0.7061,
+      "step": 6450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5100829484648838e-06,
+      "loss": 0.7556,
+      "step": 6451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.50788670384864e-06,
+      "loss": 0.6242,
+      "step": 6452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5056919272512061e-06,
+      "loss": 0.7619,
+      "step": 6453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5034986190519918e-06,
+      "loss": 0.7524,
+      "step": 6454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5013067796301473e-06,
+      "loss": 0.6597,
+      "step": 6455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499116409364576e-06,
+      "loss": 0.6811,
+      "step": 6456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4969275086339185e-06,
+      "loss": 0.702,
+      "step": 6457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4947400778165722e-06,
+      "loss": 0.7205,
+      "step": 6458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.492554117290672e-06,
+      "loss": 0.6808,
+      "step": 6459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.490369627434103e-06,
+      "loss": 0.7134,
+      "step": 6460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4881866086244957e-06,
+      "loss": 0.7131,
+      "step": 6461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.486005061239223e-06,
+      "loss": 0.685,
+      "step": 6462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4838249856554076e-06,
+      "loss": 0.7316,
+      "step": 6463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4816463822499127e-06,
+      "loss": 0.6854,
+      "step": 6464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4794692513993525e-06,
+      "loss": 0.7491,
+      "step": 6465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4772935934800847e-06,
+      "loss": 0.6489,
+      "step": 6466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475119408868213e-06,
+      "loss": 0.6249,
+      "step": 6467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4729466979395823e-06,
+      "loss": 0.6993,
+      "step": 6468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4707754610697877e-06,
+      "loss": 0.6925,
+      "step": 6469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468605698634169e-06,
+      "loss": 0.6644,
+      "step": 6470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4664374110078062e-06,
+      "loss": 0.7204,
+      "step": 6471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4642705985655291e-06,
+      "loss": 0.6572,
+      "step": 6472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462105261681912e-06,
+      "loss": 0.6996,
+      "step": 6473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4599414007312762e-06,
+      "loss": 0.6202,
+      "step": 6474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577790160876803e-06,
+      "loss": 0.646,
+      "step": 6475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4556181081249364e-06,
+      "loss": 0.6397,
+      "step": 6476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4534586772165938e-06,
+      "loss": 0.7638,
+      "step": 6477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451300723735951e-06,
+      "loss": 0.7466,
+      "step": 6478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491442480560514e-06,
+      "loss": 0.7045,
+      "step": 6479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446989250549683e-06,
+      "loss": 0.6054,
+      "step": 6480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4448357315893736e-06,
+      "loss": 0.6521,
+      "step": 6481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4426836915473997e-06,
+      "loss": 0.7098,
+      "step": 6482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4405331307957848e-06,
+      "loss": 0.6254,
+      "step": 6483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.438384049706285e-06,
+      "loss": 0.6123,
+      "step": 6484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4362364486504188e-06,
+      "loss": 0.7178,
+      "step": 6485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4340903279994312e-06,
+      "loss": 0.6652,
+      "step": 6486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4319456881243242e-06,
+      "loss": 0.6864,
+      "step": 6487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.429802529395833e-06,
+      "loss": 0.6258,
+      "step": 6488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4276608521844482e-06,
+      "loss": 0.6555,
+      "step": 6489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4255206568603897e-06,
+      "loss": 0.6829,
+      "step": 6490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4233819437936392e-06,
+      "loss": 0.7079,
+      "step": 6491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4212447133539077e-06,
+      "loss": 0.6739,
+      "step": 6492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419108965910655e-06,
+      "loss": 0.7133,
+      "step": 6493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4169747018330882e-06,
+      "loss": 0.7217,
+      "step": 6494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.41484192149015e-06,
+      "loss": 0.746,
+      "step": 6495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4127106252505317e-06,
+      "loss": 0.7155,
+      "step": 6496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4105808134826693e-06,
+      "loss": 0.6746,
+      "step": 6497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40845248655474e-06,
+      "loss": 0.7386,
+      "step": 6498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.406325644834663e-06,
+      "loss": 0.7044,
+      "step": 6499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404200288690103e-06,
+      "loss": 0.7128,
+      "step": 6500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402076418488466e-06,
+      "loss": 0.7933,
+      "step": 6501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3999540345969031e-06,
+      "loss": 0.624,
+      "step": 6502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3978331373823073e-06,
+      "loss": 0.677,
+      "step": 6503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3957137272113141e-06,
+      "loss": 0.7288,
+      "step": 6504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3935958044503072e-06,
+      "loss": 0.5649,
+      "step": 6505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3914793694654028e-06,
+      "loss": 0.6472,
+      "step": 6506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3893644226224713e-06,
+      "loss": 0.713,
+      "step": 6507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3872509642871146e-06,
+      "loss": 0.7244,
+      "step": 6508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3851389948246851e-06,
+      "loss": 0.6407,
+      "step": 6509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3830285146002775e-06,
+      "loss": 0.6669,
+      "step": 6510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3809195239787288e-06,
+      "loss": 0.72,
+      "step": 6511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3788120233246115e-06,
+      "loss": 0.711,
+      "step": 6512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3767060130022525e-06,
+      "loss": 0.722,
+      "step": 6513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.374601493375709e-06,
+      "loss": 0.2757,
+      "step": 6514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372498464808787e-06,
+      "loss": 0.6542,
+      "step": 6515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3703969276650365e-06,
+      "loss": 0.6903,
+      "step": 6516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368296882307746e-06,
+      "loss": 0.7222,
+      "step": 6517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366198329099948e-06,
+      "loss": 0.2258,
+      "step": 6518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3641012684044141e-06,
+      "loss": 0.7254,
+      "step": 6519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3620057005836628e-06,
+      "loss": 0.7196,
+      "step": 6520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3599116259999489e-06,
+      "loss": 0.6629,
+      "step": 6521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3578190450152727e-06,
+      "loss": 0.7433,
+      "step": 6522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355727957991375e-06,
+      "loss": 0.6886,
+      "step": 6523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3536383652897433e-06,
+      "loss": 0.6512,
+      "step": 6524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3515502672715963e-06,
+      "loss": 0.6817,
+      "step": 6525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3494636642979043e-06,
+      "loss": 0.7285,
+      "step": 6526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3473785567293717e-06,
+      "loss": 0.6734,
+      "step": 6527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3452949449264507e-06,
+      "loss": 0.7345,
+      "step": 6528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3432128292493317e-06,
+      "loss": 0.7122,
+      "step": 6529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.341132210057946e-06,
+      "loss": 0.6568,
+      "step": 6530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3390530877119701e-06,
+      "loss": 0.7438,
+      "step": 6531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3369754625708142e-06,
+      "loss": 0.7109,
+      "step": 6532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3348993349936379e-06,
+      "loss": 0.712,
+      "step": 6533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332824705339335e-06,
+      "loss": 0.7323,
+      "step": 6534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3307515739665455e-06,
+      "loss": 0.7377,
+      "step": 6535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3286799412336482e-06,
+      "loss": 0.6964,
+      "step": 6536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3266098074987654e-06,
+      "loss": 0.7008,
+      "step": 6537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3245411731197543e-06,
+      "loss": 0.7265,
+      "step": 6538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322474038454218e-06,
+      "loss": 0.7234,
+      "step": 6539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3204084038595e-06,
+      "loss": 0.7235,
+      "step": 6540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3183442696926795e-06,
+      "loss": 0.7094,
+      "step": 6541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3162816363105867e-06,
+      "loss": 0.7353,
+      "step": 6542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3142205040697808e-06,
+      "loss": 0.7229,
+      "step": 6543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3121608733265712e-06,
+      "loss": 0.7358,
+      "step": 6544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101027444369974e-06,
+      "loss": 0.7666,
+      "step": 6545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3080461177568471e-06,
+      "loss": 0.7121,
+      "step": 6546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3059909936416481e-06,
+      "loss": 0.6775,
+      "step": 6547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3039373724466675e-06,
+      "loss": 0.7241,
+      "step": 6548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301885254526908e-06,
+      "loss": 0.6122,
+      "step": 6549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2998346402371177e-06,
+      "loss": 0.6623,
+      "step": 6550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2977855299317854e-06,
+      "loss": 0.659,
+      "step": 6551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2957379239651334e-06,
+      "loss": 0.2449,
+      "step": 6552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29369182269113e-06,
+      "loss": 0.7025,
+      "step": 6553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2916472264634827e-06,
+      "loss": 0.6972,
+      "step": 6554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2896041356356404e-06,
+      "loss": 0.6675,
+      "step": 6555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2875625505607825e-06,
+      "loss": 0.7041,
+      "step": 6556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2855224715918413e-06,
+      "loss": 0.6443,
+      "step": 6557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.283483899081477e-06,
+      "loss": 0.614,
+      "step": 6558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2814468333820952e-06,
+      "loss": 0.7622,
+      "step": 6559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279411274845843e-06,
+      "loss": 0.2289,
+      "step": 6560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2773772238246051e-06,
+      "loss": 0.6623,
+      "step": 6561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275344680670001e-06,
+      "loss": 0.7694,
+      "step": 6562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273313645733395e-06,
+      "loss": 0.7096,
+      "step": 6563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2712841193658908e-06,
+      "loss": 0.7272,
+      "step": 6564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269256101918327e-06,
+      "loss": 0.7567,
+      "step": 6565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2672295937412836e-06,
+      "loss": 0.5968,
+      "step": 6566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2652045951850822e-06,
+      "loss": 0.7313,
+      "step": 6567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2631811065997835e-06,
+      "loss": 0.7434,
+      "step": 6568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2611591283351786e-06,
+      "loss": 0.724,
+      "step": 6569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2591386607408095e-06,
+      "loss": 0.6078,
+      "step": 6570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2571197041659466e-06,
+      "loss": 0.702,
+      "step": 6571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2551022589596062e-06,
+      "loss": 0.7251,
+      "step": 6572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2530863254705405e-06,
+      "loss": 0.7256,
+      "step": 6573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251071904047243e-06,
+      "loss": 0.6698,
+      "step": 6574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2490589950379429e-06,
+      "loss": 0.7284,
+      "step": 6575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2470475987906061e-06,
+      "loss": 0.6616,
+      "step": 6576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2450377156529437e-06,
+      "loss": 0.6987,
+      "step": 6577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2430293459723975e-06,
+      "loss": 0.7106,
+      "step": 6578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2410224900961521e-06,
+      "loss": 0.6671,
+      "step": 6579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2390171483711299e-06,
+      "loss": 0.7162,
+      "step": 6580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2370133211439949e-06,
+      "loss": 0.7074,
+      "step": 6581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.23501100876114e-06,
+      "loss": 0.6699,
+      "step": 6582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.233010211568706e-06,
+      "loss": 0.6404,
+      "step": 6583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.231010929912564e-06,
+      "loss": 0.7119,
+      "step": 6584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290131641383274e-06,
+      "loss": 0.2606,
+      "step": 6585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2270169145913525e-06,
+      "loss": 0.6073,
+      "step": 6586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2250221816167217e-06,
+      "loss": 0.7932,
+      "step": 6587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2230289655592663e-06,
+      "loss": 0.6412,
+      "step": 6588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2210372667635461e-06,
+      "loss": 0.7551,
+      "step": 6589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2190470855738656e-06,
+      "loss": 0.6798,
+      "step": 6590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2170584223342607e-06,
+      "loss": 0.7091,
+      "step": 6591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2150712773885153e-06,
+      "loss": 0.6406,
+      "step": 6592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2130856510801393e-06,
+      "loss": 0.7485,
+      "step": 6593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2111015437523877e-06,
+      "loss": 0.6704,
+      "step": 6594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2091189557482463e-06,
+      "loss": 0.7068,
+      "step": 6595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2071378874104467e-06,
+      "loss": 0.6919,
+      "step": 6596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.205158339081447e-06,
+      "loss": 0.7277,
+      "step": 6597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2031803111034534e-06,
+      "loss": 0.6693,
+      "step": 6598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201203803818407e-06,
+      "loss": 0.6897,
+      "step": 6599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1992288175679779e-06,
+      "loss": 0.6809,
+      "step": 6600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1972553526935838e-06,
+      "loss": 0.6798,
+      "step": 6601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1952834095363708e-06,
+      "loss": 0.6878,
+      "step": 6602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1933129884372263e-06,
+      "loss": 0.7282,
+      "step": 6603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1913440897367756e-06,
+      "loss": 0.7238,
+      "step": 6604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1893767137753809e-06,
+      "loss": 0.7639,
+      "step": 6605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1874108608931345e-06,
+      "loss": 0.6119,
+      "step": 6606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1854465314298758e-06,
+      "loss": 0.6377,
+      "step": 6607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1834837257251696e-06,
+      "loss": 0.6321,
+      "step": 6608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.181522444118328e-06,
+      "loss": 0.6546,
+      "step": 6609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1795626869483924e-06,
+      "loss": 0.7451,
+      "step": 6610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1776044545541442e-06,
+      "loss": 0.7093,
+      "step": 6611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1756477472741012e-06,
+      "loss": 0.7246,
+      "step": 6612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1736925654465137e-06,
+      "loss": 0.6933,
+      "step": 6613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1717389094093735e-06,
+      "loss": 0.7144,
+      "step": 6614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1697867795004036e-06,
+      "loss": 0.2627,
+      "step": 6615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1678361760570667e-06,
+      "loss": 0.708,
+      "step": 6616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1658870994165605e-06,
+      "loss": 0.6692,
+      "step": 6617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1639395499158224e-06,
+      "loss": 0.6732,
+      "step": 6618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1619935278915174e-06,
+      "loss": 0.6591,
+      "step": 6619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160049033680053e-06,
+      "loss": 0.6847,
+      "step": 6620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1581060676175748e-06,
+      "loss": 0.7933,
+      "step": 6621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1561646300399531e-06,
+      "loss": 0.6747,
+      "step": 6622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1542247212828062e-06,
+      "loss": 0.6993,
+      "step": 6623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1522863416814823e-06,
+      "loss": 0.6959,
+      "step": 6624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1503494915710689e-06,
+      "loss": 0.7247,
+      "step": 6625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1484141712863805e-06,
+      "loss": 0.244,
+      "step": 6626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1464803811619784e-06,
+      "loss": 0.5982,
+      "step": 6627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1445481215321497e-06,
+      "loss": 0.6838,
+      "step": 6628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1426173927309226e-06,
+      "loss": 0.7256,
+      "step": 6629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14068819509206e-06,
+      "loss": 0.7473,
+      "step": 6630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.138760528949061e-06,
+      "loss": 0.7924,
+      "step": 6631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1368343946351545e-06,
+      "loss": 0.6864,
+      "step": 6632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1349097924833107e-06,
+      "loss": 0.749,
+      "step": 6633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1329867228262348e-06,
+      "loss": 0.6529,
+      "step": 6634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1310651859963607e-06,
+      "loss": 0.6712,
+      "step": 6635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291451823258626e-06,
+      "loss": 0.7223,
+      "step": 6636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1272267121466507e-06,
+      "loss": 0.7019,
+      "step": 6637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1253097757903675e-06,
+      "loss": 0.7301,
+      "step": 6638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.123394373588389e-06,
+      "loss": 0.6888,
+      "step": 6639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214805058718314e-06,
+      "loss": 0.6564,
+      "step": 6640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1195681729715347e-06,
+      "loss": 0.7673,
+      "step": 6641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1176573752180908e-06,
+      "loss": 0.7037,
+      "step": 6642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1157481129418102e-06,
+      "loss": 0.7272,
+      "step": 6643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1138403864727454e-06,
+      "loss": 0.2638,
+      "step": 6644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1119341961406848e-06,
+      "loss": 0.627,
+      "step": 6645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110029542275144e-06,
+      "loss": 0.6817,
+      "step": 6646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1081264252053804e-06,
+      "loss": 0.2664,
+      "step": 6647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1062248452603819e-06,
+      "loss": 0.5968,
+      "step": 6648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1043248027688746e-06,
+      "loss": 0.7268,
+      "step": 6649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1024262980593126e-06,
+      "loss": 0.7279,
+      "step": 6650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1005293314598896e-06,
+      "loss": 0.7075,
+      "step": 6651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0986339032985294e-06,
+      "loss": 0.607,
+      "step": 6652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0967400139028927e-06,
+      "loss": 0.5906,
+      "step": 6653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0948476636003734e-06,
+      "loss": 0.6106,
+      "step": 6654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0929568527181e-06,
+      "loss": 0.2505,
+      "step": 6655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0910675815829353e-06,
+      "loss": 0.6966,
+      "step": 6656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0891798505214712e-06,
+      "loss": 0.7344,
+      "step": 6657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0872936598600403e-06,
+      "loss": 0.7383,
+      "step": 6658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0854090099247016e-06,
+      "loss": 0.759,
+      "step": 6659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0835259010412547e-06,
+      "loss": 0.683,
+      "step": 6660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0816443335352288e-06,
+      "loss": 0.7107,
+      "step": 6661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0797643077318899e-06,
+      "loss": 0.7047,
+      "step": 6662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778858239562305e-06,
+      "loss": 0.696,
+      "step": 6663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0760088825329872e-06,
+      "loss": 0.6243,
+      "step": 6664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.074133483786619e-06,
+      "loss": 0.267,
+      "step": 6665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0722596280413244e-06,
+      "loss": 0.7399,
+      "step": 6666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0703873156210342e-06,
+      "loss": 0.7112,
+      "step": 6667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068516546849414e-06,
+      "loss": 0.6378,
+      "step": 6668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0666473220498608e-06,
+      "loss": 0.7321,
+      "step": 6669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0647796415455014e-06,
+      "loss": 0.6191,
+      "step": 6670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0629135056592032e-06,
+      "loss": 0.708,
+      "step": 6671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0610489147135571e-06,
+      "loss": 0.6823,
+      "step": 6672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0591858690308954e-06,
+      "loss": 0.7035,
+      "step": 6673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.05732436893328e-06,
+      "loss": 0.6992,
+      "step": 6674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0554644147425076e-06,
+      "loss": 0.7542,
+      "step": 6675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0536060067801001e-06,
+      "loss": 0.7016,
+      "step": 6676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0517491453673234e-06,
+      "loss": 0.6968,
+      "step": 6677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0498938308251661e-06,
+      "loss": 0.6711,
+      "step": 6678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0480400634743549e-06,
+      "loss": 0.7586,
+      "step": 6679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.046187843635349e-06,
+      "loss": 0.722,
+      "step": 6680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0443371716283368e-06,
+      "loss": 0.7339,
+      "step": 6681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0424880477732457e-06,
+      "loss": 0.7162,
+      "step": 6682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0406404723897246e-06,
+      "loss": 0.7467,
+      "step": 6683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0387944457971666e-06,
+      "loss": 0.6679,
+      "step": 6684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0369499683146867e-06,
+      "loss": 0.6347,
+      "step": 6685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0351070402611396e-06,
+      "loss": 0.7608,
+      "step": 6686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0332656619551097e-06,
+      "loss": 0.6797,
+      "step": 6687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0314258337149142e-06,
+      "loss": 0.6336,
+      "step": 6688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0295875558585976e-06,
+      "loss": 0.7457,
+      "step": 6689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0277508287039429e-06,
+      "loss": 0.6663,
+      "step": 6690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0259156525684644e-06,
+      "loss": 0.7328,
+      "step": 6691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0240820277693996e-06,
+      "loss": 0.6336,
+      "step": 6692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.022249954623733e-06,
+      "loss": 0.6647,
+      "step": 6693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0204194334481665e-06,
+      "loss": 0.657,
+      "step": 6694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0185904645591428e-06,
+      "loss": 0.7371,
+      "step": 6695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167630482728286e-06,
+      "loss": 0.6709,
+      "step": 6696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0149371849051316e-06,
+      "loss": 0.6989,
+      "step": 6697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01311287477168e-06,
+      "loss": 0.7374,
+      "step": 6698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0112901181878477e-06,
+      "loss": 0.611,
+      "step": 6699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0094689154687254e-06,
+      "loss": 0.7146,
+      "step": 6700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0076492669291427e-06,
+      "loss": 0.7667,
+      "step": 6701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0058311728836635e-06,
+      "loss": 0.7015,
+      "step": 6702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.004014633646574e-06,
+      "loss": 0.6675,
+      "step": 6703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0021996495318986e-06,
+      "loss": 0.7055,
+      "step": 6704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.00038622085339e-06,
+      "loss": 0.7231,
+      "step": 6705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.98574347924537e-07,
+      "loss": 0.6061,
+      "step": 6706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.96764031058549e-07,
+      "loss": 0.7316,
+      "step": 6707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.949552705683785e-07,
+      "loss": 0.7361,
+      "step": 6708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.931480667666982e-07,
+      "loss": 0.714,
+      "step": 6709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.913424199659182e-07,
+      "loss": 0.6756,
+      "step": 6710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.895383304781803e-07,
+      "loss": 0.7082,
+      "step": 6711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.877357986153545e-07,
+      "loss": 0.7676,
+      "step": 6712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.859348246890387e-07,
+      "loss": 0.7268,
+      "step": 6713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84135409010566e-07,
+      "loss": 0.6381,
+      "step": 6714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.823375518910006e-07,
+      "loss": 0.7239,
+      "step": 6715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.805412536411317e-07,
+      "loss": 0.7195,
+      "step": 6716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78746514571486e-07,
+      "loss": 0.7124,
+      "step": 6717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.769533349923144e-07,
+      "loss": 0.6325,
+      "step": 6718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.751617152136062e-07,
+      "loss": 0.2714,
+      "step": 6719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73371655545069e-07,
+      "loss": 0.7032,
+      "step": 6720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71583156296154e-07,
+      "loss": 0.6638,
+      "step": 6721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.697962177760323e-07,
+      "loss": 0.7421,
+      "step": 6722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.680108402936107e-07,
+      "loss": 0.6493,
+      "step": 6723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.662270241575233e-07,
+      "loss": 0.7691,
+      "step": 6724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.644447696761382e-07,
+      "loss": 0.277,
+      "step": 6725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.626640771575512e-07,
+      "loss": 0.622,
+      "step": 6726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.608849469095849e-07,
+      "loss": 0.7817,
+      "step": 6727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.591073792397976e-07,
+      "loss": 0.7034,
+      "step": 6728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.573313744554736e-07,
+      "loss": 0.6916,
+      "step": 6729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.555569328636272e-07,
+      "loss": 0.6744,
+      "step": 6730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.537840547710042e-07,
+      "loss": 0.7574,
+      "step": 6731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.520127404840818e-07,
+      "loss": 0.6544,
+      "step": 6732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.502429903090604e-07,
+      "loss": 0.6911,
+      "step": 6733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.484748045518777e-07,
+      "loss": 0.6967,
+      "step": 6734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467081835181947e-07,
+      "loss": 0.7127,
+      "step": 6735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.449431275134024e-07,
+      "loss": 0.6858,
+      "step": 6736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.431796368426304e-07,
+      "loss": 0.7203,
+      "step": 6737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.414177118107237e-07,
+      "loss": 0.7563,
+      "step": 6738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.396573527222696e-07,
+      "loss": 0.6518,
+      "step": 6739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.378985598815737e-07,
+      "loss": 0.6223,
+      "step": 6740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.361413335926795e-07,
+      "loss": 0.7181,
+      "step": 6741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.343856741593504e-07,
+      "loss": 0.7313,
+      "step": 6742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.326315818850929e-07,
+      "loss": 0.6487,
+      "step": 6743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.308790570731274e-07,
+      "loss": 0.7189,
+      "step": 6744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.291281000264151e-07,
+      "loss": 0.6887,
+      "step": 6745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.273787110476363e-07,
+      "loss": 0.7672,
+      "step": 6746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256308904392086e-07,
+      "loss": 0.7374,
+      "step": 6747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.238846385032741e-07,
+      "loss": 0.6499,
+      "step": 6748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.221399555417054e-07,
+      "loss": 0.7203,
+      "step": 6749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.203968418561038e-07,
+      "loss": 0.6469,
+      "step": 6750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.186552977477969e-07,
+      "loss": 0.6654,
+      "step": 6751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.169153235178452e-07,
+      "loss": 0.6828,
+      "step": 6752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151769194670323e-07,
+      "loss": 0.7287,
+      "step": 6753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.134400858958747e-07,
+      "loss": 0.6948,
+      "step": 6754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.117048231046166e-07,
+      "loss": 0.7307,
+      "step": 6755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.099711313932325e-07,
+      "loss": 0.6982,
+      "step": 6756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.082390110614181e-07,
+      "loss": 0.7236,
+      "step": 6757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.065084624086073e-07,
+      "loss": 0.6526,
+      "step": 6758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.047794857339543e-07,
+      "loss": 0.6802,
+      "step": 6759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030520813363442e-07,
+      "loss": 0.7198,
+      "step": 6760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.013262495143938e-07,
+      "loss": 0.6609,
+      "step": 6761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.996019905664421e-07,
+      "loss": 0.6695,
+      "step": 6762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978793047905632e-07,
+      "loss": 0.7101,
+      "step": 6763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.961581924845509e-07,
+      "loss": 0.7265,
+      "step": 6764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.944386539459338e-07,
+      "loss": 0.7539,
+      "step": 6765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.927206894719642e-07,
+      "loss": 0.6896,
+      "step": 6766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.910042993596246e-07,
+      "loss": 0.6162,
+      "step": 6767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.892894839056243e-07,
+      "loss": 0.6952,
+      "step": 6768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87576243406404e-07,
+      "loss": 0.7184,
+      "step": 6769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.858645781581254e-07,
+      "loss": 0.65,
+      "step": 6770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.841544884566822e-07,
+      "loss": 0.7391,
+      "step": 6771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824459745976965e-07,
+      "loss": 0.6401,
+      "step": 6772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.807390368765145e-07,
+      "loss": 0.6899,
+      "step": 6773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.790336755882134e-07,
+      "loss": 0.6897,
+      "step": 6774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773298910275951e-07,
+      "loss": 0.6169,
+      "step": 6775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.75627683489193e-07,
+      "loss": 0.6526,
+      "step": 6776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.739270532672628e-07,
+      "loss": 0.7412,
+      "step": 6777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.722280006557904e-07,
+      "loss": 0.7027,
+      "step": 6778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.705305259484875e-07,
+      "loss": 0.6798,
+      "step": 6779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.68834629438795e-07,
+      "loss": 0.6708,
+      "step": 6780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.671403114198806e-07,
+      "loss": 0.6552,
+      "step": 6781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.654475721846367e-07,
+      "loss": 0.6486,
+      "step": 6782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.63756412025688e-07,
+      "loss": 0.2363,
+      "step": 6783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.620668312353786e-07,
+      "loss": 0.7309,
+      "step": 6784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.603788301057881e-07,
+      "loss": 0.7156,
+      "step": 6785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58692408928714e-07,
+      "loss": 0.6459,
+      "step": 6786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.570075679956868e-07,
+      "loss": 0.6558,
+      "step": 6787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.553243075979645e-07,
+      "loss": 0.6816,
+      "step": 6788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.536426280265297e-07,
+      "loss": 0.6986,
+      "step": 6789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.51962529572089e-07,
+      "loss": 0.699,
+      "step": 6790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.502840125250821e-07,
+      "loss": 0.6496,
+      "step": 6791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.486070771756682e-07,
+      "loss": 0.7155,
+      "step": 6792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.469317238137387e-07,
+      "loss": 0.7258,
+      "step": 6793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.452579527289073e-07,
+      "loss": 0.7197,
+      "step": 6794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.435857642105194e-07,
+      "loss": 0.69,
+      "step": 6795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.419151585476426e-07,
+      "loss": 0.7035,
+      "step": 6796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.402461360290704e-07,
+      "loss": 0.6258,
+      "step": 6797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385786969433274e-07,
+      "loss": 0.6861,
+      "step": 6798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.369128415786554e-07,
+      "loss": 0.7026,
+      "step": 6799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.352485702230351e-07,
+      "loss": 0.6993,
+      "step": 6800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.335858831641619e-07,
+      "loss": 0.7331,
+      "step": 6801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319247806894659e-07,
+      "loss": 0.6038,
+      "step": 6802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.302652630860952e-07,
+      "loss": 0.6739,
+      "step": 6803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.286073306409304e-07,
+      "loss": 0.6309,
+      "step": 6804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269509836405754e-07,
+      "loss": 0.6576,
+      "step": 6805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.252962223713601e-07,
+      "loss": 0.6833,
+      "step": 6806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.236430471193435e-07,
+      "loss": 0.721,
+      "step": 6807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.219914581703026e-07,
+      "loss": 0.6091,
+      "step": 6808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.203414558097489e-07,
+      "loss": 0.7699,
+      "step": 6809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.186930403229131e-07,
+      "loss": 0.6594,
+      "step": 6810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.170462119947553e-07,
+      "loss": 0.6103,
+      "step": 6811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154009711099608e-07,
+      "loss": 0.6938,
+      "step": 6812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.137573179529412e-07,
+      "loss": 0.7001,
+      "step": 6813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.121152528078291e-07,
+      "loss": 0.2516,
+      "step": 6814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.104747759584886e-07,
+      "loss": 0.2641,
+      "step": 6815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.08835887688505e-07,
+      "loss": 0.7066,
+      "step": 6816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.071985882811895e-07,
+      "loss": 0.7078,
+      "step": 6817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.055628780195813e-07,
+      "loss": 0.6691,
+      "step": 6818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.039287571864418e-07,
+      "loss": 0.7419,
+      "step": 6819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.02296226064263e-07,
+      "loss": 0.7213,
+      "step": 6820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.006652849352526e-07,
+      "loss": 0.6875,
+      "step": 6821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.990359340813525e-07,
+      "loss": 0.7154,
+      "step": 6822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.974081737842243e-07,
+      "loss": 0.6968,
+      "step": 6823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.957820043252562e-07,
+      "loss": 0.7072,
+      "step": 6824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.941574259855634e-07,
+      "loss": 0.2538,
+      "step": 6825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.925344390459844e-07,
+      "loss": 0.7413,
+      "step": 6826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.909130437870805e-07,
+      "loss": 0.6525,
+      "step": 6827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.892932404891429e-07,
+      "loss": 0.7016,
+      "step": 6828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.8767502943218e-07,
+      "loss": 0.6895,
+      "step": 6829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.860584108959324e-07,
+      "loss": 0.6759,
+      "step": 6830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.844433851598632e-07,
+      "loss": 0.6842,
+      "step": 6831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.828299525031568e-07,
+      "loss": 0.7162,
+      "step": 6832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.812181132047292e-07,
+      "loss": 0.6643,
+      "step": 6833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.796078675432128e-07,
+      "loss": 0.6775,
+      "step": 6834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779992157969696e-07,
+      "loss": 0.6703,
+      "step": 6835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.763921582440847e-07,
+      "loss": 0.7188,
+      "step": 6836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.747866951623662e-07,
+      "loss": 0.2633,
+      "step": 6837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.731828268293506e-07,
+      "loss": 0.7091,
+      "step": 6838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.715805535222964e-07,
+      "loss": 0.6953,
+      "step": 6839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.699798755181831e-07,
+      "loss": 0.7205,
+      "step": 6840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683807930937204e-07,
+      "loss": 0.6498,
+      "step": 6841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.667833065253394e-07,
+      "loss": 0.6308,
+      "step": 6842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.651874160891903e-07,
+      "loss": 0.679,
+      "step": 6843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.635931220611592e-07,
+      "loss": 0.7077,
+      "step": 6844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.620004247168444e-07,
+      "loss": 0.7483,
+      "step": 6845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.604093243315769e-07,
+      "loss": 0.6991,
+      "step": 6846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.588198211804043e-07,
+      "loss": 0.6015,
+      "step": 6847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.572319155381036e-07,
+      "loss": 0.6875,
+      "step": 6848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.556456076791707e-07,
+      "loss": 0.6969,
+      "step": 6849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54060897877833e-07,
+      "loss": 0.746,
+      "step": 6850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.524777864080334e-07,
+      "loss": 0.6594,
+      "step": 6851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.508962735434433e-07,
+      "loss": 0.6474,
+      "step": 6852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.493163595574571e-07,
+      "loss": 0.7221,
+      "step": 6853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.477380447231908e-07,
+      "loss": 0.7105,
+      "step": 6854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46161329313485e-07,
+      "loss": 0.7267,
+      "step": 6855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.445862136009052e-07,
+      "loss": 0.7841,
+      "step": 6856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.430126978577401e-07,
+      "loss": 0.6655,
+      "step": 6857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.414407823559977e-07,
+      "loss": 0.7482,
+      "step": 6858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398704673674173e-07,
+      "loss": 0.7483,
+      "step": 6859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.383017531634529e-07,
+      "loss": 0.7223,
+      "step": 6860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.367346400152875e-07,
+      "loss": 0.6411,
+      "step": 6861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.351691281938244e-07,
+      "loss": 0.5771,
+      "step": 6862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.336052179696929e-07,
+      "loss": 0.7033,
+      "step": 6863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.320429096132442e-07,
+      "loss": 0.7028,
+      "step": 6864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.304822033945502e-07,
+      "loss": 0.6635,
+      "step": 6865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289230995834096e-07,
+      "loss": 0.6793,
+      "step": 6866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.273655984493399e-07,
+      "loss": 0.2419,
+      "step": 6867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25809700261586e-07,
+      "loss": 0.6598,
+      "step": 6868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242554052891137e-07,
+      "loss": 0.63,
+      "step": 6869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.227027138006115e-07,
+      "loss": 0.7314,
+      "step": 6870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.211516260644891e-07,
+      "loss": 0.7266,
+      "step": 6871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196021423488842e-07,
+      "loss": 0.2394,
+      "step": 6872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.180542629216492e-07,
+      "loss": 0.7231,
+      "step": 6873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165079880503667e-07,
+      "loss": 0.6529,
+      "step": 6874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149633180023375e-07,
+      "loss": 0.6562,
+      "step": 6875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.134202530445877e-07,
+      "loss": 0.7152,
+      "step": 6876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.118787934438653e-07,
+      "loss": 0.6709,
+      "step": 6877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103389394666382e-07,
+      "loss": 0.715,
+      "step": 6878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088006913791001e-07,
+      "loss": 0.6955,
+      "step": 6879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.072640494471628e-07,
+      "loss": 0.6774,
+      "step": 6880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.057290139364658e-07,
+      "loss": 0.7117,
+      "step": 6881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.041955851123672e-07,
+      "loss": 0.6978,
+      "step": 6882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.026637632399513e-07,
+      "loss": 0.7276,
+      "step": 6883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.011335485840187e-07,
+      "loss": 0.6901,
+      "step": 6884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.996049414090967e-07,
+      "loss": 0.7233,
+      "step": 6885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.980779419794315e-07,
+      "loss": 0.6829,
+      "step": 6886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965525505589954e-07,
+      "loss": 0.7256,
+      "step": 6887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950287674114786e-07,
+      "loss": 0.6515,
+      "step": 6888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.93506592800297e-07,
+      "loss": 0.6615,
+      "step": 6889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.919860269885881e-07,
+      "loss": 0.7307,
+      "step": 6890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.904670702392047e-07,
+      "loss": 0.7328,
+      "step": 6891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.889497228147323e-07,
+      "loss": 0.5746,
+      "step": 6892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.874339849774659e-07,
+      "loss": 0.6475,
+      "step": 6893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.859198569894354e-07,
+      "loss": 0.704,
+      "step": 6894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.844073391123818e-07,
+      "loss": 0.7226,
+      "step": 6895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.828964316077747e-07,
+      "loss": 0.7117,
+      "step": 6896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.813871347367984e-07,
+      "loss": 0.697,
+      "step": 6897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.798794487603655e-07,
+      "loss": 0.694,
+      "step": 6898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.783733739391075e-07,
+      "loss": 0.7218,
+      "step": 6899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.768689105333759e-07,
+      "loss": 0.6477,
+      "step": 6900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.753660588032473e-07,
+      "loss": 0.7372,
+      "step": 6901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.738648190085151e-07,
+      "loss": 0.7445,
+      "step": 6902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.723651914086981e-07,
+      "loss": 0.7293,
+      "step": 6903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.708671762630336e-07,
+      "loss": 0.6813,
+      "step": 6904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6937077383048e-07,
+      "loss": 0.2726,
+      "step": 6905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.678759843697203e-07,
+      "loss": 0.7332,
+      "step": 6906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.663828081391566e-07,
+      "loss": 0.7075,
+      "step": 6907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.648912453969103e-07,
+      "loss": 0.7363,
+      "step": 6908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.634012964008274e-07,
+      "loss": 0.7705,
+      "step": 6909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.619129614084707e-07,
+      "loss": 0.7138,
+      "step": 6910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604262406771278e-07,
+      "loss": 0.6601,
+      "step": 6911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.589411344638063e-07,
+      "loss": 0.6875,
+      "step": 6912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.574576430252333e-07,
+      "loss": 0.6232,
+      "step": 6913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.559757666178612e-07,
+      "loss": 0.6721,
+      "step": 6914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.544955054978541e-07,
+      "loss": 0.6734,
+      "step": 6915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530168599211073e-07,
+      "loss": 0.6807,
+      "step": 6916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.515398301432285e-07,
+      "loss": 0.7149,
+      "step": 6917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.500644164195513e-07,
+      "loss": 0.7184,
+      "step": 6918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.485906190051283e-07,
+      "loss": 0.6624,
+      "step": 6919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.471184381547335e-07,
+      "loss": 0.6058,
+      "step": 6920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.456478741228578e-07,
+      "loss": 0.7046,
+      "step": 6921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.441789271637177e-07,
+      "loss": 0.6494,
+      "step": 6922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427115975312493e-07,
+      "loss": 0.7458,
+      "step": 6923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412458854791026e-07,
+      "loss": 0.7491,
+      "step": 6924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.397817912606563e-07,
+      "loss": 0.7656,
+      "step": 6925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.383193151290068e-07,
+      "loss": 0.6885,
+      "step": 6926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.368584573369707e-07,
+      "loss": 0.6474,
+      "step": 6927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.353992181370805e-07,
+      "loss": 0.7374,
+      "step": 6928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.339415977815966e-07,
+      "loss": 0.6888,
+      "step": 6929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324855965224929e-07,
+      "loss": 0.6486,
+      "step": 6930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.310312146114672e-07,
+      "loss": 0.7501,
+      "step": 6931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.29578452299936e-07,
+      "loss": 0.7101,
+      "step": 6932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.281273098390372e-07,
+      "loss": 0.6462,
+      "step": 6933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.266777874796271e-07,
+      "loss": 0.7641,
+      "step": 6934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.252298854722816e-07,
+      "loss": 0.6375,
+      "step": 6935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.237836040672995e-07,
+      "loss": 0.6483,
+      "step": 6936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223389435146943e-07,
+      "loss": 0.6558,
+      "step": 6937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.208959040642037e-07,
+      "loss": 0.6884,
+      "step": 6938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194544859652841e-07,
+      "loss": 0.6893,
+      "step": 6939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.180146894671124e-07,
+      "loss": 0.7478,
+      "step": 6940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.16576514818581e-07,
+      "loss": 0.7472,
+      "step": 6941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.151399622683074e-07,
+      "loss": 0.6561,
+      "step": 6942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.137050320646243e-07,
+      "loss": 0.6727,
+      "step": 6943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.122717244555875e-07,
+      "loss": 0.6982,
+      "step": 6944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.108400396889702e-07,
+      "loss": 0.6959,
+      "step": 6945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.094099780122664e-07,
+      "loss": 0.2445,
+      "step": 6946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.079815396726885e-07,
+      "loss": 0.6814,
+      "step": 6947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065547249171666e-07,
+      "loss": 0.7138,
+      "step": 6948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.051295339923546e-07,
+      "loss": 0.6959,
+      "step": 6949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0370596714462e-07,
+      "loss": 0.6708,
+      "step": 6950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022840246200567e-07,
+      "loss": 0.7043,
+      "step": 6951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008637066644707e-07,
+      "loss": 0.6684,
+      "step": 6952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.99445013523392e-07,
+      "loss": 0.2473,
+      "step": 6953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980279454420656e-07,
+      "loss": 0.748,
+      "step": 6954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.966125026654612e-07,
+      "loss": 0.6692,
+      "step": 6955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.951986854382596e-07,
+      "loss": 0.6737,
+      "step": 6956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.937864940048699e-07,
+      "loss": 0.623,
+      "step": 6957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.923759286094144e-07,
+      "loss": 0.6747,
+      "step": 6958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.909669894957337e-07,
+      "loss": 0.6759,
+      "step": 6959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.895596769073908e-07,
+      "loss": 0.7444,
+      "step": 6960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.881539910876643e-07,
+      "loss": 0.7414,
+      "step": 6961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.867499322795534e-07,
+      "loss": 0.6932,
+      "step": 6962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.853475007257758e-07,
+      "loss": 0.6877,
+      "step": 6963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.839466966687691e-07,
+      "loss": 0.7002,
+      "step": 6964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.825475203506847e-07,
+      "loss": 0.7533,
+      "step": 6965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811499720134006e-07,
+      "loss": 0.6818,
+      "step": 6966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.797540518985045e-07,
+      "loss": 0.7142,
+      "step": 6967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.783597602473101e-07,
+      "loss": 0.6249,
+      "step": 6968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769670973008445e-07,
+      "loss": 0.7041,
+      "step": 6969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.755760632998564e-07,
+      "loss": 0.6508,
+      "step": 6970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.741866584848122e-07,
+      "loss": 0.7815,
+      "step": 6971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.727988830958942e-07,
+      "loss": 0.6801,
+      "step": 6972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.714127373730083e-07,
+      "loss": 0.7795,
+      "step": 6973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700282215557717e-07,
+      "loss": 0.6873,
+      "step": 6974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686453358835265e-07,
+      "loss": 0.6686,
+      "step": 6975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.672640805953278e-07,
+      "loss": 0.7445,
+      "step": 6976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.658844559299537e-07,
+      "loss": 0.2474,
+      "step": 6977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.645064621258944e-07,
+      "loss": 0.714,
+      "step": 6978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63130099421364e-07,
+      "loss": 0.6897,
+      "step": 6979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617553680542931e-07,
+      "loss": 0.7595,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.603822682623272e-07,
+      "loss": 0.6749,
+      "step": 6981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.590108002828321e-07,
+      "loss": 0.7127,
+      "step": 6982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.576409643528924e-07,
+      "loss": 0.7781,
+      "step": 6983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.56272760709311e-07,
+      "loss": 0.6351,
+      "step": 6984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.549061895886043e-07,
+      "loss": 0.6323,
+      "step": 6985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.535412512270121e-07,
+      "loss": 0.704,
+      "step": 6986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521779458604859e-07,
+      "loss": 0.6679,
+      "step": 6987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.508162737247003e-07,
+      "loss": 0.6826,
+      "step": 6988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.494562350550459e-07,
+      "loss": 0.6647,
+      "step": 6989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480978300866313e-07,
+      "loss": 0.6587,
+      "step": 6990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467410590542788e-07,
+      "loss": 0.2537,
+      "step": 6991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.453859221925329e-07,
+      "loss": 0.6781,
+      "step": 6992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.440324197356561e-07,
+      "loss": 0.6577,
+      "step": 6993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.426805519176215e-07,
+      "loss": 0.6916,
+      "step": 6994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.413303189721297e-07,
+      "loss": 0.6942,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.399817211325908e-07,
+      "loss": 0.6595,
+      "step": 6996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.386347586321372e-07,
+      "loss": 0.7152,
+      "step": 6997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.372894317036114e-07,
+      "loss": 0.6858,
+      "step": 6998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.359457405795832e-07,
+      "loss": 0.7236,
+      "step": 6999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346036854923298e-07,
+      "loss": 0.7089,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332632666738558e-07,
+      "loss": 0.6751,
+      "step": 7001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319244843558713e-07,
+      "loss": 0.7407,
+      "step": 7002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.305873387698135e-07,
+      "loss": 0.6047,
+      "step": 7003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.292518301468341e-07,
+      "loss": 0.6585,
+      "step": 7004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279179587177952e-07,
+      "loss": 0.661,
+      "step": 7005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.265857247132856e-07,
+      "loss": 0.6324,
+      "step": 7006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.252551283636043e-07,
+      "loss": 0.6766,
+      "step": 7007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.239261698987708e-07,
+      "loss": 0.6834,
+      "step": 7008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.22598849548519e-07,
+      "loss": 0.691,
+      "step": 7009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212731675423033e-07,
+      "loss": 0.6605,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.19949124109288e-07,
+      "loss": 0.7121,
+      "step": 7011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.186267194783623e-07,
+      "loss": 0.5627,
+      "step": 7012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.173059538781255e-07,
+      "loss": 0.2483,
+      "step": 7013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.159868275368985e-07,
+      "loss": 0.729,
+      "step": 7014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146693406827174e-07,
+      "loss": 0.6612,
+      "step": 7015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.133534935433304e-07,
+      "loss": 0.6917,
+      "step": 7016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.120392863462109e-07,
+      "loss": 0.682,
+      "step": 7017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.107267193185394e-07,
+      "loss": 0.686,
+      "step": 7018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09415792687219e-07,
+      "loss": 0.6109,
+      "step": 7019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.081065066788682e-07,
+      "loss": 0.676,
+      "step": 7020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.067988615198227e-07,
+      "loss": 0.2342,
+      "step": 7021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.054928574361295e-07,
+      "loss": 0.6397,
+      "step": 7022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.041884946535591e-07,
+      "loss": 0.6546,
+      "step": 7023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.028857733975922e-07,
+      "loss": 0.726,
+      "step": 7024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.015846938934288e-07,
+      "loss": 0.7049,
+      "step": 7025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.002852563659855e-07,
+      "loss": 0.6803,
+      "step": 7026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.989874610398925e-07,
+      "loss": 0.6446,
+      "step": 7027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976913081395008e-07,
+      "loss": 0.7128,
+      "step": 7028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963967978888706e-07,
+      "loss": 0.7783,
+      "step": 7029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.951039305117856e-07,
+      "loss": 0.639,
+      "step": 7030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.938127062317389e-07,
+      "loss": 0.732,
+      "step": 7031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.925231252719431e-07,
+      "loss": 0.7013,
+      "step": 7032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.912351878553256e-07,
+      "loss": 0.6199,
+      "step": 7033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899488942045339e-07,
+      "loss": 0.6979,
+      "step": 7034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886642445419232e-07,
+      "loss": 0.7133,
+      "step": 7035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.873812390895716e-07,
+      "loss": 0.63,
+      "step": 7036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.860998780692683e-07,
+      "loss": 0.7713,
+      "step": 7037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.848201617025205e-07,
+      "loss": 0.6455,
+      "step": 7038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.835420902105525e-07,
+      "loss": 0.7034,
+      "step": 7039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.82265663814302e-07,
+      "loss": 0.7084,
+      "step": 7040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.809908827344234e-07,
+      "loss": 0.6405,
+      "step": 7041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.797177471912839e-07,
+      "loss": 0.7348,
+      "step": 7042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.784462574049708e-07,
+      "loss": 0.6495,
+      "step": 7043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.771764135952839e-07,
+      "loss": 0.6752,
+      "step": 7044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.759082159817385e-07,
+      "loss": 0.7018,
+      "step": 7045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.746416647835661e-07,
+      "loss": 0.7541,
+      "step": 7046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.733767602197159e-07,
+      "loss": 0.698,
+      "step": 7047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721135025088475e-07,
+      "loss": 0.6693,
+      "step": 7048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7085189186933833e-07,
+      "loss": 0.7514,
+      "step": 7049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6959192851928294e-07,
+      "loss": 0.763,
+      "step": 7050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.683336126764859e-07,
+      "loss": 0.703,
+      "step": 7051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.670769445584755e-07,
+      "loss": 0.6645,
+      "step": 7052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658219243824846e-07,
+      "loss": 0.7191,
+      "step": 7053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.645685523654719e-07,
+      "loss": 0.598,
+      "step": 7054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6331682872410187e-07,
+      "loss": 0.7106,
+      "step": 7055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.620667536747592e-07,
+      "loss": 0.7128,
+      "step": 7056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6081832743354205e-07,
+      "loss": 0.5955,
+      "step": 7057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5957155021626677e-07,
+      "loss": 0.6317,
+      "step": 7058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5832642223845867e-07,
+      "loss": 0.6787,
+      "step": 7059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.570829437153623e-07,
+      "loss": 0.6703,
+      "step": 7060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.558411148619368e-07,
+      "loss": 0.7466,
+      "step": 7061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546009358928527e-07,
+      "loss": 0.6691,
+      "step": 7062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5336240702249955e-07,
+      "loss": 0.6325,
+      "step": 7063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.521255284649806e-07,
+      "loss": 0.6941,
+      "step": 7064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5089030043411366e-07,
+      "loss": 0.7469,
+      "step": 7065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.496567231434279e-07,
+      "loss": 0.6867,
+      "step": 7066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.484247968061728e-07,
+      "loss": 0.6331,
+      "step": 7067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4719452163530797e-07,
+      "loss": 0.69,
+      "step": 7068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.459658978435089e-07,
+      "loss": 0.6626,
+      "step": 7069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.447389256431667e-07,
+      "loss": 0.7398,
+      "step": 7070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.435136052463873e-07,
+      "loss": 0.6599,
+      "step": 7071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42289936864988e-07,
+      "loss": 0.6659,
+      "step": 7072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.41067920710504e-07,
+      "loss": 0.6777,
+      "step": 7073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.398475569941829e-07,
+      "loss": 0.7014,
+      "step": 7074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.386288459269872e-07,
+      "loss": 0.7411,
+      "step": 7075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.374117877195927e-07,
+      "loss": 0.7313,
+      "step": 7076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3619638258239117e-07,
+      "loss": 0.6172,
+      "step": 7077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3498263072548897e-07,
+      "loss": 0.662,
+      "step": 7078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.337705323587027e-07,
+      "loss": 0.6912,
+      "step": 7079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.325600876915703e-07,
+      "loss": 0.6599,
+      "step": 7080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3135129693333443e-07,
+      "loss": 0.6929,
+      "step": 7081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.301441602929601e-07,
+      "loss": 0.6735,
+      "step": 7082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289386779791227e-07,
+      "loss": 0.6574,
+      "step": 7083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.277348502002121e-07,
+      "loss": 0.6977,
+      "step": 7084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.26532677164333e-07,
+      "loss": 0.6675,
+      "step": 7085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2533215907930135e-07,
+      "loss": 0.6532,
+      "step": 7086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.241332961526512e-07,
+      "loss": 0.6886,
+      "step": 7087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.229360885916245e-07,
+      "loss": 0.6445,
+      "step": 7088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.217405366031846e-07,
+      "loss": 0.6568,
+      "step": 7089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2054664039400175e-07,
+      "loss": 0.7677,
+      "step": 7090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1935440017046637e-07,
+      "loss": 0.6617,
+      "step": 7091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.18163816138677e-07,
+      "loss": 0.6891,
+      "step": 7092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.169748885044478e-07,
+      "loss": 0.6537,
+      "step": 7093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1578761747330775e-07,
+      "loss": 0.7015,
+      "step": 7094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1460200325049716e-07,
+      "loss": 0.6297,
+      "step": 7095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.134180460409742e-07,
+      "loss": 0.7094,
+      "step": 7096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.122357460494053e-07,
+      "loss": 0.6946,
+      "step": 7097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.110551034801746e-07,
+      "loss": 0.6556,
+      "step": 7098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098761185373767e-07,
+      "loss": 0.7099,
+      "step": 7099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.086987914248219e-07,
+      "loss": 0.7072,
+      "step": 7100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.075231223460296e-07,
+      "loss": 0.6754,
+      "step": 7101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.063491115042417e-07,
+      "loss": 0.709,
+      "step": 7102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.051767591024036e-07,
+      "loss": 0.6638,
+      "step": 7103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.04006065343181e-07,
+      "loss": 0.7185,
+      "step": 7104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.028370304289464e-07,
+      "loss": 0.6969,
+      "step": 7105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0166965456179263e-07,
+      "loss": 0.7031,
+      "step": 7106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0050393794351715e-07,
+      "loss": 0.7223,
+      "step": 7107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99339880775641e-07,
+      "loss": 0.691,
+      "step": 7108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.981774832593921e-07,
+      "loss": 0.7719,
+      "step": 7109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.970167455957097e-07,
+      "loss": 0.6633,
+      "step": 7110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9585766798525105e-07,
+      "loss": 0.7709,
+      "step": 7111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.947002506283826e-07,
+      "loss": 0.6579,
+      "step": 7112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.935444937251864e-07,
+      "loss": 0.716,
+      "step": 7113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9239039747545504e-07,
+      "loss": 0.6736,
+      "step": 7114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.912379620786988e-07,
+      "loss": 0.6548,
+      "step": 7115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.900871877341328e-07,
+      "loss": 0.6205,
+      "step": 7116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8893807464069455e-07,
+      "loss": 0.6979,
+      "step": 7117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.877906229970252e-07,
+      "loss": 0.6495,
+      "step": 7118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.866448330014838e-07,
+      "loss": 0.6372,
+      "step": 7119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8550070485214306e-07,
+      "loss": 0.7749,
+      "step": 7120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.843582387467859e-07,
+      "loss": 0.7126,
+      "step": 7121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8321743488290984e-07,
+      "loss": 0.7276,
+      "step": 7122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.820782934577216e-07,
+      "loss": 0.2414,
+      "step": 7123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.809408146681448e-07,
+      "loss": 0.7225,
+      "step": 7124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7980499871081213e-07,
+      "loss": 0.7259,
+      "step": 7125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.786708457820709e-07,
+      "loss": 0.6179,
+      "step": 7126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.775383560779811e-07,
+      "loss": 0.7042,
+      "step": 7127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7640752979431484e-07,
+      "loss": 0.6456,
+      "step": 7128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7527836712655473e-07,
+      "loss": 0.7209,
+      "step": 7129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7415086826989797e-07,
+      "loss": 0.6751,
+      "step": 7130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7302503341925535e-07,
+      "loss": 0.2379,
+      "step": 7131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.719008627692444e-07,
+      "loss": 0.6795,
+      "step": 7132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.70778356514202e-07,
+      "loss": 0.7361,
+      "step": 7133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6965751484817403e-07,
+      "loss": 0.6386,
+      "step": 7134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.685383379649177e-07,
+      "loss": 0.7025,
+      "step": 7135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.674208260579037e-07,
+      "loss": 0.7036,
+      "step": 7136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.663049793203155e-07,
+      "loss": 0.6675,
+      "step": 7137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.651907979450453e-07,
+      "loss": 0.6078,
+      "step": 7138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6407828212470244e-07,
+      "loss": 0.674,
+      "step": 7139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6296743205160413e-07,
+      "loss": 0.6115,
+      "step": 7140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6185824791778347e-07,
+      "loss": 0.7373,
+      "step": 7141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.607507299149815e-07,
+      "loss": 0.6476,
+      "step": 7142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.59644878234654e-07,
+      "loss": 0.6787,
+      "step": 7143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.585406930679691e-07,
+      "loss": 0.7756,
+      "step": 7144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.57438174605802e-07,
+      "loss": 0.6807,
+      "step": 7145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.563373230387468e-07,
+      "loss": 0.6947,
+      "step": 7146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5523813855710465e-07,
+      "loss": 0.5904,
+      "step": 7147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5414062135089134e-07,
+      "loss": 0.7308,
+      "step": 7148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.530447716098306e-07,
+      "loss": 0.689,
+      "step": 7149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.519505895233632e-07,
+      "loss": 0.7234,
+      "step": 7150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508580752806345e-07,
+      "loss": 0.673,
+      "step": 7151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.497672290705101e-07,
+      "loss": 0.559,
+      "step": 7152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4867805108156127e-07,
+      "loss": 0.7022,
+      "step": 7153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4759054150207196e-07,
+      "loss": 0.69,
+      "step": 7154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465047005200406e-07,
+      "loss": 0.6544,
+      "step": 7155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4542052832317154e-07,
+      "loss": 0.6792,
+      "step": 7156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.443380250988859e-07,
+      "loss": 0.7264,
+      "step": 7157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.43257191034313e-07,
+      "loss": 0.6827,
+      "step": 7158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4217802631629883e-07,
+      "loss": 0.6806,
+      "step": 7159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.411005311313931e-07,
+      "loss": 0.7119,
+      "step": 7160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.400247056658623e-07,
+      "loss": 0.6314,
+      "step": 7161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.389505501056811e-07,
+      "loss": 0.6251,
+      "step": 7162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.378780646365387e-07,
+      "loss": 0.7099,
+      "step": 7163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.368072494438346e-07,
+      "loss": 0.6228,
+      "step": 7164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3573810471267734e-07,
+      "loss": 0.7335,
+      "step": 7165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.346706306278902e-07,
+      "loss": 0.646,
+      "step": 7166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3360482737400444e-07,
+      "loss": 0.6913,
+      "step": 7167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32540695135265e-07,
+      "loss": 0.7631,
+      "step": 7168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314782340956246e-07,
+      "loss": 0.7107,
+      "step": 7169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.304174444387509e-07,
+      "loss": 0.7487,
+      "step": 7170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.293583263480216e-07,
+      "loss": 0.2734,
+      "step": 7171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.283008800065246e-07,
+      "loss": 0.2367,
+      "step": 7172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.272451055970571e-07,
+      "loss": 0.7247,
+      "step": 7173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2619100330213207e-07,
+      "loss": 0.7459,
+      "step": 7174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.25138573303968e-07,
+      "loss": 0.7037,
+      "step": 7175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2408781578449846e-07,
+      "loss": 0.6516,
+      "step": 7176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2303873092536577e-07,
+      "loss": 0.7331,
+      "step": 7177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2199131890792377e-07,
+      "loss": 0.6421,
+      "step": 7178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.209455799132377e-07,
+      "loss": 0.6795,
+      "step": 7179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1990151412208184e-07,
+      "loss": 0.6703,
+      "step": 7180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1885912171494395e-07,
+      "loss": 0.6891,
+      "step": 7181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.178184028720177e-07,
+      "loss": 0.6527,
+      "step": 7182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1677935777321255e-07,
+      "loss": 0.7072,
+      "step": 7183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.157419865981459e-07,
+      "loss": 0.6729,
+      "step": 7184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.147062895261488e-07,
+      "loss": 0.7262,
+      "step": 7185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1367226673625797e-07,
+      "loss": 0.7123,
+      "step": 7186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1263991840722596e-07,
+      "loss": 0.6774,
+      "step": 7187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.116092447175101e-07,
+      "loss": 0.718,
+      "step": 7188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1058024584528224e-07,
+      "loss": 0.7024,
+      "step": 7189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0955292196842566e-07,
+      "loss": 0.6631,
+      "step": 7190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.085272732645306e-07,
+      "loss": 0.6492,
+      "step": 7191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.075032999109018e-07,
+      "loss": 0.738,
+      "step": 7192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.064810020845488e-07,
+      "loss": 0.6801,
+      "step": 7193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.054603799621969e-07,
+      "loss": 0.6688,
+      "step": 7194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0444143372027833e-07,
+      "loss": 0.6799,
+      "step": 7195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0342416353493663e-07,
+      "loss": 0.7333,
+      "step": 7196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.024085695820267e-07,
+      "loss": 0.6578,
+      "step": 7197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.013946520371125e-07,
+      "loss": 0.2542,
+      "step": 7198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0038241107546826e-07,
+      "loss": 0.7197,
+      "step": 7199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9937184687207853e-07,
+      "loss": 0.6665,
+      "step": 7200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.98362959601638e-07,
+      "loss": 0.6624,
+      "step": 7201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.973557494385504e-07,
+      "loss": 0.6631,
+      "step": 7202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.963502165569343e-07,
+      "loss": 0.8212,
+      "step": 7203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9534636113060956e-07,
+      "loss": 0.662,
+      "step": 7204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.943441833331162e-07,
+      "loss": 0.6653,
+      "step": 7205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.933436833376946e-07,
+      "loss": 0.6689,
+      "step": 7206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.92344861317303e-07,
+      "loss": 0.771,
+      "step": 7207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.913477174446033e-07,
+      "loss": 0.6882,
+      "step": 7208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9035225189197436e-07,
+      "loss": 0.7553,
+      "step": 7209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8935846483149734e-07,
+      "loss": 0.7582,
+      "step": 7210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8836635643496703e-07,
+      "loss": 0.6857,
+      "step": 7211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8737592687389074e-07,
+      "loss": 0.2249,
+      "step": 7212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.863871763194781e-07,
+      "loss": 0.6938,
+      "step": 7213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.854001049426558e-07,
+      "loss": 0.7153,
+      "step": 7214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.844147129140562e-07,
+      "loss": 0.622,
+      "step": 7215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8343100040402415e-07,
+      "loss": 0.6834,
+      "step": 7216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8244896758261033e-07,
+      "loss": 0.7345,
+      "step": 7217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8146861461958e-07,
+      "loss": 0.6892,
+      "step": 7218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8048994168440203e-07,
+      "loss": 0.7344,
+      "step": 7219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7951294894626004e-07,
+      "loss": 0.6998,
+      "step": 7220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.785376365740444e-07,
+      "loss": 0.7382,
+      "step": 7221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7756400473635816e-07,
+      "loss": 0.6728,
+      "step": 7222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7659205360150875e-07,
+      "loss": 0.6771,
+      "step": 7223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.756217833375174e-07,
+      "loss": 0.6289,
+      "step": 7224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.746531941121133e-07,
+      "loss": 0.2526,
+      "step": 7225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7368628609273474e-07,
+      "loss": 0.7264,
+      "step": 7226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727210594465279e-07,
+      "loss": 0.7032,
+      "step": 7227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717575143403528e-07,
+      "loss": 0.7087,
+      "step": 7228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707956509407761e-07,
+      "loss": 0.6733,
+      "step": 7229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6983546941407144e-07,
+      "loss": 0.6597,
+      "step": 7230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6887696992622505e-07,
+      "loss": 0.6443,
+      "step": 7231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6792015264293094e-07,
+      "loss": 0.6988,
+      "step": 7232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.669650177295935e-07,
+      "loss": 0.7644,
+      "step": 7233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.66011565351324e-07,
+      "loss": 0.7365,
+      "step": 7234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6505979567294615e-07,
+      "loss": 0.7252,
+      "step": 7235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6410970885899056e-07,
+      "loss": 0.6325,
+      "step": 7236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6316130507369584e-07,
+      "loss": 0.7858,
+      "step": 7237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.622145844810131e-07,
+      "loss": 0.6857,
+      "step": 7238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.612695472445981e-07,
+      "loss": 0.7521,
+      "step": 7239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6032619352782007e-07,
+      "loss": 0.7668,
+      "step": 7240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.593845234937542e-07,
+      "loss": 0.6974,
+      "step": 7241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5844453730518695e-07,
+      "loss": 0.6678,
+      "step": 7242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.575062351246094e-07,
+      "loss": 0.7358,
+      "step": 7243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5656961711422844e-07,
+      "loss": 0.6315,
+      "step": 7244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.556346834359513e-07,
+      "loss": 0.7343,
+      "step": 7245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.547014342514009e-07,
+      "loss": 0.6372,
+      "step": 7246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.537698697219082e-07,
+      "loss": 0.7314,
+      "step": 7247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5283999000850876e-07,
+      "loss": 0.6374,
+      "step": 7248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.519117952719519e-07,
+      "loss": 0.6951,
+      "step": 7249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.509852856726902e-07,
+      "loss": 0.719,
+      "step": 7250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5006046137089014e-07,
+      "loss": 0.5866,
+      "step": 7251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4913732252642266e-07,
+      "loss": 0.783,
+      "step": 7252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4821586929887344e-07,
+      "loss": 0.7663,
+      "step": 7253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.472961018475284e-07,
+      "loss": 0.7248,
+      "step": 7254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.463780203313881e-07,
+      "loss": 0.659,
+      "step": 7255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4546162490916016e-07,
+      "loss": 0.7028,
+      "step": 7256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4454691573925995e-07,
+      "loss": 0.6779,
+      "step": 7257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.436338929798121e-07,
+      "loss": 0.6953,
+      "step": 7258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427225567886493e-07,
+      "loss": 0.6829,
+      "step": 7259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.418129073233144e-07,
+      "loss": 0.6838,
+      "step": 7260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4090494474105497e-07,
+      "loss": 0.6685,
+      "step": 7261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3999866919882983e-07,
+      "loss": 0.6871,
+      "step": 7262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.39094080853306e-07,
+      "loss": 0.701,
+      "step": 7263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3819117986085605e-07,
+      "loss": 0.6937,
+      "step": 7264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.372899663775663e-07,
+      "loss": 0.654,
+      "step": 7265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3639044055922766e-07,
+      "loss": 0.7227,
+      "step": 7266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.35492602561338e-07,
+      "loss": 0.7282,
+      "step": 7267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3459645253910644e-07,
+      "loss": 0.6922,
+      "step": 7268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3370199064744803e-07,
+      "loss": 0.6631,
+      "step": 7269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3280921704098792e-07,
+      "loss": 0.615,
+      "step": 7270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.319181318740582e-07,
+      "loss": 0.6584,
+      "step": 7271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3102873530070013e-07,
+      "loss": 0.5914,
+      "step": 7272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3014102747466293e-07,
+      "loss": 0.7778,
+      "step": 7273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2925500854940052e-07,
+      "loss": 0.6677,
+      "step": 7274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.283706786780815e-07,
+      "loss": 0.669,
+      "step": 7275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2748803801357467e-07,
+      "loss": 0.7241,
+      "step": 7276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2660708670846244e-07,
+      "loss": 0.6463,
+      "step": 7277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.257278249150341e-07,
+      "loss": 0.6694,
+      "step": 7278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2485025278528694e-07,
+      "loss": 0.6361,
+      "step": 7279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2397437047092407e-07,
+      "loss": 0.6033,
+      "step": 7280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231001781233577e-07,
+      "loss": 0.6798,
+      "step": 7281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222276758937103e-07,
+      "loss": 0.702,
+      "step": 7282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2135686393280675e-07,
+      "loss": 0.6777,
+      "step": 7283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2048774239118554e-07,
+      "loss": 0.7012,
+      "step": 7284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196203114190898e-07,
+      "loss": 0.6156,
+      "step": 7285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.187545711664707e-07,
+      "loss": 0.7538,
+      "step": 7286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1789052178298853e-07,
+      "loss": 0.7518,
+      "step": 7287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1702816341800826e-07,
+      "loss": 0.6937,
+      "step": 7288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.161674962206062e-07,
+      "loss": 0.2506,
+      "step": 7289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.153085203395633e-07,
+      "loss": 0.2647,
+      "step": 7290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.144512359233697e-07,
+      "loss": 0.727,
+      "step": 7291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1359564312022352e-07,
+      "loss": 0.6786,
+      "step": 7292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1274174207803088e-07,
+      "loss": 0.6764,
+      "step": 7293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.118895329444015e-07,
+      "loss": 0.6839,
+      "step": 7294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1103901586665753e-07,
+      "loss": 0.6769,
+      "step": 7295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.101901909918258e-07,
+      "loss": 0.6831,
+      "step": 7296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0934305846664117e-07,
+      "loss": 0.6222,
+      "step": 7297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0849761843754645e-07,
+      "loss": 0.6818,
+      "step": 7298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0765387105069256e-07,
+      "loss": 0.7044,
+      "step": 7299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0681181645193616e-07,
+      "loss": 0.6887,
+      "step": 7300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0597145478684188e-07,
+      "loss": 0.6978,
+      "step": 7301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0513278620068022e-07,
+      "loss": 0.7247,
+      "step": 7302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0429581083843299e-07,
+      "loss": 0.7258,
+      "step": 7303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.034605288447855e-07,
+      "loss": 0.6511,
+      "step": 7304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0262694036413232e-07,
+      "loss": 0.6595,
+      "step": 7305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0179504554057483e-07,
+      "loss": 0.7171,
+      "step": 7306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.009648445179202e-07,
+      "loss": 0.7164,
+      "step": 7307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0013633743968585e-07,
+      "loss": 0.7023,
+      "step": 7308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9930952444909057e-07,
+      "loss": 0.7438,
+      "step": 7309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9848440568907002e-07,
+      "loss": 0.6539,
+      "step": 7310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9766098130225785e-07,
+      "loss": 0.6852,
+      "step": 7311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9683925143099913e-07,
+      "loss": 0.7362,
+      "step": 7312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9601921621734465e-07,
+      "loss": 0.6642,
+      "step": 7313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9520087580305214e-07,
+      "loss": 0.6215,
+      "step": 7314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.943842303295873e-07,
+      "loss": 0.6606,
+      "step": 7315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935692799381217e-07,
+      "loss": 0.6589,
+      "step": 7316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9275602476953813e-07,
+      "loss": 0.6598,
+      "step": 7317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9194446496441753e-07,
+      "loss": 0.6796,
+      "step": 7318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9113460066305766e-07,
+      "loss": 0.7109,
+      "step": 7319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9032643200545432e-07,
+      "loss": 0.6869,
+      "step": 7320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8951995913131793e-07,
+      "loss": 0.6949,
+      "step": 7321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8871518218006036e-07,
+      "loss": 0.6787,
+      "step": 7322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.879121012908025e-07,
+      "loss": 0.7639,
+      "step": 7323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711071660237223e-07,
+      "loss": 0.2571,
+      "step": 7324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8631102825330428e-07,
+      "loss": 0.225,
+      "step": 7325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8551303638183692e-07,
+      "loss": 0.6667,
+      "step": 7326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8471674112591985e-07,
+      "loss": 0.7291,
+      "step": 7327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.839221426232074e-07,
+      "loss": 0.6289,
+      "step": 7328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8312924101106077e-07,
+      "loss": 0.6823,
+      "step": 7329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8233803642654702e-07,
+      "loss": 0.6284,
+      "step": 7330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8154852900644115e-07,
+      "loss": 0.7192,
+      "step": 7331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8076071888722402e-07,
+      "loss": 0.7351,
+      "step": 7332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7997460620508112e-07,
+      "loss": 0.6663,
+      "step": 7333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7919019109590929e-07,
+      "loss": 0.6916,
+      "step": 7334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7840747369530787e-07,
+      "loss": 0.2774,
+      "step": 7335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7762645413858527e-07,
+      "loss": 0.6754,
+      "step": 7336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7684713256075238e-07,
+      "loss": 0.6663,
+      "step": 7337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7606950909653365e-07,
+      "loss": 0.732,
+      "step": 7338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7529358388035155e-07,
+      "loss": 0.6352,
+      "step": 7339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7451935704633993e-07,
+      "loss": 0.78,
+      "step": 7340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.737468287283406e-07,
+      "loss": 0.6594,
+      "step": 7341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7297599905989671e-07,
+      "loss": 0.7273,
+      "step": 7342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722068681742628e-07,
+      "loss": 0.6825,
+      "step": 7343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7143943620439585e-07,
+      "loss": 0.2702,
+      "step": 7344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7067370328296085e-07,
+      "loss": 0.6306,
+      "step": 7345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6990966954232858e-07,
+      "loss": 0.6305,
+      "step": 7346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6914733511457671e-07,
+      "loss": 0.6507,
+      "step": 7347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6838670013148872e-07,
+      "loss": 0.6815,
+      "step": 7348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.67627764724555e-07,
+      "loss": 0.6861,
+      "step": 7349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.668705290249717e-07,
+      "loss": 0.6415,
+      "step": 7350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6611499316363965e-07,
+      "loss": 0.7243,
+      "step": 7351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6536115727116887e-07,
+      "loss": 0.6311,
+      "step": 7352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6460902147787173e-07,
+      "loss": 0.6866,
+      "step": 7353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.638585859137709e-07,
+      "loss": 0.653,
+      "step": 7354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6310985070859152e-07,
+      "loss": 0.6764,
+      "step": 7355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6236281599176896e-07,
+      "loss": 0.6446,
+      "step": 7356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6161748189243877e-07,
+      "loss": 0.6457,
+      "step": 7357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6087384853944677e-07,
+      "loss": 0.7712,
+      "step": 7358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6013191606134348e-07,
+      "loss": 0.7196,
+      "step": 7359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5939168458638743e-07,
+      "loss": 0.6705,
+      "step": 7360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.586531542425396e-07,
+      "loss": 0.3034,
+      "step": 7361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5791632515746891e-07,
+      "loss": 0.6979,
+      "step": 7362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5718119745855022e-07,
+      "loss": 0.6534,
+      "step": 7363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5644777127286292e-07,
+      "loss": 0.7217,
+      "step": 7364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5571604672719454e-07,
+      "loss": 0.5876,
+      "step": 7365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.549860239480372e-07,
+      "loss": 0.6884,
+      "step": 7366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5425770306158884e-07,
+      "loss": 0.6969,
+      "step": 7367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5353108419375208e-07,
+      "loss": 0.6748,
+      "step": 7368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5280616747013753e-07,
+      "loss": 0.7214,
+      "step": 7369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5208295301605947e-07,
+      "loss": 0.6996,
+      "step": 7370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5136144095654005e-07,
+      "loss": 0.6565,
+      "step": 7371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5064163141630506e-07,
+      "loss": 0.6811,
+      "step": 7372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.499235245197861e-07,
+      "loss": 0.699,
+      "step": 7373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4920712039112385e-07,
+      "loss": 0.6969,
+      "step": 7374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4849241915416036e-07,
+      "loss": 0.7328,
+      "step": 7375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.477794209324457e-07,
+      "loss": 0.6502,
+      "step": 7376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4706812584923235e-07,
+      "loss": 0.6496,
+      "step": 7377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4635853402748312e-07,
+      "loss": 0.7557,
+      "step": 7378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4565064558986432e-07,
+      "loss": 0.7259,
+      "step": 7379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.449444606587469e-07,
+      "loss": 0.6885,
+      "step": 7380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.442399793562077e-07,
+      "loss": 0.7306,
+      "step": 7381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4353720180403042e-07,
+      "loss": 0.2468,
+      "step": 7382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4283612812370118e-07,
+      "loss": 0.6648,
+      "step": 7383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4213675843641416e-07,
+      "loss": 0.7592,
+      "step": 7384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4143909286306934e-07,
+      "loss": 0.6537,
+      "step": 7385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4074313152427023e-07,
+      "loss": 0.6422,
+      "step": 7386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4004887454032833e-07,
+      "loss": 0.6532,
+      "step": 7387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.393563220312566e-07,
+      "loss": 0.6511,
+      "step": 7388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.38665474116777e-07,
+      "loss": 0.7536,
+      "step": 7389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3797633091631402e-07,
+      "loss": 0.7882,
+      "step": 7390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3728889254900013e-07,
+      "loss": 0.278,
+      "step": 7391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3660315913367029e-07,
+      "loss": 0.687,
+      "step": 7392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3591913078886853e-07,
+      "loss": 0.6663,
+      "step": 7393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.352368076328392e-07,
+      "loss": 0.6978,
+      "step": 7394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.345561897835368e-07,
+      "loss": 0.667,
+      "step": 7395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3387727735861612e-07,
+      "loss": 0.6809,
+      "step": 7396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3320007047544215e-07,
+      "loss": 0.7368,
+      "step": 7397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3252456925108016e-07,
+      "loss": 0.5827,
+      "step": 7398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.318507738023056e-07,
+      "loss": 0.6832,
+      "step": 7399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3117868424559644e-07,
+      "loss": 0.7087,
+      "step": 7400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3050830069713305e-07,
+      "loss": 0.6493,
+      "step": 7401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.298396232728072e-07,
+      "loss": 0.2644,
+      "step": 7402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.291726520882086e-07,
+      "loss": 0.6957,
+      "step": 7403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2850738725863733e-07,
+      "loss": 0.7066,
+      "step": 7404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2784382889909575e-07,
+      "loss": 0.6689,
+      "step": 7405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2718197712429547e-07,
+      "loss": 0.652,
+      "step": 7406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.26521832048645e-07,
+      "loss": 0.6994,
+      "step": 7407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2586339378626634e-07,
+      "loss": 0.6199,
+      "step": 7408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2520666245098068e-07,
+      "loss": 0.688,
+      "step": 7409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.245516381563172e-07,
+      "loss": 0.5886,
+      "step": 7410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2389832101550869e-07,
+      "loss": 0.7368,
+      "step": 7411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2324671114149257e-07,
+      "loss": 0.7325,
+      "step": 7412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2259680864691314e-07,
+      "loss": 0.7069,
+      "step": 7413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2194861364411615e-07,
+      "loss": 0.6182,
+      "step": 7414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2130212624515525e-07,
+      "loss": 0.6775,
+      "step": 7415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2065734656178772e-07,
+      "loss": 0.6975,
+      "step": 7416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2001427470547555e-07,
+      "loss": 0.5793,
+      "step": 7417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.193729107873842e-07,
+      "loss": 0.604,
+      "step": 7418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1873325491838838e-07,
+      "loss": 0.7077,
+      "step": 7419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.180953072090607e-07,
+      "loss": 0.7199,
+      "step": 7420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1745906776968519e-07,
+      "loss": 0.6759,
+      "step": 7421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1682453671024607e-07,
+      "loss": 0.6738,
+      "step": 7422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1619171414043451e-07,
+      "loss": 0.2609,
+      "step": 7423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.155606001696452e-07,
+      "loss": 0.7322,
+      "step": 7424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1493119490697758e-07,
+      "loss": 0.6428,
+      "step": 7425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1430349846123789e-07,
+      "loss": 0.6403,
+      "step": 7426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.136775109409316e-07,
+      "loss": 0.6595,
+      "step": 7427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1305323245427546e-07,
+      "loss": 0.5994,
+      "step": 7428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1243066310918538e-07,
+      "loss": 0.6972,
+      "step": 7429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1180980301328526e-07,
+      "loss": 0.6565,
+      "step": 7430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1119065227390147e-07,
+      "loss": 0.6447,
+      "step": 7431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1057321099806729e-07,
+      "loss": 0.7379,
+      "step": 7432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0995747929251621e-07,
+      "loss": 0.6835,
+      "step": 7433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0934345726369089e-07,
+      "loss": 0.7151,
+      "step": 7434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.087311450177353e-07,
+      "loss": 0.7898,
+      "step": 7435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0812054266050032e-07,
+      "loss": 0.634,
+      "step": 7436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0751165029753818e-07,
+      "loss": 0.6653,
+      "step": 7437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0690446803410804e-07,
+      "loss": 0.6898,
+      "step": 7438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0629899597517257e-07,
+      "loss": 0.7377,
+      "step": 7439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0569523422539918e-07,
+      "loss": 0.6716,
+      "step": 7440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0509318288915771e-07,
+      "loss": 0.6581,
+      "step": 7441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0449284207052602e-07,
+      "loss": 0.6857,
+      "step": 7442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.038942118732822e-07,
+      "loss": 0.6656,
+      "step": 7443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0329729240091124e-07,
+      "loss": 0.6031,
+      "step": 7444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0270208375660285e-07,
+      "loss": 0.6623,
+      "step": 7445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0210858604324803e-07,
+      "loss": 0.6867,
+      "step": 7446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0151679936344472e-07,
+      "loss": 0.684,
+      "step": 7447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0092672381949443e-07,
+      "loss": 0.2403,
+      "step": 7448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0033835951340221e-07,
+      "loss": 0.6696,
+      "step": 7449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.975170654687893e-08,
+      "loss": 0.6623,
+      "step": 7450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.916676502133571e-08,
+      "loss": 0.732,
+      "step": 7451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.858353503789276e-08,
+      "loss": 0.8116,
+      "step": 7452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.800201669737163e-08,
+      "loss": 0.6495,
+      "step": 7453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.742221010029751e-08,
+      "loss": 0.6567,
+      "step": 7454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.684411534690241e-08,
+      "loss": 0.6346,
+      "step": 7455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.626773253711974e-08,
+      "loss": 0.6554,
+      "step": 7456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.569306177058757e-08,
+      "loss": 0.6708,
+      "step": 7457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.512010314664976e-08,
+      "loss": 0.6452,
+      "step": 7458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.454885676435155e-08,
+      "loss": 0.591,
+      "step": 7459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.39793227224428e-08,
+      "loss": 0.6845,
+      "step": 7460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.341150111938035e-08,
+      "loss": 0.6905,
+      "step": 7461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.28453920533201e-08,
+      "loss": 0.6397,
+      "step": 7462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.228099562212712e-08,
+      "loss": 0.6669,
+      "step": 7463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.171831192336555e-08,
+      "loss": 0.6619,
+      "step": 7464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.115734105430652e-08,
+      "loss": 0.7007,
+      "step": 7465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.059808311192353e-08,
+      "loss": 0.6945,
+      "step": 7466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.004053819289482e-08,
+      "loss": 0.6737,
+      "step": 7467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.948470639360441e-08,
+      "loss": 0.6636,
+      "step": 7468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.893058781013541e-08,
+      "loss": 0.6691,
+      "step": 7469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.8378182538279e-08,
+      "loss": 0.6974,
+      "step": 7470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.78274906735277e-08,
+      "loss": 0.762,
+      "step": 7471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.727851231107976e-08,
+      "loss": 0.6944,
+      "step": 7472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.673124754583595e-08,
+      "loss": 0.7813,
+      "step": 7473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.618569647240061e-08,
+      "loss": 0.6905,
+      "step": 7474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.564185918508161e-08,
+      "loss": 0.6568,
+      "step": 7475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.509973577789377e-08,
+      "loss": 0.6493,
+      "step": 7476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.4559326344551e-08,
+      "loss": 0.6728,
+      "step": 7477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.402063097847413e-08,
+      "loss": 0.6933,
+      "step": 7478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.348364977278534e-08,
+      "loss": 0.6957,
+      "step": 7479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.294838282031258e-08,
+      "loss": 0.621,
+      "step": 7480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.241483021358742e-08,
+      "loss": 0.6338,
+      "step": 7481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.188299204484385e-08,
+      "loss": 0.7517,
+      "step": 7482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.135286840601942e-08,
+      "loss": 0.6841,
+      "step": 7483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.082445938875527e-08,
+      "loss": 0.6226,
+      "step": 7484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.029776508439835e-08,
+      "loss": 0.6755,
+      "step": 7485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.977278558399581e-08,
+      "loss": 0.6396,
+      "step": 7486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.924952097830174e-08,
+      "loss": 0.6578,
+      "step": 7487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.872797135777044e-08,
+      "loss": 0.7272,
+      "step": 7488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.820813681256201e-08,
+      "loss": 0.6652,
+      "step": 7489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.769001743254013e-08,
+      "loss": 0.6625,
+      "step": 7490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.717361330727092e-08,
+      "loss": 0.6617,
+      "step": 7491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.665892452602297e-08,
+      "loss": 0.6559,
+      "step": 7492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.614595117777179e-08,
+      "loss": 0.6999,
+      "step": 7493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.563469335119422e-08,
+      "loss": 0.6974,
+      "step": 7494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.512515113466955e-08,
+      "loss": 0.6678,
+      "step": 7495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.46173246162829e-08,
+      "loss": 0.6152,
+      "step": 7496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.411121388381959e-08,
+      "loss": 0.6341,
+      "step": 7497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.360681902477073e-08,
+      "loss": 0.6698,
+      "step": 7498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.310414012633104e-08,
+      "loss": 0.692,
+      "step": 7499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.260317727539879e-08,
+      "loss": 0.6839,
+      "step": 7500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.210393055857246e-08,
+      "loss": 0.7069,
+      "step": 7501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.160640006215747e-08,
+      "loss": 0.6208,
+      "step": 7502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111058587216057e-08,
+      "loss": 0.659,
+      "step": 7503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.061648807429211e-08,
+      "loss": 0.6298,
+      "step": 7504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.012410675396708e-08,
+      "loss": 0.7311,
+      "step": 7505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.963344199630184e-08,
+      "loss": 0.7113,
+      "step": 7506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.914449388611854e-08,
+      "loss": 0.699,
+      "step": 7507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.865726250793736e-08,
+      "loss": 0.6983,
+      "step": 7508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.81717479459898e-08,
+      "loss": 0.6533,
+      "step": 7509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.768795028420095e-08,
+      "loss": 0.6933,
+      "step": 7510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.720586960620945e-08,
+      "loss": 0.6483,
+      "step": 7511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.672550599534755e-08,
+      "loss": 0.6503,
+      "step": 7512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.624685953465771e-08,
+      "loss": 0.6521,
+      "step": 7513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.57699303068815e-08,
+      "loss": 0.6623,
+      "step": 7514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.529471839446522e-08,
+      "loss": 0.6273,
+      "step": 7515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.482122387955869e-08,
+      "loss": 0.2477,
+      "step": 7516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.434944684401312e-08,
+      "loss": 0.6072,
+      "step": 7517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.387938736938548e-08,
+      "loss": 0.7225,
+      "step": 7518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.34110455369319e-08,
+      "loss": 0.6577,
+      "step": 7519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.294442142761536e-08,
+      "loss": 0.7546,
+      "step": 7520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.247951512210026e-08,
+      "loss": 0.7483,
+      "step": 7521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.201632670075452e-08,
+      "loss": 0.6764,
+      "step": 7522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.155485624364854e-08,
+      "loss": 0.7107,
+      "step": 7523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.109510383055629e-08,
+      "loss": 0.722,
+      "step": 7524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.063706954095417e-08,
+      "loss": 0.7608,
+      "step": 7525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.018075345402219e-08,
+      "loss": 0.7765,
+      "step": 7526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.972615564864282e-08,
+      "loss": 0.7147,
+      "step": 7527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.927327620340095e-08,
+      "loss": 0.6372,
+      "step": 7528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8822115196586205e-08,
+      "loss": 0.6502,
+      "step": 7529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.837267270618952e-08,
+      "loss": 0.7136,
+      "step": 7530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7924948809906515e-08,
+      "loss": 0.7425,
+      "step": 7531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.747894358513306e-08,
+      "loss": 0.2221,
+      "step": 7532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.703465710897082e-08,
+      "loss": 0.6839,
+      "step": 7533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6592089458221697e-08,
+      "loss": 0.6591,
+      "step": 7534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6151240709391154e-08,
+      "loss": 0.6841,
+      "step": 7535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5712110938689335e-08,
+      "loss": 0.6398,
+      "step": 7536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5274700222027744e-08,
+      "loss": 0.7853,
+      "step": 7537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.483900863502145e-08,
+      "loss": 0.7241,
+      "step": 7538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.440503625298688e-08,
+      "loss": 0.6716,
+      "step": 7539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.397278315094401e-08,
+      "loss": 0.6602,
+      "step": 7540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.354224940361641e-08,
+      "loss": 0.7092,
+      "step": 7541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.311343508542899e-08,
+      "loss": 0.6885,
+      "step": 7542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2686340270511336e-08,
+      "loss": 0.6798,
+      "step": 7543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.22609650326944e-08,
+      "loss": 0.6934,
+      "step": 7544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1837309445511575e-08,
+      "loss": 0.6817,
+      "step": 7545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1415373582199835e-08,
+      "loss": 0.6716,
+      "step": 7546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0995157515698614e-08,
+      "loss": 0.7019,
+      "step": 7547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.057666131865091e-08,
+      "loss": 0.6906,
+      "step": 7548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.015988506339997e-08,
+      "loss": 0.7012,
+      "step": 7549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.974482882199372e-08,
+      "loss": 0.7636,
+      "step": 7550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.933149266618364e-08,
+      "loss": 0.731,
+      "step": 7551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8919876667420376e-08,
+      "loss": 0.7703,
+      "step": 7552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.850998089686143e-08,
+      "loss": 0.7487,
+      "step": 7553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8101805425363466e-08,
+      "loss": 0.6767,
+      "step": 7554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.769535032348782e-08,
+      "loss": 0.6323,
+      "step": 7555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.729061566149717e-08,
+      "loss": 0.6549,
+      "step": 7556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.688760150935889e-08,
+      "loss": 0.749,
+      "step": 7557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.648630793673836e-08,
+      "loss": 0.7078,
+      "step": 7558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.608673501301009e-08,
+      "loss": 0.7468,
+      "step": 7559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.568888280724548e-08,
+      "loss": 0.7228,
+      "step": 7560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.529275138822175e-08,
+      "loss": 0.6565,
+      "step": 7561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.489834082441635e-08,
+      "loss": 0.7095,
+      "step": 7562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.45056511840114e-08,
+      "loss": 0.6743,
+      "step": 7563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4114682534891485e-08,
+      "loss": 0.7054,
+      "step": 7564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.372543494464032e-08,
+      "loss": 0.7252,
+      "step": 7565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.333790848054853e-08,
+      "loss": 0.6654,
+      "step": 7566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2952103209604744e-08,
+      "loss": 0.6162,
+      "step": 7567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.256801919850673e-08,
+      "loss": 0.7236,
+      "step": 7568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.218565651364692e-08,
+      "loss": 0.6793,
+      "step": 7569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.180501522112468e-08,
+      "loss": 0.5985,
+      "step": 7570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.142609538674181e-08,
+      "loss": 0.6613,
+      "step": 7571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.104889707600035e-08,
+      "loss": 0.2375,
+      "step": 7572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.067342035410704e-08,
+      "loss": 0.6123,
+      "step": 7573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0299665285969956e-08,
+      "loss": 0.7047,
+      "step": 7574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.992763193619853e-08,
+      "loss": 0.7319,
+      "step": 7575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9557320369106866e-08,
+      "loss": 0.6843,
+      "step": 7576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9188730648709316e-08,
+      "loss": 0.6957,
+      "step": 7577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.882186283872269e-08,
+      "loss": 0.7118,
+      "step": 7578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.845671700256848e-08,
+      "loss": 0.7414,
+      "step": 7579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.809329320336841e-08,
+      "loss": 0.5961,
+      "step": 7580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7731591503947786e-08,
+      "loss": 0.672,
+      "step": 7581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.737161196683103e-08,
+      "loss": 0.6721,
+      "step": 7582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.701335465424949e-08,
+      "loss": 0.6946,
+      "step": 7583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.665681962813361e-08,
+      "loss": 0.6576,
+      "step": 7584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.630200695011743e-08,
+      "loss": 0.6795,
+      "step": 7585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.594891668153633e-08,
+      "loss": 0.7244,
+      "step": 7586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.559754888342926e-08,
+      "loss": 0.6114,
+      "step": 7587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.524790361653541e-08,
+      "loss": 0.7233,
+      "step": 7588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.489998094129976e-08,
+      "loss": 0.6667,
+      "step": 7589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.455378091786532e-08,
+      "loss": 0.6654,
+      "step": 7590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4209303606078655e-08,
+      "loss": 0.7295,
+      "step": 7591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386654906548992e-08,
+      "loss": 0.7116,
+      "step": 7592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.352551735535059e-08,
+      "loss": 0.6868,
+      "step": 7593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.318620853461463e-08,
+      "loss": 0.7073,
+      "step": 7594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2848622661938445e-08,
+      "loss": 0.7388,
+      "step": 7595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2512759795677585e-08,
+      "loss": 0.7161,
+      "step": 7596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.21786199938956e-08,
+      "loss": 0.2488,
+      "step": 7597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.184620331435073e-08,
+      "loss": 0.6246,
+      "step": 7598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.151550981451035e-08,
+      "loss": 0.7881,
+      "step": 7599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.118653955153983e-08,
+      "loss": 0.6806,
+      "step": 7600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.085929258230702e-08,
+      "loss": 0.6702,
+      "step": 7601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.053376896338445e-08,
+      "loss": 0.6799,
+      "step": 7602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.020996875104376e-08,
+      "loss": 0.6866,
+      "step": 7603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.988789200126019e-08,
+      "loss": 0.7609,
+      "step": 7604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9567538769710303e-08,
+      "loss": 0.7963,
+      "step": 7605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9248909111773138e-08,
+      "loss": 0.6696,
+      "step": 7606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8932003082529082e-08,
+      "loss": 0.7507,
+      "step": 7607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8616820736763197e-08,
+      "loss": 0.727,
+      "step": 7608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8303362128958568e-08,
+      "loss": 0.689,
+      "step": 7609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.799162731330407e-08,
+      "loss": 0.2272,
+      "step": 7610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7681616343686602e-08,
+      "loss": 0.7232,
+      "step": 7611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7373329273698847e-08,
+      "loss": 0.6948,
+      "step": 7612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7066766156633728e-08,
+      "loss": 0.6858,
+      "step": 7613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6761927045486637e-08,
+      "loss": 0.7175,
+      "step": 7614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6458811992953194e-08,
+      "loss": 0.7479,
+      "step": 7615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6157421051434818e-08,
+      "loss": 0.7162,
+      "step": 7616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5857754273029834e-08,
+      "loss": 0.6799,
+      "step": 7617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.555981170954347e-08,
+      "loss": 0.7255,
+      "step": 7618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.526359341247897e-08,
+      "loss": 0.6771,
+      "step": 7619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4969099433044263e-08,
+      "loss": 0.6424,
+      "step": 7620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4676329822147514e-08,
+      "loss": 0.7322,
+      "step": 7621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4385284630399352e-08,
+      "loss": 0.2581,
+      "step": 7622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.409596390811286e-08,
+      "loss": 0.7258,
+      "step": 7623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.380836770530137e-08,
+      "loss": 0.6462,
+      "step": 7624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3522496071682886e-08,
+      "loss": 0.7314,
+      "step": 7625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3238349056674546e-08,
+      "loss": 0.6478,
+      "step": 7626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2955926709395947e-08,
+      "loss": 0.6437,
+      "step": 7627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2675229078669148e-08,
+      "loss": 0.6361,
+      "step": 7628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2396256213018665e-08,
+      "loss": 0.7156,
+      "step": 7629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2119008160670365e-08,
+      "loss": 0.6462,
+      "step": 7630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1843484969550355e-08,
+      "loss": 0.7294,
+      "step": 7631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1569686687289425e-08,
+      "loss": 0.6593,
+      "step": 7632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1297613361217494e-08,
+      "loss": 0.7178,
+      "step": 7633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.102726503836805e-08,
+      "loss": 0.6896,
+      "step": 7634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.075864176547593e-08,
+      "loss": 0.6823,
+      "step": 7635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.049174358897732e-08,
+      "loss": 0.6975,
+      "step": 7636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022657055500976e-08,
+      "loss": 0.6949,
+      "step": 7637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9963122709414363e-08,
+      "loss": 0.7344,
+      "step": 7638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.970140009773358e-08,
+      "loss": 0.713,
+      "step": 7639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9441402765209006e-08,
+      "loss": 0.6665,
+      "step": 7640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9183130756788014e-08,
+      "loss": 0.6384,
+      "step": 7641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.892658411711601e-08,
+      "loss": 0.7375,
+      "step": 7642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8671762890543065e-08,
+      "loss": 0.7138,
+      "step": 7643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8418667121118393e-08,
+      "loss": 0.6357,
+      "step": 7644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8167296852595884e-08,
+      "loss": 0.6689,
+      "step": 7645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7917652128428552e-08,
+      "loss": 0.6385,
+      "step": 7646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7669732991772994e-08,
+      "loss": 0.5939,
+      "step": 7647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.742353948548603e-08,
+      "loss": 0.6715,
+      "step": 7648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7179071652125846e-08,
+      "loss": 0.6827,
+      "step": 7649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6936329533954187e-08,
+      "loss": 0.6927,
+      "step": 7650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6695313172934156e-08,
+      "loss": 0.6483,
+      "step": 7651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6456022610729094e-08,
+      "loss": 0.2391,
+      "step": 7652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6218457888703686e-08,
+      "loss": 0.66,
+      "step": 7653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5982619047928416e-08,
+      "loss": 0.6825,
+      "step": 7654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574850612916845e-08,
+      "loss": 0.7557,
+      "step": 7655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5516119172898082e-08,
+      "loss": 0.6849,
+      "step": 7656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5285458219287398e-08,
+      "loss": 0.6577,
+      "step": 7657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5056523308212278e-08,
+      "loss": 0.7347,
+      "step": 7658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4829314479246627e-08,
+      "loss": 0.6406,
+      "step": 7659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.460383177166902e-08,
+      "loss": 0.7231,
+      "step": 7660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4380075224456058e-08,
+      "loss": 0.7068,
+      "step": 7661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4158044876291243e-08,
+      "loss": 0.735,
+      "step": 7662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3937740765554986e-08,
+      "loss": 0.6513,
+      "step": 7663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3719162930331265e-08,
+      "loss": 0.7213,
+      "step": 7664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3502311408404301e-08,
+      "loss": 0.6709,
+      "step": 7665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3287186237262994e-08,
+      "loss": 0.6682,
+      "step": 7666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3073787454093157e-08,
+      "loss": 0.7422,
+      "step": 7667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2862115095786387e-08,
+      "loss": 0.6869,
+      "step": 7668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2652169198934528e-08,
+      "loss": 0.7752,
+      "step": 7669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2443949799829658e-08,
+      "loss": 0.694,
+      "step": 7670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2237456934466319e-08,
+      "loss": 0.7018,
+      "step": 7671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2032690638541511e-08,
+      "loss": 0.709,
+      "step": 7672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1829650947451365e-08,
+      "loss": 0.672,
+      "step": 7673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1628337896296693e-08,
+      "loss": 0.7178,
+      "step": 7674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1428751519876324e-08,
+      "loss": 0.5518,
+      "step": 7675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1230891852694881e-08,
+      "loss": 0.6911,
+      "step": 7676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1034758928953892e-08,
+      "loss": 0.6965,
+      "step": 7677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0840352782559572e-08,
+      "loss": 0.6642,
+      "step": 7678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0647673447118367e-08,
+      "loss": 0.6767,
+      "step": 7679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0456720955939192e-08,
+      "loss": 0.7534,
+      "step": 7680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0267495342030087e-08,
+      "loss": 0.6491,
+      "step": 7681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0079996638103773e-08,
+      "loss": 0.6823,
+      "step": 7682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.8942248765721e-09,
+      "loss": 0.6986,
+      "step": 7683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.710180089549382e-09,
+      "loss": 0.6865,
+      "step": 7684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.527862308851276e-09,
+      "loss": 0.6729,
+      "step": 7685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.347271565993687e-09,
+      "loss": 0.6868,
+      "step": 7686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.168407892197196e-09,
+      "loss": 0.7207,
+      "step": 7687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.991271318380401e-09,
+      "loss": 0.7117,
+      "step": 7688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.815861875164367e-09,
+      "loss": 0.7083,
+      "step": 7689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.642179592871503e-09,
+      "loss": 0.7029,
+      "step": 7690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.47022450152779e-09,
+      "loss": 0.606,
+      "step": 7691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.299996630857232e-09,
+      "loss": 0.7059,
+      "step": 7692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.131496010287398e-09,
+      "loss": 0.2496,
+      "step": 7693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.964722668947211e-09,
+      "loss": 0.6173,
+      "step": 7694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.799676635665831e-09,
+      "loss": 0.6871,
+      "step": 7695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.63635793897488e-09,
+      "loss": 0.7203,
+      "step": 7696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.47476660710622e-09,
+      "loss": 0.6796,
+      "step": 7697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.314902667995283e-09,
+      "loss": 0.7158,
+      "step": 7698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.156766149276628e-09,
+      "loss": 0.6762,
+      "step": 7699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0003570782872785e-09,
+      "loss": 0.7271,
+      "step": 7700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8456754820644956e-09,
+      "loss": 0.2474,
+      "step": 7701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.69272138735022e-09,
+      "loss": 0.7192,
+      "step": 7702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.541494820583305e-09,
+      "loss": 0.736,
+      "step": 7703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3919958079061705e-09,
+      "loss": 0.6533,
+      "step": 7704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2442243751636985e-09,
+      "loss": 0.6891,
+      "step": 7705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.098180547899901e-09,
+      "loss": 0.6561,
+      "step": 7706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9538643513612495e-09,
+      "loss": 0.5804,
+      "step": 7707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.811275810495565e-09,
+      "loss": 0.6805,
+      "step": 7708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.670414949953129e-09,
+      "loss": 0.6818,
+      "step": 7709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.531281794082244e-09,
+      "loss": 0.6535,
+      "step": 7710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.393876366935891e-09,
+      "loss": 0.7743,
+      "step": 7711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.258198692266181e-09,
+      "loss": 0.6371,
+      "step": 7712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1242487935287965e-09,
+      "loss": 0.6702,
+      "step": 7713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.992026693879659e-09,
+      "loss": 0.7162,
+      "step": 7714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.86153241617382e-09,
+      "loss": 0.6586,
+      "step": 7715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7327659829710105e-09,
+      "loss": 0.6426,
+      "step": 7716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.605727416531203e-09,
+      "loss": 0.6797,
+      "step": 7717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.480416738814608e-09,
+      "loss": 0.6222,
+      "step": 7718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.356833971483898e-09,
+      "loss": 0.6559,
+      "step": 7719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.234979135903095e-09,
+      "loss": 0.6051,
+      "step": 7720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.114852253135349e-09,
+      "loss": 0.6712,
+      "step": 7721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.996453343948492e-09,
+      "loss": 0.7354,
+      "step": 7722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.879782428809486e-09,
+      "loss": 0.236,
+      "step": 7723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7648395278866435e-09,
+      "loss": 0.7113,
+      "step": 7724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6516246610507344e-09,
+      "loss": 0.7118,
+      "step": 7725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.540137847871661e-09,
+      "loss": 0.7668,
+      "step": 7726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4303791076240046e-09,
+      "loss": 0.7239,
+      "step": 7727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.322348459280367e-09,
+      "loss": 0.6822,
+      "step": 7728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2160459215146987e-09,
+      "loss": 0.6039,
+      "step": 7729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1114715127056326e-09,
+      "loss": 0.633,
+      "step": 7730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0086252509298196e-09,
+      "loss": 0.7216,
+      "step": 7731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.907507153966371e-09,
+      "loss": 0.7306,
+      "step": 7732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.808117239294639e-09,
+      "loss": 0.6696,
+      "step": 7733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.710455524097544e-09,
+      "loss": 0.6155,
+      "step": 7734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6145220252560277e-09,
+      "loss": 0.6175,
+      "step": 7735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5203167593546017e-09,
+      "loss": 0.7205,
+      "step": 7736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4278397426780178e-09,
+      "loss": 0.6641,
+      "step": 7737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3370909912134867e-09,
+      "loss": 0.6305,
+      "step": 7738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.24807052064846e-09,
+      "loss": 0.7093,
+      "step": 7739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1607783463706288e-09,
+      "loss": 0.6895,
+      "step": 7740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0752144834723653e-09,
+      "loss": 0.6726,
+      "step": 7741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9913789467418397e-09,
+      "loss": 0.7044,
+      "step": 7742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.909271750674124e-09,
+      "loss": 0.7663,
+      "step": 7743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8288929094623099e-09,
+      "loss": 0.6881,
+      "step": 7744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7502424370008375e-09,
+      "loss": 0.7189,
+      "step": 7745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6733203468866089e-09,
+      "loss": 0.7347,
+      "step": 7746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5981266524167648e-09,
+      "loss": 0.6859,
+      "step": 7747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5246613665886868e-09,
+      "loss": 0.6069,
+      "step": 7748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4529245021044357e-09,
+      "loss": 0.6945,
+      "step": 7749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.382916071362983e-09,
+      "loss": 0.6549,
+      "step": 7750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3146360864690899e-09,
+      "loss": 0.6833,
+      "step": 7751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.248084559223317e-09,
+      "loss": 0.2323,
+      "step": 7752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1832615011320159e-09,
+      "loss": 0.7116,
+      "step": 7753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1201669234006673e-09,
+      "loss": 0.286,
+      "step": 7754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0588008369361025e-09,
+      "loss": 0.6351,
+      "step": 7755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.991632523476125e-10,
+      "loss": 0.7099,
+      "step": 7756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.412541799436182e-10,
+      "loss": 0.6993,
+      "step": 7757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.850736297350005e-10,
+      "loss": 0.721,
+      "step": 7758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.306216114328802e-10,
+      "loss": 0.7167,
+      "step": 7759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.778981344519487e-10,
+      "loss": 0.2453,
+      "step": 7760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.269032079049165e-10,
+      "loss": 0.6733,
+      "step": 7761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.776368406091749e-10,
+      "loss": 0.6851,
+      "step": 7762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.300990410790242e-10,
+      "loss": 0.6597,
+      "step": 7763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.842898175334455e-10,
+      "loss": 0.64,
+      "step": 7764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.402091778905493e-10,
+      "loss": 0.6513,
+      "step": 7765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.978571297720169e-10,
+      "loss": 0.6394,
+      "step": 7766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5723368049754856e-10,
+      "loss": 0.2571,
+      "step": 7767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1833883709041513e-10,
+      "loss": 0.7222,
+      "step": 7768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8117260627523743e-10,
+      "loss": 0.6232,
+      "step": 7769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4573499447465576e-10,
+      "loss": 0.2499,
+      "step": 7770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1202600781710114e-10,
+      "loss": 0.7024,
+      "step": 7771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.80045652129024e-10,
+      "loss": 0.2161,
+      "step": 7772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.497939329371146e-10,
+      "loss": 0.62,
+      "step": 7773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2127085547274384e-10,
+      "loss": 0.7018,
+      "step": 7774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9447642466641215e-10,
+      "loss": 0.6624,
+      "step": 7775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6941064514996996e-10,
+      "loss": 0.7184,
+      "step": 7776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4607352125661777e-10,
+      "loss": 0.7326,
+      "step": 7777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2446505702090605e-10,
+      "loss": 0.6922,
+      "step": 7778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0458525617651482e-10,
+      "loss": 0.7156,
+      "step": 7779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.643412216180481e-11,
+      "loss": 0.6707,
+      "step": 7780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.001165811426625e-11,
+      "loss": 0.7096,
+      "step": 7781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.531786687162921e-11,
+      "loss": 0.7195,
+      "step": 7782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2352750975194203e-11,
+      "loss": 0.6696,
+      "step": 7783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1116312666501504e-11,
+      "loss": 0.6075,
+      "step": 7784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1608553887331185e-11,
+      "loss": 0.7367,
+      "step": 7785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3829476280813325e-11,
+      "loss": 0.6518,
+      "step": 7786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.779081192538229e-12,
+      "loss": 0.7341,
+      "step": 7787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.457369668335986e-12,
+      "loss": 0.6801,
+      "step": 7788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.643424542764678e-13,
+      "loss": 0.4821,
+      "step": 7789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3132,
+      "step": 7790
+    },
+    {
+      "epoch": 1.0,
+      "step": 7790,
+      "total_flos": 2951136165888000.0,
+      "train_loss": 0.7439419657068473,
+      "train_runtime": 130828.2071,
+      "train_samples_per_second": 7.621,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7790,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 2951136165888000.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}