diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31482 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998331068354672,
+  "eval_steps": 500,
+  "global_step": 5242,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2658227848101266e-07,
+      "loss": 2.085,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5316455696202533e-07,
+      "loss": 2.251,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.79746835443038e-07,
+      "loss": 2.2003,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.063291139240507e-07,
+      "loss": 2.1613,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.329113924050634e-07,
+      "loss": 2.1948,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.59493670886076e-07,
+      "loss": 2.6278,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.860759493670887e-07,
+      "loss": 2.2479,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0126582278481013e-06,
+      "loss": 2.3188,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.139240506329114e-06,
+      "loss": 2.1006,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2658227848101267e-06,
+      "loss": 2.141,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3924050632911392e-06,
+      "loss": 2.0675,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.518987341772152e-06,
+      "loss": 2.1356,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6455696202531647e-06,
+      "loss": 2.1316,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7721518987341774e-06,
+      "loss": 1.8515,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 1.7799,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0253164556962026e-06,
+      "loss": 2.0186,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1518987341772153e-06,
+      "loss": 1.7791,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.278481012658228e-06,
+      "loss": 1.5859,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4050632911392408e-06,
+      "loss": 1.4581,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5316455696202535e-06,
+      "loss": 1.5511,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6582278481012658e-06,
+      "loss": 1.4654,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7848101265822785e-06,
+      "loss": 1.4344,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9113924050632912e-06,
+      "loss": 1.4582,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.037974683544304e-06,
+      "loss": 1.3856,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.164556962025317e-06,
+      "loss": 1.3708,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2911392405063294e-06,
+      "loss": 1.4288,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.417721518987342e-06,
+      "loss": 1.3541,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.544303797468355e-06,
+      "loss": 1.2619,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6708860759493675e-06,
+      "loss": 1.2327,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 1.244,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.924050632911393e-06,
+      "loss": 1.3264,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.050632911392405e-06,
+      "loss": 1.362,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.177215189873418e-06,
+      "loss": 1.2698,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.303797468354431e-06,
+      "loss": 1.1997,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 1.1728,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.556962025316456e-06,
+      "loss": 1.3371,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683544303797468e-06,
+      "loss": 1.2468,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8101265822784815e-06,
+      "loss": 1.342,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.936708860759495e-06,
+      "loss": 1.1622,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.063291139240507e-06,
+      "loss": 1.029,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.189873417721519e-06,
+      "loss": 1.191,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3164556962025316e-06,
+      "loss": 1.041,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.443037974683545e-06,
+      "loss": 1.2018,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.569620253164557e-06,
+      "loss": 1.0709,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 1.2299,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8227848101265824e-06,
+      "loss": 1.0832,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.949367088607595e-06,
+      "loss": 1.0159,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.075949367088608e-06,
+      "loss": 1.0539,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.20253164556962e-06,
+      "loss": 1.0721,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.329113924050634e-06,
+      "loss": 1.1867,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4556962025316464e-06,
+      "loss": 1.087,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.582278481012659e-06,
+      "loss": 1.1018,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.708860759493672e-06,
+      "loss": 1.0682,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.835443037974684e-06,
+      "loss": 1.152,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.962025316455697e-06,
+      "loss": 1.0493,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.08860759493671e-06,
+      "loss": 1.0922,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.215189873417722e-06,
+      "loss": 1.0245,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.341772151898735e-06,
+      "loss": 0.953,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.468354430379747e-06,
+      "loss": 1.0033,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 1.0078,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.721518987341773e-06,
+      "loss": 0.9438,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.848101265822786e-06,
+      "loss": 1.002,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.974683544303799e-06,
+      "loss": 0.945,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.10126582278481e-06,
+      "loss": 0.8913,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.227848101265824e-06,
+      "loss": 0.9923,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.354430379746837e-06,
+      "loss": 0.9831,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.481012658227848e-06,
+      "loss": 0.986,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.607594936708861e-06,
+      "loss": 0.9615,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.734177215189874e-06,
+      "loss": 0.9902,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.9995,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.987341772151899e-06,
+      "loss": 0.8453,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.113924050632912e-06,
+      "loss": 0.9632,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.240506329113925e-06,
+      "loss": 1.0637,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.367088607594937e-06,
+      "loss": 0.9949,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 1.1207,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.620253164556963e-06,
+      "loss": 1.0071,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.746835443037975e-06,
+      "loss": 0.9051,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.87341772151899e-06,
+      "loss": 0.8692,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.8676,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0126582278481014e-05,
+      "loss": 0.998,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0253164556962025e-05,
+      "loss": 0.9027,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0379746835443039e-05,
+      "loss": 0.9426,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0506329113924052e-05,
+      "loss": 0.9816,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0632911392405063e-05,
+      "loss": 0.9852,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.956,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.088607594936709e-05,
+      "loss": 0.9748,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1012658227848103e-05,
+      "loss": 0.9313,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1139240506329114e-05,
+      "loss": 0.9891,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1265822784810127e-05,
+      "loss": 0.9462,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.8584,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1518987341772152e-05,
+      "loss": 1.0351,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1645569620253165e-05,
+      "loss": 0.8837,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1772151898734178e-05,
+      "loss": 1.0448,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.189873417721519e-05,
+      "loss": 0.8235,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.9322,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2151898734177216e-05,
+      "loss": 1.0184,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.227848101265823e-05,
+      "loss": 0.9077,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.240506329113924e-05,
+      "loss": 0.9255,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2531645569620255e-05,
+      "loss": 0.8475,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2658227848101268e-05,
+      "loss": 0.9057,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2784810126582278e-05,
+      "loss": 0.9491,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2911392405063293e-05,
+      "loss": 1.0422,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3037974683544306e-05,
+      "loss": 0.8978,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3164556962025317e-05,
+      "loss": 0.8195,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.329113924050633e-05,
+      "loss": 0.9889,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3417721518987344e-05,
+      "loss": 0.9977,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3544303797468355e-05,
+      "loss": 1.0563,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3670886075949368e-05,
+      "loss": 0.8708,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3797468354430381e-05,
+      "loss": 0.9525,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3924050632911395e-05,
+      "loss": 0.8196,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4050632911392406e-05,
+      "loss": 0.9388,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.417721518987342e-05,
+      "loss": 0.9412,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4303797468354432e-05,
+      "loss": 0.8369,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4430379746835444e-05,
+      "loss": 0.9378,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 1.0179,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.468354430379747e-05,
+      "loss": 0.9305,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4810126582278482e-05,
+      "loss": 0.8531,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4936708860759495e-05,
+      "loss": 0.8393,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5063291139240508e-05,
+      "loss": 0.9769,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.9349,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.531645569620253e-05,
+      "loss": 0.951,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5443037974683546e-05,
+      "loss": 0.8923,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.556962025316456e-05,
+      "loss": 0.8854,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5696202531645572e-05,
+      "loss": 0.8073,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.943,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5949367088607598e-05,
+      "loss": 0.8304,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.607594936708861e-05,
+      "loss": 0.9482,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.620253164556962e-05,
+      "loss": 0.8649,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6329113924050636e-05,
+      "loss": 1.0226,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6455696202531647e-05,
+      "loss": 0.9419,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.658227848101266e-05,
+      "loss": 0.9479,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6708860759493674e-05,
+      "loss": 0.8854,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6835443037974685e-05,
+      "loss": 0.9087,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6962025316455696e-05,
+      "loss": 0.96,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.708860759493671e-05,
+      "loss": 0.8741,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7215189873417723e-05,
+      "loss": 0.8881,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7341772151898734e-05,
+      "loss": 0.993,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.746835443037975e-05,
+      "loss": 0.9282,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.759493670886076e-05,
+      "loss": 0.8936,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.9401,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7848101265822787e-05,
+      "loss": 0.8699,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7974683544303798e-05,
+      "loss": 0.9632,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8101265822784813e-05,
+      "loss": 0.9291,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8227848101265824e-05,
+      "loss": 0.9166,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.8498,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.848101265822785e-05,
+      "loss": 0.8511,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8607594936708862e-05,
+      "loss": 0.895,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8734177215189874e-05,
+      "loss": 0.9202,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.886075949367089e-05,
+      "loss": 0.9498,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.7869,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.911392405063291e-05,
+      "loss": 0.7714,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9240506329113926e-05,
+      "loss": 0.8241,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.936708860759494e-05,
+      "loss": 0.842,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.949367088607595e-05,
+      "loss": 0.9444,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9620253164556964e-05,
+      "loss": 0.9724,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.974683544303798e-05,
+      "loss": 0.9152,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9873417721518987e-05,
+      "loss": 0.8807,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9853,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998090768233e-05,
+      "loss": 0.9975,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992363073656e-05,
+      "loss": 0.8967,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982816918462e-05,
+      "loss": 1.0061,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996945230629e-05,
+      "loss": 0.9436,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999952269242243e-05,
+      "loss": 0.9843,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999931267732887e-05,
+      "loss": 0.9997,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990644778624e-05,
+      "loss": 0.862,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999877809411777e-05,
+      "loss": 1.0233,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999845352620437e-05,
+      "loss": 0.9732,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980907742461e-05,
+      "loss": 0.9212,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999768983838154e-05,
+      "loss": 0.9099,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972507187637e-05,
+      "loss": 0.9107,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999967734155603e-05,
+      "loss": 1.0595,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999625792895357e-05,
+      "loss": 1.0027,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999957042591404e-05,
+      "loss": 1.0065,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999511240633215e-05,
+      "loss": 0.8519,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999448237075487e-05,
+      "loss": 0.7415,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938141526491e-05,
+      "loss": 0.8996,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999310775226997e-05,
+      "loss": 0.9315,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999923631698873e-05,
+      "loss": 0.8709,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915804057853e-05,
+      "loss": 0.9306,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99990759460263e-05,
+      "loss": 0.9007,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998990033363373e-05,
+      "loss": 0.97,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998900302622567e-05,
+      "loss": 0.8856,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998806753838136e-05,
+      "loss": 0.9241,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998709387045803e-05,
+      "loss": 0.843,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999860820228275e-05,
+      "loss": 0.9839,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998503199587617e-05,
+      "loss": 0.9221,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998394379000493e-05,
+      "loss": 0.8016,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999828174056293e-05,
+      "loss": 0.9689,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998165284317944e-05,
+      "loss": 0.8875,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999804501031e-05,
+      "loss": 0.8945,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997920918585024e-05,
+      "loss": 0.9237,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997793009190403e-05,
+      "loss": 0.8553,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997661282174976e-05,
+      "loss": 0.8878,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997525737589045e-05,
+      "loss": 0.8981,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999738637548436e-05,
+      "loss": 0.9891,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997243195914143e-05,
+      "loss": 0.8996,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997096198933068e-05,
+      "loss": 0.9707,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996945384597256e-05,
+      "loss": 0.9183,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996790752964305e-05,
+      "loss": 0.8894,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996632304093256e-05,
+      "loss": 0.8659,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996470038044612e-05,
+      "loss": 0.8718,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999630395488034e-05,
+      "loss": 0.9034,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996134054663843e-05,
+      "loss": 0.9408,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995960337460014e-05,
+      "loss": 0.9695,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999578280333517e-05,
+      "loss": 0.8561,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995601452357117e-05,
+      "loss": 0.973,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995416284595097e-05,
+      "loss": 0.9637,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995227300119814e-05,
+      "loss": 0.8834,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999503449900343e-05,
+      "loss": 0.9224,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999483788131957e-05,
+      "loss": 0.8819,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994637447143307e-05,
+      "loss": 1.0245,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994433196551183e-05,
+      "loss": 0.8975,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994225129621182e-05,
+      "loss": 0.8831,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994013246432758e-05,
+      "loss": 0.7574,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993797547066817e-05,
+      "loss": 0.8653,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993578031605724e-05,
+      "loss": 0.9642,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99933547001333e-05,
+      "loss": 0.9011,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999312755273482e-05,
+      "loss": 0.9055,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992896589497024e-05,
+      "loss": 0.8807,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99926618105081e-05,
+      "loss": 0.9415,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992423215857705e-05,
+      "loss": 0.7888,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992180805636936e-05,
+      "loss": 0.8351,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991934579938364e-05,
+      "loss": 0.8143,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991684538856005e-05,
+      "loss": 0.9831,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991430682485337e-05,
+      "loss": 0.9697,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99911730109233e-05,
+      "loss": 0.8247,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990911524268273e-05,
+      "loss": 0.7876,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999064622262011e-05,
+      "loss": 0.951,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999037710608012e-05,
+      "loss": 0.9726,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999010417475106e-05,
+      "loss": 0.9667,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998982742873715e-05,
+      "loss": 0.9056,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998954686814406e-05,
+      "loss": 0.8083,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989262493078924e-05,
+      "loss": 0.7514,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988974303650327e-05,
+      "loss": 0.9177,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988682299968314e-05,
+      "loss": 0.8688,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998838648214439e-05,
+      "loss": 0.8181,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988086850291508e-05,
+      "loss": 0.9134,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998778340452408e-05,
+      "loss": 0.971,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998747614495798e-05,
+      "loss": 0.8749,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998716507171053e-05,
+      "loss": 1.0306,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986850184900516e-05,
+      "loss": 0.9698,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998653148464817e-05,
+      "loss": 0.8632,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986208971075193e-05,
+      "loss": 0.9349,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985882644304734e-05,
+      "loss": 0.8836,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985552504461398e-05,
+      "loss": 0.8282,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998521855167125e-05,
+      "loss": 0.7886,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984880786061806e-05,
+      "loss": 0.8674,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984539207762042e-05,
+      "loss": 0.9011,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984193816902388e-05,
+      "loss": 0.894,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998384461361473e-05,
+      "loss": 0.9194,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998349159803241e-05,
+      "loss": 0.7983,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983134770290232e-05,
+      "loss": 0.8272,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998277413052444e-05,
+      "loss": 0.8949,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982409678872745e-05,
+      "loss": 0.8098,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982041415474314e-05,
+      "loss": 0.8853,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981669340469766e-05,
+      "loss": 0.9605,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981293454001178e-05,
+      "loss": 0.8965,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998091375621208e-05,
+      "loss": 0.8342,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980530247247455e-05,
+      "loss": 0.7993,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980142927253747e-05,
+      "loss": 0.8637,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979751796378856e-05,
+      "loss": 0.9392,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979356854772128e-05,
+      "loss": 0.8834,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978958102584375e-05,
+      "loss": 0.9399,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997855553996786e-05,
+      "loss": 0.9309,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978149167076292e-05,
+      "loss": 0.9042,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997773898406485e-05,
+      "loss": 0.8631,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977324991090156e-05,
+      "loss": 0.8646,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99769071883103e-05,
+      "loss": 0.7917,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997648557588481e-05,
+      "loss": 0.8992,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997606015397468e-05,
+      "loss": 0.8579,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975630922742356e-05,
+      "loss": 0.8564,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997519788235174e-05,
+      "loss": 0.8275,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974761032968186e-05,
+      "loss": 0.7729,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99743203747585e-05,
+      "loss": 1.0397,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997387590789095e-05,
+      "loss": 0.9935,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997342763253525e-05,
+      "loss": 0.8763,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997297554886258e-05,
+      "loss": 0.9634,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972519657045555e-05,
+      "loss": 0.8955,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997205995725827e-05,
+      "loss": 0.8844,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971596449676242e-05,
+      "loss": 0.8966,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971129134476474e-05,
+      "loss": 0.842,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970658011837404e-05,
+      "loss": 0.86,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997018308193893e-05,
+      "loss": 0.9316,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969704344962398e-05,
+      "loss": 0.8947,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969221801090618e-05,
+      "loss": 0.9104,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968735450507842e-05,
+      "loss": 0.8956,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996824529339978e-05,
+      "loss": 1.0158,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967751329953605e-05,
+      "loss": 0.8169,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996725356035793e-05,
+      "loss": 0.9377,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966751984802825e-05,
+      "loss": 0.8307,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966246603479815e-05,
+      "loss": 0.7905,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996573741658188e-05,
+      "loss": 0.8756,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965224424303454e-05,
+      "loss": 0.8738,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964707626840415e-05,
+      "loss": 0.82,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964187024390106e-05,
+      "loss": 0.914,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996366261715131e-05,
+      "loss": 0.8073,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963134405324278e-05,
+      "loss": 0.7806,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99626023891107e-05,
+      "loss": 0.8826,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996206656871373e-05,
+      "loss": 0.9394,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961526944337964e-05,
+      "loss": 0.8505,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960983516189455e-05,
+      "loss": 0.9137,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960436284475712e-05,
+      "loss": 0.8358,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959885249405694e-05,
+      "loss": 0.8999,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995933041118981e-05,
+      "loss": 0.8396,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958771770039922e-05,
+      "loss": 0.9135,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958209326169348e-05,
+      "loss": 0.8667,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957643079792853e-05,
+      "loss": 0.9333,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957073031126657e-05,
+      "loss": 0.8519,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956499180388428e-05,
+      "loss": 0.9115,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955921527797296e-05,
+      "loss": 0.9438,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995534007357383e-05,
+      "loss": 0.7439,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954754817940054e-05,
+      "loss": 0.8625,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995416576111945e-05,
+      "loss": 0.8283,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995357290333695e-05,
+      "loss": 0.8585,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952976244818927e-05,
+      "loss": 0.79,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952375785793218e-05,
+      "loss": 0.9328,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951771526489102e-05,
+      "loss": 0.856,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995116346713732e-05,
+      "loss": 0.922,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995055160797005e-05,
+      "loss": 0.8451,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949935949220936e-05,
+      "loss": 0.8757,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949316491125054e-05,
+      "loss": 0.9382,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 0.7886,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948066177840615e-05,
+      "loss": 0.9951,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947435323129484e-05,
+      "loss": 1.0173,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994680067002644e-05,
+      "loss": 0.9075,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946162218773836e-05,
+      "loss": 0.9863,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994551996961545e-05,
+      "loss": 0.8426,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994487392279653e-05,
+      "loss": 0.8494,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944224078563766e-05,
+      "loss": 0.9248,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943570437165296e-05,
+      "loss": 0.8674,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942912998850712e-05,
+      "loss": 0.8931,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942251763871056e-05,
+      "loss": 0.9923,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941586732478816e-05,
+      "loss": 0.8166,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994091790492793e-05,
+      "loss": 0.9534,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940245281473794e-05,
+      "loss": 0.8573,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993956886237324e-05,
+      "loss": 0.8094,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938888647884558e-05,
+      "loss": 0.7379,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938204638267487e-05,
+      "loss": 0.9742,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993751683378321e-05,
+      "loss": 0.9259,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993682523469437e-05,
+      "loss": 1.0267,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936129841265046e-05,
+      "loss": 0.9165,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935430653760772e-05,
+      "loss": 0.8586,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993472767244853e-05,
+      "loss": 0.8392,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934020897596752e-05,
+      "loss": 0.8932,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933310329475317e-05,
+      "loss": 1.0344,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993259596835555e-05,
+      "loss": 0.8339,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931877814510235e-05,
+      "loss": 0.8046,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993115586821359e-05,
+      "loss": 0.8346,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930430129741284e-05,
+      "loss": 0.9228,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929700599370444e-05,
+      "loss": 0.9464,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992896727737964e-05,
+      "loss": 0.8238,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928230164048885e-05,
+      "loss": 0.8253,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927489259659642e-05,
+      "loss": 0.9527,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926744564494825e-05,
+      "loss": 0.8834,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925996078838792e-05,
+      "loss": 0.907,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992524380297735e-05,
+      "loss": 0.8116,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924487737197748e-05,
+      "loss": 0.7658,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923727881788695e-05,
+      "loss": 0.8981,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922964237040338e-05,
+      "loss": 0.8273,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922196803244264e-05,
+      "loss": 0.8312,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921425580693524e-05,
+      "loss": 0.7599,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99206505696826e-05,
+      "loss": 0.9041,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991987177050743e-05,
+      "loss": 0.8762,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99190891834654e-05,
+      "loss": 0.9716,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991830280885533e-05,
+      "loss": 0.8846,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917512646977494e-05,
+      "loss": 0.8052,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916718698133623e-05,
+      "loss": 0.8742,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991592096262687e-05,
+      "loss": 0.892,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915119440761858e-05,
+      "loss": 0.7269,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914314132844644e-05,
+      "loss": 0.8377,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913505039182723e-05,
+      "loss": 0.7549,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912692160085054e-05,
+      "loss": 0.8101,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991187549586203e-05,
+      "loss": 0.874,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911055046825486e-05,
+      "loss": 0.8819,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910230813288713e-05,
+      "loss": 0.8749,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909402795566438e-05,
+      "loss": 0.883,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908570993974843e-05,
+      "loss": 0.7547,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990773540883154e-05,
+      "loss": 0.8506,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906896040455598e-05,
+      "loss": 0.8359,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906052889167528e-05,
+      "loss": 1.0027,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990520595528928e-05,
+      "loss": 0.8862,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990435523914426e-05,
+      "loss": 0.8742,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903500741057304e-05,
+      "loss": 0.8897,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902642461354702e-05,
+      "loss": 1.0068,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901780400364185e-05,
+      "loss": 0.8926,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900914558414923e-05,
+      "loss": 0.9336,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900044935837545e-05,
+      "loss": 0.8886,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899171532964103e-05,
+      "loss": 0.8957,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989829435012811e-05,
+      "loss": 0.8949,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897413387664506e-05,
+      "loss": 0.8441,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896528645909692e-05,
+      "loss": 0.9804,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895640125201498e-05,
+      "loss": 0.8643,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894747825879207e-05,
+      "loss": 0.8541,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893851748283538e-05,
+      "loss": 0.9127,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892951892756657e-05,
+      "loss": 0.9343,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892048259642168e-05,
+      "loss": 0.9273,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989114084928512e-05,
+      "loss": 0.8965,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890229662032e-05,
+      "loss": 0.8891,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889314698230754e-05,
+      "loss": 0.9071,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888395958230745e-05,
+      "loss": 0.8005,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98874734423828e-05,
+      "loss": 0.8678,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988654715103917e-05,
+      "loss": 0.7468,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988561708455356e-05,
+      "loss": 0.8631,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884683243281117e-05,
+      "loss": 0.7985,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883745627578417e-05,
+      "loss": 0.8094,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882804237803487e-05,
+      "loss": 0.8844,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881859074315796e-05,
+      "loss": 0.8079,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880910137476248e-05,
+      "loss": 0.8202,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879957427647198e-05,
+      "loss": 0.8269,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879000945192425e-05,
+      "loss": 0.7645,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878040690477163e-05,
+      "loss": 0.8737,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877076663868084e-05,
+      "loss": 0.9529,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876108865733296e-05,
+      "loss": 0.8213,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875137296442348e-05,
+      "loss": 0.8422,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874161956366232e-05,
+      "loss": 0.8557,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873182845877376e-05,
+      "loss": 0.849,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872199965349653e-05,
+      "loss": 0.8419,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871213315158367e-05,
+      "loss": 0.9361,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870222895680277e-05,
+      "loss": 0.8624,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869228707293557e-05,
+      "loss": 0.8496,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868230750377844e-05,
+      "loss": 0.7979,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867229025314204e-05,
+      "loss": 0.8177,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866223532485134e-05,
+      "loss": 0.806,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865214272274586e-05,
+      "loss": 0.8343,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986420124506794e-05,
+      "loss": 0.8338,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986318445125202e-05,
+      "loss": 0.9522,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862163891215077e-05,
+      "loss": 0.8374,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986113956534682e-05,
+      "loss": 0.9264,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860111474038367e-05,
+      "loss": 0.8167,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859079617682307e-05,
+      "loss": 0.9134,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858043996672643e-05,
+      "loss": 0.8388,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857004611404825e-05,
+      "loss": 0.9779,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855961462275733e-05,
+      "loss": 0.7697,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854914549683702e-05,
+      "loss": 0.89,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985386387402848e-05,
+      "loss": 0.8055,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985280943571127e-05,
+      "loss": 0.8452,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98517512351347e-05,
+      "loss": 0.8629,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850689272702846e-05,
+      "loss": 0.787,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849623548821212e-05,
+      "loss": 0.692,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848554063896743e-05,
+      "loss": 0.9159,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984748081833781e-05,
+      "loss": 0.8768,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984640381255424e-05,
+      "loss": 0.9299,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984532304695727e-05,
+      "loss": 0.9272,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98442385219596e-05,
+      "loss": 0.844,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843150237975343e-05,
+      "loss": 0.7654,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842058195420058e-05,
+      "loss": 0.8767,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984096239471074e-05,
+      "loss": 0.9011,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839862836265816e-05,
+      "loss": 0.7565,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838759520505143e-05,
+      "loss": 0.9207,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837652447850026e-05,
+      "loss": 0.9395,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836541618723193e-05,
+      "loss": 0.9037,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835427033548807e-05,
+      "loss": 0.8536,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834308692752475e-05,
+      "loss": 0.8862,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833186596761224e-05,
+      "loss": 0.8837,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983206074600353e-05,
+      "loss": 0.9798,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983093114090929e-05,
+      "loss": 0.824,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982979778190984e-05,
+      "loss": 0.9269,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828660669437947e-05,
+      "loss": 0.8106,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827519803927818e-05,
+      "loss": 0.9076,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826375185815087e-05,
+      "loss": 0.9331,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982522681553682e-05,
+      "loss": 0.8343,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982407469353152e-05,
+      "loss": 0.8536,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982291882023912e-05,
+      "loss": 0.92,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821759196100984e-05,
+      "loss": 0.7606,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820595821559913e-05,
+      "loss": 0.8108,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819428697060138e-05,
+      "loss": 0.8888,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981825782304732e-05,
+      "loss": 0.7788,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817083199968552e-05,
+      "loss": 0.9109,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981590482827236e-05,
+      "loss": 0.9902,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98147227084087e-05,
+      "loss": 0.9627,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981353684082896e-05,
+      "loss": 0.8629,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812347225985966e-05,
+      "loss": 0.8719,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981115386433396e-05,
+      "loss": 0.992,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980995675632863e-05,
+      "loss": 0.9176,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808755902427075e-05,
+      "loss": 0.9414,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807551303087846e-05,
+      "loss": 0.8387,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806342958770917e-05,
+      "loss": 0.9284,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805130869937687e-05,
+      "loss": 0.9018,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803915037050983e-05,
+      "loss": 0.8736,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980269546057507e-05,
+      "loss": 0.9078,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980147214097564e-05,
+      "loss": 0.7421,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800245078719814e-05,
+      "loss": 0.8439,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979901427427614e-05,
+      "loss": 0.8168,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979777972811459e-05,
+      "loss": 0.858,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796541440706582e-05,
+      "loss": 0.8893,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 0.8986,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794053644043947e-05,
+      "loss": 0.7316,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792804135739275e-05,
+      "loss": 0.8958,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979155088808805e-05,
+      "loss": 0.9101,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979029390156882e-05,
+      "loss": 0.9764,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789033176661566e-05,
+      "loss": 0.79,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787768713847685e-05,
+      "loss": 0.875,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978650051361001e-05,
+      "loss": 0.8419,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97852285764328e-05,
+      "loss": 0.7961,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783952902801736e-05,
+      "loss": 0.8984,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782673493203933e-05,
+      "loss": 0.8753,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978139034812793e-05,
+      "loss": 0.9221,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780103468063683e-05,
+      "loss": 0.8694,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778812853502592e-05,
+      "loss": 0.8118,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977751850493747e-05,
+      "loss": 0.8738,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776220422862558e-05,
+      "loss": 0.9876,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774918607773524e-05,
+      "loss": 0.8504,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977361306016746e-05,
+      "loss": 0.8682,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977230378054289e-05,
+      "loss": 0.8015,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770990769399755e-05,
+      "loss": 0.8752,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976967402723942e-05,
+      "loss": 0.6871,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768353554564685e-05,
+      "loss": 0.8271,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767029351879755e-05,
+      "loss": 0.8624,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976570141969029e-05,
+      "loss": 0.8852,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764369758503342e-05,
+      "loss": 0.8924,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976303436882741e-05,
+      "loss": 0.8074,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761695251172398e-05,
+      "loss": 0.8528,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976035240604965e-05,
+      "loss": 0.865,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759005833971928e-05,
+      "loss": 0.8535,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757655535453408e-05,
+      "loss": 0.9436,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756301511009704e-05,
+      "loss": 0.8589,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754943761157844e-05,
+      "loss": 0.9067,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753582286416274e-05,
+      "loss": 0.8665,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752217087304875e-05,
+      "loss": 0.7335,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750848164344946e-05,
+      "loss": 0.8763,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749475518059194e-05,
+      "loss": 0.8321,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748099148971766e-05,
+      "loss": 0.7964,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746719057608222e-05,
+      "loss": 0.8417,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745335244495546e-05,
+      "loss": 0.9174,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743947710162145e-05,
+      "loss": 0.8874,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974255645513784e-05,
+      "loss": 0.9759,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741161479953872e-05,
+      "loss": 0.7536,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739762785142916e-05,
+      "loss": 0.8778,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738360371239055e-05,
+      "loss": 0.8271,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736954238777793e-05,
+      "loss": 0.8372,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973554438829606e-05,
+      "loss": 0.8423,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97341308203322e-05,
+      "loss": 0.8273,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973271353542598e-05,
+      "loss": 0.7794,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973129253411859e-05,
+      "loss": 0.8248,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729867816952624e-05,
+      "loss": 0.88,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972843938447211e-05,
+      "loss": 1.0367,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727007237222492e-05,
+      "loss": 0.8328,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972557137575063e-05,
+      "loss": 0.8816,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97241318006048e-05,
+      "loss": 0.8729,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722688512334696e-05,
+      "loss": 0.8096,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972124151149144e-05,
+      "loss": 0.8563,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719790798627555e-05,
+      "loss": 0.9342,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718336374296994e-05,
+      "loss": 0.88,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971687823905513e-05,
+      "loss": 0.8379,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971541639345874e-05,
+      "loss": 0.8938,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713950838066024e-05,
+      "loss": 0.8644,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712481573436602e-05,
+      "loss": 0.778,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711008600131503e-05,
+      "loss": 0.8745,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970953191871318e-05,
+      "loss": 0.8941,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708051529745506e-05,
+      "loss": 0.8006,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706567433793743e-05,
+      "loss": 0.8334,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705079631424605e-05,
+      "loss": 0.8333,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703588123206193e-05,
+      "loss": 0.8266,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702092909708043e-05,
+      "loss": 0.765,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970059399150109e-05,
+      "loss": 0.8149,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699091369157695e-05,
+      "loss": 0.8365,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697585043251625e-05,
+      "loss": 0.9393,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696075014358067e-05,
+      "loss": 0.8231,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694561283053616e-05,
+      "loss": 0.9404,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693043849916295e-05,
+      "loss": 0.9082,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969152271552552e-05,
+      "loss": 0.9073,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689997880462134e-05,
+      "loss": 0.9301,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968846934530839e-05,
+      "loss": 0.7999,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686937110647955e-05,
+      "loss": 0.9224,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685401177065908e-05,
+      "loss": 0.8258,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683861545148738e-05,
+      "loss": 0.839,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682318215484344e-05,
+      "loss": 0.8844,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680771188662044e-05,
+      "loss": 0.7882,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679220465272566e-05,
+      "loss": 0.8968,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677666045908047e-05,
+      "loss": 0.924,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676107931162037e-05,
+      "loss": 0.7986,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674546121629495e-05,
+      "loss": 0.8265,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672980617906793e-05,
+      "loss": 0.8592,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967141142059171e-05,
+      "loss": 0.8554,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669838530283445e-05,
+      "loss": 0.8876,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668261947582598e-05,
+      "loss": 0.9342,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966668167309117e-05,
+      "loss": 0.8689,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665097707412604e-05,
+      "loss": 0.6963,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663510051151716e-05,
+      "loss": 0.9039,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661918704914752e-05,
+      "loss": 0.8406,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966032366930936e-05,
+      "loss": 0.9266,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658724944944597e-05,
+      "loss": 0.9213,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657122532430937e-05,
+      "loss": 0.806,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655516432380252e-05,
+      "loss": 0.8634,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965390664540582e-05,
+      "loss": 0.84,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965229317212234e-05,
+      "loss": 0.879,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650676013145906e-05,
+      "loss": 0.8061,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649055169094026e-05,
+      "loss": 0.8807,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647430640585613e-05,
+      "loss": 0.7826,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645802428240986e-05,
+      "loss": 0.8482,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644170532681875e-05,
+      "loss": 0.8978,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964253495453141e-05,
+      "loss": 0.916,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640895694414137e-05,
+      "loss": 0.7523,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639252752955992e-05,
+      "loss": 0.9795,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637606130784333e-05,
+      "loss": 0.8361,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635955828527917e-05,
+      "loss": 0.8567,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96343018468169e-05,
+      "loss": 0.9508,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632644186282853e-05,
+      "loss": 0.8264,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630982847558748e-05,
+      "loss": 0.811,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962931783127896e-05,
+      "loss": 0.8174,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962764913807927e-05,
+      "loss": 0.8888,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625976768596862e-05,
+      "loss": 0.8856,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624300723470323e-05,
+      "loss": 0.866,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962262100333965e-05,
+      "loss": 0.8663,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962093760884623e-05,
+      "loss": 0.884,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619250540632867e-05,
+      "loss": 0.8558,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617559799343757e-05,
+      "loss": 0.8925,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961586538562451e-05,
+      "loss": 0.8965,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614167300122126e-05,
+      "loss": 0.9117,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612465543485016e-05,
+      "loss": 0.8997,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610760116362987e-05,
+      "loss": 0.8604,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609051019407254e-05,
+      "loss": 0.9466,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607338253270422e-05,
+      "loss": 0.7942,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605621818606516e-05,
+      "loss": 0.7554,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960390171607094e-05,
+      "loss": 0.9433,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960217794632051e-05,
+      "loss": 0.837,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960045051001345e-05,
+      "loss": 0.8773,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959871940780936e-05,
+      "loss": 0.9087,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596984640369273e-05,
+      "loss": 0.9395,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595246208355588e-05,
+      "loss": 0.9535,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959350411243213e-05,
+      "loss": 0.8224,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591758353264106e-05,
+      "loss": 0.8244,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590008931518133e-05,
+      "loss": 0.8913,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588255847862213e-05,
+      "loss": 0.805,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958649910296576e-05,
+      "loss": 0.8603,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958473869749958e-05,
+      "loss": 0.9472,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582974632135882e-05,
+      "loss": 0.8291,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958120690754826e-05,
+      "loss": 0.8197,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579435524411715e-05,
+      "loss": 0.8096,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577660483402643e-05,
+      "loss": 0.8477,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575881785198842e-05,
+      "loss": 0.89,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574099430479498e-05,
+      "loss": 0.8762,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572313419925194e-05,
+      "loss": 0.9178,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570523754217917e-05,
+      "loss": 0.8054,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568730434041044e-05,
+      "loss": 0.8859,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956693346007934e-05,
+      "loss": 0.8427,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956513283301898e-05,
+      "loss": 0.8665,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563328553547526e-05,
+      "loss": 0.7718,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561520622353935e-05,
+      "loss": 0.8511,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955970904012856e-05,
+      "loss": 0.941,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557893807563145e-05,
+      "loss": 0.9027,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556074925350826e-05,
+      "loss": 0.8824,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554252394186147e-05,
+      "loss": 0.9203,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552426214765025e-05,
+      "loss": 0.8586,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955059638778479e-05,
+      "loss": 0.8687,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548762913944143e-05,
+      "loss": 0.8433,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546925793943198e-05,
+      "loss": 0.8114,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545085028483447e-05,
+      "loss": 0.8214,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543240618267785e-05,
+      "loss": 0.8372,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 0.8443,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539540866387234e-05,
+      "loss": 0.9573,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537685526135088e-05,
+      "loss": 0.8909,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953582654395249e-05,
+      "loss": 0.8022,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533963920549307e-05,
+      "loss": 0.8664,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953209765663676e-05,
+      "loss": 0.8765,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953022775292748e-05,
+      "loss": 0.9619,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528354210135486e-05,
+      "loss": 0.8267,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526477028976178e-05,
+      "loss": 0.801,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524596210166352e-05,
+      "loss": 0.9379,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522711754424194e-05,
+      "loss": 0.952,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952082366246928e-05,
+      "loss": 0.8058,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951893193502256e-05,
+      "loss": 0.8924,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951703657280639e-05,
+      "loss": 0.8922,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515137576544512e-05,
+      "loss": 0.7967,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951323494696204e-05,
+      "loss": 0.9335,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511328684785494e-05,
+      "loss": 0.8403,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509418790742772e-05,
+      "loss": 0.9339,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507505265563158e-05,
+      "loss": 0.8543,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505588109977324e-05,
+      "loss": 0.7862,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503667324717335e-05,
+      "loss": 1.0143,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501742910516625e-05,
+      "loss": 0.8129,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499814868110035e-05,
+      "loss": 0.759,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949788319823378e-05,
+      "loss": 0.8619,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949594790162545e-05,
+      "loss": 0.8756,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494008979024047e-05,
+      "loss": 0.7913,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949206643116993e-05,
+      "loss": 0.8479,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490120258804855e-05,
+      "loss": 0.905,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948817046267197e-05,
+      "loss": 0.7462,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486217043515785e-05,
+      "loss": 0.9182,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484260002082213e-05,
+      "loss": 0.9308,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482299339118543e-05,
+      "loss": 0.8511,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480335055373444e-05,
+      "loss": 0.8897,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478367151596976e-05,
+      "loss": 0.7717,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476395628540566e-05,
+      "loss": 0.8871,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474420486957045e-05,
+      "loss": 0.9168,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472441727600606e-05,
+      "loss": 0.9214,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947045935122683e-05,
+      "loss": 0.8248,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468473358592687e-05,
+      "loss": 0.9178,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466483750456515e-05,
+      "loss": 0.8942,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946449052757804e-05,
+      "loss": 0.8313,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462493690718373e-05,
+      "loss": 0.7773,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460493240639985e-05,
+      "loss": 0.8485,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458489178106753e-05,
+      "loss": 0.8364,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945648150388392e-05,
+      "loss": 0.9246,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454470218738103e-05,
+      "loss": 0.8367,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452455323437306e-05,
+      "loss": 0.7416,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945043681875091e-05,
+      "loss": 0.8028,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944841470544968e-05,
+      "loss": 0.7701,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944638898430574e-05,
+      "loss": 0.7732,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444359656092616e-05,
+      "loss": 0.9713,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944232672158519e-05,
+      "loss": 0.9422,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440290181559737e-05,
+      "loss": 0.8649,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94382500367939e-05,
+      "loss": 0.8811,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94362062880667e-05,
+      "loss": 0.918,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943415893615854e-05,
+      "loss": 0.7408,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432107981851186e-05,
+      "loss": 0.8818,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430053425927794e-05,
+      "loss": 1.014,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427995269172885e-05,
+      "loss": 0.7456,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425933512372364e-05,
+      "loss": 0.8399,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423868156313498e-05,
+      "loss": 0.8426,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421799201784937e-05,
+      "loss": 0.7727,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419726649576707e-05,
+      "loss": 0.7775,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417650500480207e-05,
+      "loss": 0.9041,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94155707552882e-05,
+      "loss": 0.9073,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941348741479483e-05,
+      "loss": 0.8466,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411400479795618e-05,
+      "loss": 0.8306,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409309951087447e-05,
+      "loss": 0.7874,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407215829468585e-05,
+      "loss": 0.7925,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405118115738657e-05,
+      "loss": 0.877,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940301681069867e-05,
+      "loss": 0.8734,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400911915151e-05,
+      "loss": 0.8589,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93988034298994e-05,
+      "loss": 0.9274,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396691355748975e-05,
+      "loss": 0.7162,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394575693506224e-05,
+      "loss": 0.7826,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392456443978995e-05,
+      "loss": 0.8567,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390333607976522e-05,
+      "loss": 0.9274,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388207186309404e-05,
+      "loss": 0.9104,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93860771797896e-05,
+      "loss": 0.8176,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938394358923045e-05,
+      "loss": 0.7258,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938180641544666e-05,
+      "loss": 0.7345,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93796656592543e-05,
+      "loss": 0.8032,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377521321470806e-05,
+      "loss": 0.821,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937537340291499e-05,
+      "loss": 0.7751,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373221904407026e-05,
+      "loss": 0.8405,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371066826768455e-05,
+      "loss": 0.9738,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368908170822187e-05,
+      "loss": 0.9642,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366745937392496e-05,
+      "loss": 0.8139,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364580127305023e-05,
+      "loss": 0.8747,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362410741386773e-05,
+      "loss": 0.7929,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936023778046612e-05,
+      "loss": 0.824,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93580612453728e-05,
+      "loss": 0.8532,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935588113693792e-05,
+      "loss": 0.9249,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353697455993942e-05,
+      "loss": 0.8415,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351510203374694e-05,
+      "loss": 0.8111,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934931937991537e-05,
+      "loss": 0.7958,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934712498645254e-05,
+      "loss": 0.8857,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344927023824112e-05,
+      "loss": 0.8223,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342725492869375e-05,
+      "loss": 0.8344,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340520394428973e-05,
+      "loss": 0.7711,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933831172934492e-05,
+      "loss": 0.8884,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933609949846058e-05,
+      "loss": 0.8438,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333883702620692e-05,
+      "loss": 0.8,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331664342671348e-05,
+      "loss": 0.8226,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932944141946e-05,
+      "loss": 0.8557,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327214933835463e-05,
+      "loss": 0.7826,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932498488664791e-05,
+      "loss": 0.8345,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322751278748883e-05,
+      "loss": 0.9069,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320514110991275e-05,
+      "loss": 0.7814,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318273384229338e-05,
+      "loss": 0.8984,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316029099318688e-05,
+      "loss": 0.7874,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931378125711629e-05,
+      "loss": 0.7952,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311529858480488e-05,
+      "loss": 0.8134,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930927490427096e-05,
+      "loss": 0.8126,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307016395348753e-05,
+      "loss": 0.9244,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930475433257627e-05,
+      "loss": 0.9365,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302488716817275e-05,
+      "loss": 0.7872,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300219548936883e-05,
+      "loss": 0.8682,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929794682980157e-05,
+      "loss": 0.7919,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295670560279162e-05,
+      "loss": 0.7786,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293390741238844e-05,
+      "loss": 0.7463,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929110737355116e-05,
+      "loss": 0.9107,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288820458088004e-05,
+      "loss": 0.9787,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286529995722624e-05,
+      "loss": 0.855,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284235987329626e-05,
+      "loss": 0.8221,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281938433784973e-05,
+      "loss": 0.7801,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927963733596597e-05,
+      "loss": 0.8941,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927733269475129e-05,
+      "loss": 0.7562,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927502451102095e-05,
+      "loss": 0.8847,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272712785656315e-05,
+      "loss": 0.8316,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927039751954012e-05,
+      "loss": 0.8961,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926807871355643e-05,
+      "loss": 0.7676,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926575636859068e-05,
+      "loss": 0.8181,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263430485529648e-05,
+      "loss": 0.8781,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926110106526146e-05,
+      "loss": 0.7112,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258768108675596e-05,
+      "loss": 0.8886,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256431616662895e-05,
+      "loss": 0.784,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925409159011553e-05,
+      "loss": 0.747,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251748029927038e-05,
+      "loss": 0.8438,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249400936992294e-05,
+      "loss": 0.9242,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247050312207526e-05,
+      "loss": 0.8176,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924469615647032e-05,
+      "loss": 0.8905,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924233847067959e-05,
+      "loss": 0.9231,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239977255735617e-05,
+      "loss": 0.8539,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237612512540022e-05,
+      "loss": 0.8846,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235244241995774e-05,
+      "loss": 0.918,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232872445007184e-05,
+      "loss": 0.9467,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230497122479917e-05,
+      "loss": 0.8647,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922811827532098e-05,
+      "loss": 0.9315,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922573590443873e-05,
+      "loss": 0.9049,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223350010742864e-05,
+      "loss": 0.8442,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922096059514443e-05,
+      "loss": 0.8559,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218567658555813e-05,
+      "loss": 0.8142,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216171201890748e-05,
+      "loss": 0.736,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213771226064318e-05,
+      "loss": 0.8253,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921136773199294e-05,
+      "loss": 0.8846,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920896072059438e-05,
+      "loss": 0.8662,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920655019278775e-05,
+      "loss": 0.8169,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92041361494935e-05,
+      "loss": 0.7746,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201718591633417e-05,
+      "loss": 0.7855,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199297520130647e-05,
+      "loss": 0.8227,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919687293590966e-05,
+      "loss": 0.8599,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919444483989628e-05,
+      "loss": 0.8873,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919201323301766e-05,
+      "loss": 0.8039,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 0.8246,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187139490380056e-05,
+      "loss": 0.8098,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184697356482094e-05,
+      "loss": 0.8131,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182251715440934e-05,
+      "loss": 0.8268,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179802568190437e-05,
+      "loss": 0.8521,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177349915665805e-05,
+      "loss": 0.8739,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174893758803565e-05,
+      "loss": 0.8401,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172434098541603e-05,
+      "loss": 1.0006,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169970935819123e-05,
+      "loss": 0.8921,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916750427157668e-05,
+      "loss": 0.8645,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916503410675616e-05,
+      "loss": 0.9354,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916256044230078e-05,
+      "loss": 0.8281,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916008327915511e-05,
+      "loss": 0.8659,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915760261826504e-05,
+      "loss": 0.9498,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91551184605778e-05,
+      "loss": 0.8927,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152630807041956e-05,
+      "loss": 0.9063,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150139658607414e-05,
+      "loss": 0.9363,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147645016225407e-05,
+      "loss": 0.8232,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145146880848505e-05,
+      "loss": 0.9036,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142645253430613e-05,
+      "loss": 0.7792,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914014013492697e-05,
+      "loss": 0.827,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137631526294146e-05,
+      "loss": 0.8052,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913511942849004e-05,
+      "loss": 0.8094,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913260384247389e-05,
+      "loss": 0.7725,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913008476920626e-05,
+      "loss": 0.7286,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127562209649056e-05,
+      "loss": 0.9557,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125036164765502e-05,
+      "loss": 0.9138,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122506635520162e-05,
+      "loss": 0.9406,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119973622878928e-05,
+      "loss": 0.8838,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117437127809017e-05,
+      "loss": 0.8448,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114897151278987e-05,
+      "loss": 0.7937,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112353694258715e-05,
+      "loss": 0.7945,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109806757719412e-05,
+      "loss": 0.8942,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107256342633615e-05,
+      "loss": 0.81,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910470244997519e-05,
+      "loss": 0.8406,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102145080719334e-05,
+      "loss": 0.8975,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909958423584257e-05,
+      "loss": 0.8998,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909701991632274e-05,
+      "loss": 0.8325,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094452123139034e-05,
+      "loss": 0.8536,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091880857271944e-05,
+      "loss": 0.8241,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089306119703296e-05,
+      "loss": 0.8308,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908672791141625e-05,
+      "loss": 0.8554,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084146233395285e-05,
+      "loss": 0.9533,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081561086626206e-05,
+      "loss": 0.8815,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907897247209614e-05,
+      "loss": 0.8467,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076380390793535e-05,
+      "loss": 0.8733,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907378484370818e-05,
+      "loss": 0.8615,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071185831831166e-05,
+      "loss": 0.8285,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068583356154917e-05,
+      "loss": 0.7282,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065977417673183e-05,
+      "loss": 0.8135,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063368017381028e-05,
+      "loss": 0.8638,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060755156274844e-05,
+      "loss": 0.8183,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058138835352343e-05,
+      "loss": 0.8526,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055519055612554e-05,
+      "loss": 0.8022,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052895818055833e-05,
+      "loss": 0.8968,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050269123683854e-05,
+      "loss": 0.7697,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047638973499612e-05,
+      "loss": 0.7711,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045005368507418e-05,
+      "loss": 0.8163,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042368309712906e-05,
+      "loss": 0.8446,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039727798123026e-05,
+      "loss": 0.88,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037083834746048e-05,
+      "loss": 0.7875,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034436420591554e-05,
+      "loss": 0.9,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903178555667046e-05,
+      "loss": 0.8121,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029131243994985e-05,
+      "loss": 0.804,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026473483578666e-05,
+      "loss": 0.8202,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023812276436363e-05,
+      "loss": 0.8433,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902114762358424e-05,
+      "loss": 0.8424,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018479526039797e-05,
+      "loss": 0.8459,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015807984821827e-05,
+      "loss": 0.8383,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013133000950455e-05,
+      "loss": 0.7628,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901045457544711e-05,
+      "loss": 0.7495,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007772709334543e-05,
+      "loss": 0.9799,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900508740363681e-05,
+      "loss": 0.7964,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900239865937929e-05,
+      "loss": 0.8489,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999706477588665e-05,
+      "loss": 0.8227,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997010859292935e-05,
+      "loss": 0.8737,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994311805521416e-05,
+      "loss": 0.876,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991609317304733e-05,
+      "loss": 0.8487,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988903395674814e-05,
+      "loss": 0.8106,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986194041664915e-05,
+      "loss": 0.8828,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983481256309585e-05,
+      "loss": 0.8469,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980765040644694e-05,
+      "loss": 0.7637,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897804539570742e-05,
+      "loss": 0.8068,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975322322536246e-05,
+      "loss": 0.8709,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897259582217097e-05,
+      "loss": 0.9078,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969865895652696e-05,
+      "loss": 0.8693,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967132544023842e-05,
+      "loss": 0.8039,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896439576832812e-05,
+      "loss": 0.8014,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961655569610557e-05,
+      "loss": 0.9531,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958911948917495e-05,
+      "loss": 0.8511,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956164907296575e-05,
+      "loss": 0.8839,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895341444579674e-05,
+      "loss": 0.7322,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895066056546824e-05,
+      "loss": 0.7201,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894790326736265e-05,
+      "loss": 0.7931,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894514255253282e-05,
+      "loss": 0.7732,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942378422032922e-05,
+      "loss": 0.8509,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893961087691843e-05,
+      "loss": 0.9019,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936839918246122e-05,
+      "loss": 0.9205,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934065547074077e-05,
+      "loss": 0.8475,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931287764461682e-05,
+      "loss": 0.6847,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892850657146962e-05,
+      "loss": 0.8022,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925721969159877e-05,
+      "loss": 0.8606,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922933958595746e-05,
+      "loss": 0.7817,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920142540841823e-05,
+      "loss": 0.9014,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917347716963995e-05,
+      "loss": 0.6944,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891454948802945e-05,
+      "loss": 0.9268,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89117478551067e-05,
+      "loss": 0.8535,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890894281926552e-05,
+      "loss": 0.7754,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906134381577008e-05,
+      "loss": 0.971,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890332254311356e-05,
+      "loss": 0.7794,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890050730494886e-05,
+      "loss": 0.8187,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897688668157903e-05,
+      "loss": 0.7396,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889486663381697e-05,
+      "loss": 0.7546,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892041203003647e-05,
+      "loss": 0.8676,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889212376796814e-05,
+      "loss": 0.8178,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888638015627665e-05,
+      "loss": 0.9569,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888354454252462e-05,
+      "loss": 0.8821,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880705536623504e-05,
+      "loss": 0.8271,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887786313965736e-05,
+      "loss": 0.9271,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875017352711547e-05,
+      "loss": 0.8523,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887216817687272e-05,
+      "loss": 0.8397,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886931561322882e-05,
+      "loss": 0.7829,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88664596628691e-05,
+      "loss": 0.8336,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863600326884085e-05,
+      "loss": 0.88,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860737606365603e-05,
+      "loss": 0.8788,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857871502406776e-05,
+      "loss": 0.8215,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885500201610202e-05,
+      "loss": 0.9155,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852129148547025e-05,
+      "loss": 0.897,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849252900838795e-05,
+      "loss": 0.8907,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846373274075612e-05,
+      "loss": 0.8604,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884349026935705e-05,
+      "loss": 0.7404,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840603887783975e-05,
+      "loss": 0.8965,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837714130458543e-05,
+      "loss": 0.8992,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834820998484193e-05,
+      "loss": 0.8364,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883192449296566e-05,
+      "loss": 0.8775,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829024615008964e-05,
+      "loss": 0.9468,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826121365721413e-05,
+      "loss": 0.8886,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88232147462116e-05,
+      "loss": 0.935,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820304757589406e-05,
+      "loss": 0.7989,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817391400966006e-05,
+      "loss": 0.9189,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881447467745385e-05,
+      "loss": 0.8053,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811554588166677e-05,
+      "loss": 0.8357,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808631134219516e-05,
+      "loss": 0.7888,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805704316728675e-05,
+      "loss": 0.8697,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802774136811746e-05,
+      "loss": 0.736,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799840595587614e-05,
+      "loss": 0.8213,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796903694176437e-05,
+      "loss": 0.8359,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793963433699662e-05,
+      "loss": 0.8648,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791019815280015e-05,
+      "loss": 0.7506,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788072840041504e-05,
+      "loss": 0.8969,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785122509109425e-05,
+      "loss": 0.8266,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878216882361035e-05,
+      "loss": 0.7975,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877921178467213e-05,
+      "loss": 0.7663,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776251393423905e-05,
+      "loss": 0.8698,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773287650996082e-05,
+      "loss": 0.8258,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770320558520363e-05,
+      "loss": 0.9131,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8767350117129717e-05,
+      "loss": 0.7702,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764376327958395e-05,
+      "loss": 0.8618,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761399192141933e-05,
+      "loss": 0.8673,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758418710817133e-05,
+      "loss": 0.8982,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875543488512209e-05,
+      "loss": 0.8434,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752447716196158e-05,
+      "loss": 0.8187,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874945720517998e-05,
+      "loss": 0.7377,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874646335321547e-05,
+      "loss": 0.8064,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 0.8578,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740465631015502e-05,
+      "loss": 0.9158,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737461763070252e-05,
+      "loss": 0.7344,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734454558757087e-05,
+      "loss": 0.8702,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731444019224296e-05,
+      "loss": 0.7952,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728430145621444e-05,
+      "loss": 0.8449,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872541293909937e-05,
+      "loss": 0.8447,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872239240081018e-05,
+      "loss": 0.8359,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719368531907253e-05,
+      "loss": 0.7806,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871634133354525e-05,
+      "loss": 0.891,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871331080688009e-05,
+      "loss": 0.7796,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710276953068967e-05,
+      "loss": 0.8072,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707239773270353e-05,
+      "loss": 0.7815,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870419926864398e-05,
+      "loss": 0.8472,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701155440350854e-05,
+      "loss": 0.8459,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869810828955325e-05,
+      "loss": 0.8067,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695057817414715e-05,
+      "loss": 0.8501,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692004025100054e-05,
+      "loss": 0.8224,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868894691377535e-05,
+      "loss": 0.8385,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868588648460795e-05,
+      "loss": 0.8318,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868282273876647e-05,
+      "loss": 0.9322,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679755677420783e-05,
+      "loss": 0.9198,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676685301742043e-05,
+      "loss": 0.8233,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867361161290266e-05,
+      "loss": 0.9551,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670534612076304e-05,
+      "loss": 0.7643,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667454300437926e-05,
+      "loss": 0.8453,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664370679163724e-05,
+      "loss": 0.8808,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661283749431178e-05,
+      "loss": 0.7291,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658193512419007e-05,
+      "loss": 0.7536,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655099969307216e-05,
+      "loss": 0.7441,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865200312127706e-05,
+      "loss": 0.8208,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648902969511064e-05,
+      "loss": 0.8756,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645799515193005e-05,
+      "loss": 0.8274,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642692759507923e-05,
+      "loss": 0.7343,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863958270364213e-05,
+      "loss": 0.815,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636469348783176e-05,
+      "loss": 0.7891,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633352696119898e-05,
+      "loss": 0.8397,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630232746842374e-05,
+      "loss": 0.8131,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627109502141942e-05,
+      "loss": 0.9266,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623982963211204e-05,
+      "loss": 0.9286,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862085313124402e-05,
+      "loss": 0.949,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617720007435497e-05,
+      "loss": 0.9374,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614583592982016e-05,
+      "loss": 0.8142,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86114438890812e-05,
+      "loss": 0.8789,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608300896931935e-05,
+      "loss": 0.8854,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605154617734363e-05,
+      "loss": 0.779,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602005052689877e-05,
+      "loss": 0.756,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598852203001123e-05,
+      "loss": 0.8172,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595696069872013e-05,
+      "loss": 0.8233,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592536654507702e-05,
+      "loss": 0.823,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589373958114598e-05,
+      "loss": 0.9103,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858620798190037e-05,
+      "loss": 0.8831,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858303872707393e-05,
+      "loss": 0.7969,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857986619484545e-05,
+      "loss": 0.8122,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857669038642635e-05,
+      "loss": 0.7465,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573511303029294e-05,
+      "loss": 0.8434,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570328945868215e-05,
+      "loss": 0.7887,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567143316158275e-05,
+      "loss": 0.913,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563954415115897e-05,
+      "loss": 0.824,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560762243958756e-05,
+      "loss": 0.7942,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557566803905764e-05,
+      "loss": 0.8413,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554368096177097e-05,
+      "loss": 0.8197,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551166121994157e-05,
+      "loss": 0.8683,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854796088257962e-05,
+      "loss": 0.7614,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544752379157383e-05,
+      "loss": 0.7496,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854154061295261e-05,
+      "loss": 0.8778,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85383255851917e-05,
+      "loss": 0.8289,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535107297102296e-05,
+      "loss": 0.8367,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853188574991329e-05,
+      "loss": 0.8495,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852866094485483e-05,
+      "loss": 0.7782,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525432883158277e-05,
+      "loss": 0.9649,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852220156605627e-05,
+      "loss": 0.8529,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518966994782666e-05,
+      "loss": 0.8134,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515729170572577e-05,
+      "loss": 0.8086,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851248809466236e-05,
+      "loss": 0.778,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509243768289596e-05,
+      "loss": 0.7992,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505996192693132e-05,
+      "loss": 0.7946,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502745369113036e-05,
+      "loss": 0.8848,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499491298790625e-05,
+      "loss": 0.8444,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849623398296846e-05,
+      "loss": 0.8411,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492973422890316e-05,
+      "loss": 0.8221,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489709619801246e-05,
+      "loss": 0.8584,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848644257494751e-05,
+      "loss": 0.8666,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483172289576624e-05,
+      "loss": 0.8697,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847989876493733e-05,
+      "loss": 0.8438,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476622002279615e-05,
+      "loss": 0.7973,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473342002854697e-05,
+      "loss": 0.7704,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470058767915028e-05,
+      "loss": 0.8251,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466772298714305e-05,
+      "loss": 0.7907,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463482596507454e-05,
+      "loss": 0.6915,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846018966255063e-05,
+      "loss": 0.8919,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845689349810124e-05,
+      "loss": 0.8371,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453594104417898e-05,
+      "loss": 0.7484,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845029148276047e-05,
+      "loss": 0.8437,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446985634390056e-05,
+      "loss": 0.8276,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443676560568975e-05,
+      "loss": 0.8147,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844036426256079e-05,
+      "loss": 0.8589,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437048741630286e-05,
+      "loss": 0.8809,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433729999043484e-05,
+      "loss": 0.9221,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430408036067636e-05,
+      "loss": 0.9512,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427082853971217e-05,
+      "loss": 0.9181,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842375445402394e-05,
+      "loss": 0.8368,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842042283749674e-05,
+      "loss": 0.9699,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841708800566178e-05,
+      "loss": 0.6868,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841374995979246e-05,
+      "loss": 0.8859,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410408701163395e-05,
+      "loss": 0.9082,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407064231050436e-05,
+      "loss": 0.7918,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403716550730656e-05,
+      "loss": 0.7642,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400365661482352e-05,
+      "loss": 0.8843,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839701156458505e-05,
+      "loss": 0.8636,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393654261319504e-05,
+      "loss": 0.8742,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839029375296768e-05,
+      "loss": 0.8762,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386930040812782e-05,
+      "loss": 0.8098,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383563126139225e-05,
+      "loss": 0.8147,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380193010232664e-05,
+      "loss": 0.8205,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376819694379953e-05,
+      "loss": 0.8855,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837344317986919e-05,
+      "loss": 0.8338,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370063467989682e-05,
+      "loss": 0.7659,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366680560031954e-05,
+      "loss": 0.8556,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363294457287766e-05,
+      "loss": 0.7828,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359905161050085e-05,
+      "loss": 0.8684,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83565126726131e-05,
+      "loss": 0.8701,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835311699327222e-05,
+      "loss": 0.8065,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349718124324075e-05,
+      "loss": 0.7746,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834631606706651e-05,
+      "loss": 0.8585,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834291082279859e-05,
+      "loss": 0.7713,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833950239282059e-05,
+      "loss": 0.8318,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336090778434014e-05,
+      "loss": 0.7992,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332675980941565e-05,
+      "loss": 0.7812,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832925800164718e-05,
+      "loss": 0.7619,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325836841855995e-05,
+      "loss": 0.8445,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322412502874373e-05,
+      "loss": 0.7922,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318984986009883e-05,
+      "loss": 0.8363,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315554292571308e-05,
+      "loss": 0.8388,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831212042386865e-05,
+      "loss": 0.8827,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308683381213117e-05,
+      "loss": 0.9567,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830524316591713e-05,
+      "loss": 0.7721,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301799779294322e-05,
+      "loss": 0.9104,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829835322265954e-05,
+      "loss": 0.7533,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829490349732884e-05,
+      "loss": 0.7433,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291450604619483e-05,
+      "loss": 0.8232,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287994545849948e-05,
+      "loss": 0.9252,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284535322339913e-05,
+      "loss": 0.8678,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281072935410278e-05,
+      "loss": 0.9092,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277607386383134e-05,
+      "loss": 0.8663,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274138676581796e-05,
+      "loss": 0.8536,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827066680733077e-05,
+      "loss": 0.8715,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267191779955784e-05,
+      "loss": 0.8315,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826371359578376e-05,
+      "loss": 0.8592,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260232256142834e-05,
+      "loss": 0.8511,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825674776236234e-05,
+      "loss": 0.8049,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825326011577282e-05,
+      "loss": 0.7981,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249769317706017e-05,
+      "loss": 0.8223,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824627536949488e-05,
+      "loss": 0.7121,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242778272473566e-05,
+      "loss": 0.832,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823927802797742e-05,
+      "loss": 0.8866,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235774637343004e-05,
+      "loss": 0.7545,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232268101908076e-05,
+      "loss": 0.7765,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228758423011587e-05,
+      "loss": 0.8769,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225245601993698e-05,
+      "loss": 0.8837,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822172964019577e-05,
+      "loss": 0.814,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218210538960358e-05,
+      "loss": 0.8036,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214688299631214e-05,
+      "loss": 0.7726,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82111629235533e-05,
+      "loss": 0.8429,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 0.8833,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204102766536956e-05,
+      "loss": 0.726,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820056798829442e-05,
+      "loss": 0.8321,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197030078694898e-05,
+      "loss": 0.9039,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193489039089335e-05,
+      "loss": 0.7633,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818994487082985e-05,
+      "loss": 0.8058,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186397575269786e-05,
+      "loss": 0.7971,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182847153763656e-05,
+      "loss": 0.8688,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179293607667177e-05,
+      "loss": 0.8646,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817573693833726e-05,
+      "loss": 0.7671,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172177147132e-05,
+      "loss": 0.8828,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168614235410703e-05,
+      "loss": 0.8596,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165048204533836e-05,
+      "loss": 0.9525,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816147905586309e-05,
+      "loss": 0.8096,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157906790761327e-05,
+      "loss": 0.7576,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81543314105926e-05,
+      "loss": 0.9461,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150752916722157e-05,
+      "loss": 0.8195,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147171310516433e-05,
+      "loss": 0.794,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814358659334305e-05,
+      "loss": 0.8765,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139998766570825e-05,
+      "loss": 0.8994,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136407831569748e-05,
+      "loss": 0.7804,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132813789711008e-05,
+      "loss": 0.6844,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812921664236698e-05,
+      "loss": 0.8459,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125616390911216e-05,
+      "loss": 0.7516,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122013036718464e-05,
+      "loss": 0.8848,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118406581164646e-05,
+      "loss": 0.9218,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114797025626883e-05,
+      "loss": 0.7625,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811118437148346e-05,
+      "loss": 0.8361,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107568620113863e-05,
+      "loss": 0.7197,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810394977289875e-05,
+      "loss": 0.8361,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100327831219968e-05,
+      "loss": 0.7382,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096702796460538e-05,
+      "loss": 0.836,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093074670004667e-05,
+      "loss": 0.8173,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089443453237746e-05,
+      "loss": 0.9274,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085809147546336e-05,
+      "loss": 0.7962,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808217175431819e-05,
+      "loss": 0.8468,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078531274942227e-05,
+      "loss": 0.683,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074887710808555e-05,
+      "loss": 0.8143,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071241063308455e-05,
+      "loss": 0.7762,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067591333834382e-05,
+      "loss": 0.8536,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806393852377998e-05,
+      "loss": 0.8562,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060282634540053e-05,
+      "loss": 0.8131,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056623667510594e-05,
+      "loss": 0.8903,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805296162408876e-05,
+      "loss": 0.871,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80492965056729e-05,
+      "loss": 0.8093,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045628313662516e-05,
+      "loss": 0.8225,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041957049458298e-05,
+      "loss": 0.8211,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038282714462103e-05,
+      "loss": 0.7937,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034605310076965e-05,
+      "loss": 0.825,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030924837707087e-05,
+      "loss": 0.8809,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802724129875784e-05,
+      "loss": 0.8092,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023554694635778e-05,
+      "loss": 0.8404,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801986502674861e-05,
+      "loss": 0.8845,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801617229650522e-05,
+      "loss": 0.8374,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012476505315674e-05,
+      "loss": 0.8387,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008777654591188e-05,
+      "loss": 0.7984,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005075745744158e-05,
+      "loss": 0.8396,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800137078018814e-05,
+      "loss": 0.8878,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799766275933787e-05,
+      "loss": 0.7886,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993951684609233e-05,
+      "loss": 0.8324,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990237557419298e-05,
+      "loss": 0.7871,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986520379186285e-05,
+      "loss": 0.909,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798280015132959e-05,
+      "loss": 0.8926,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797907687526976e-05,
+      "loss": 0.6844,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975350552428516e-05,
+      "loss": 0.9211,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971621184228748e-05,
+      "loss": 0.8772,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79678887720945e-05,
+      "loss": 0.8652,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796415331745098e-05,
+      "loss": 0.8521,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796041482172455e-05,
+      "loss": 0.8425,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795667328634275e-05,
+      "loss": 0.7576,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952928712734266e-05,
+      "loss": 0.8364,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949181102328955e-05,
+      "loss": 0.8814,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794543045655782e-05,
+      "loss": 0.8088,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794167677685304e-05,
+      "loss": 0.8018,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793792006464794e-05,
+      "loss": 0.8215,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934160321377008e-05,
+      "loss": 0.8043,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793039754847589e-05,
+      "loss": 0.7793,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792663174738138e-05,
+      "loss": 0.7509,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922862919531443e-05,
+      "loss": 0.8225,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919091066365192e-05,
+      "loss": 0.8691,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791531618932289e-05,
+      "loss": 0.8192,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911538289845964e-05,
+      "loss": 0.8628,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907757369376984e-05,
+      "loss": 0.8293,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903973429359696e-05,
+      "loss": 0.8338,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790018647123897e-05,
+      "loss": 0.9182,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896396496460846e-05,
+      "loss": 0.7881,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892603506472516e-05,
+      "loss": 0.8551,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788880750272231e-05,
+      "loss": 0.7192,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885008486659728e-05,
+      "loss": 0.8832,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881206459735407e-05,
+      "loss": 0.8218,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877401423401134e-05,
+      "loss": 0.7108,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873593379109855e-05,
+      "loss": 0.8898,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786978232831565e-05,
+      "loss": 0.9029,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786596827247376e-05,
+      "loss": 0.823,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862151213040566e-05,
+      "loss": 0.8327,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858331151473604e-05,
+      "loss": 0.9486,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854508089231538e-05,
+      "loss": 0.9144,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850682027774203e-05,
+      "loss": 0.8723,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846852968562564e-05,
+      "loss": 0.8369,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843020913058725e-05,
+      "loss": 0.8228,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839185862725953e-05,
+      "loss": 0.8723,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835347819028642e-05,
+      "loss": 0.8069,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831506783432337e-05,
+      "loss": 0.7645,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827662757403724e-05,
+      "loss": 0.7797,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823815742410627e-05,
+      "loss": 0.8455,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781996573992202e-05,
+      "loss": 0.8324,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816112751408003e-05,
+      "loss": 0.9228,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812256778339837e-05,
+      "loss": 0.8726,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808397822189903e-05,
+      "loss": 0.8015,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804535884431737e-05,
+      "loss": 0.7986,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800670966539997e-05,
+      "loss": 0.857,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796803069990492e-05,
+      "loss": 0.8288,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792932196260168e-05,
+      "loss": 0.8791,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77890583468271e-05,
+      "loss": 0.9837,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77851815231705e-05,
+      "loss": 0.7412,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781301726770726e-05,
+      "loss": 0.831,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777741895910926e-05,
+      "loss": 0.7719,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773533221668724e-05,
+      "loss": 0.7881,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776964451593287e-05,
+      "loss": 0.7856,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776575284338659e-05,
+      "loss": 0.8624,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761858205515904e-05,
+      "loss": 0.792,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775796060380796e-05,
+      "loss": 0.8991,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775406003975105e-05,
+      "loss": 0.8739,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775015651483459e-05,
+      "loss": 0.8113,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774625003054912e-05,
+      "loss": 0.9251,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742340588386325e-05,
+      "loss": 0.8451,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738428189839006e-05,
+      "loss": 0.8015,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77345128364011e-05,
+      "loss": 0.7469,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730594529567668e-05,
+      "loss": 0.705,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726673270834906e-05,
+      "loss": 0.8929,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722749061700122e-05,
+      "loss": 0.9007,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718821903661777e-05,
+      "loss": 0.7832,
+      "step": 1273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714891798219432e-05,
+      "loss": 0.8741,
+      "step": 1274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771095874687378e-05,
+      "loss": 0.8402,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770702275112665e-05,
+      "loss": 0.7225,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703083812480986e-05,
+      "loss": 0.906,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699141932440855e-05,
+      "loss": 0.7299,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695197112511452e-05,
+      "loss": 0.8183,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769124935419909e-05,
+      "loss": 0.8729,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687298659011207e-05,
+      "loss": 0.8708,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683345028456357e-05,
+      "loss": 0.8935,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679388464044228e-05,
+      "loss": 0.8869,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767542896728562e-05,
+      "loss": 0.7866,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767146653969244e-05,
+      "loss": 0.7973,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667501182777736e-05,
+      "loss": 0.9235,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663532898055663e-05,
+      "loss": 0.9386,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76595616870415e-05,
+      "loss": 0.829,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655587551251633e-05,
+      "loss": 0.78,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651610492203578e-05,
+      "loss": 0.7662,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647630511415955e-05,
+      "loss": 0.7297,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643647610408507e-05,
+      "loss": 0.7546,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639661790702088e-05,
+      "loss": 0.8241,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635673053818672e-05,
+      "loss": 0.909,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763168140128134e-05,
+      "loss": 0.8656,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627686834614298e-05,
+      "loss": 0.845,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762368935534285e-05,
+      "loss": 0.8269,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619688964993416e-05,
+      "loss": 0.7912,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615685665093537e-05,
+      "loss": 0.7652,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611679457171856e-05,
+      "loss": 0.8213,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760767034275813e-05,
+      "loss": 0.8826,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760365832338322e-05,
+      "loss": 0.8137,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599643400579104e-05,
+      "loss": 0.7462,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595625575878873e-05,
+      "loss": 0.856,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591604850816705e-05,
+      "loss": 0.9284,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 0.8455,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583554705748888e-05,
+      "loss": 0.8374,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757952528881716e-05,
+      "loss": 0.8286,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757549297767133e-05,
+      "loss": 0.8802,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571457773851132e-05,
+      "loss": 0.8509,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756741967889739e-05,
+      "loss": 0.7982,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563378694352038e-05,
+      "loss": 0.812,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755933482175811e-05,
+      "loss": 0.792,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755528806265974e-05,
+      "loss": 0.6604,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551238418602178e-05,
+      "loss": 0.7767,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754718589113176e-05,
+      "loss": 0.7456,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543130481795925e-05,
+      "loss": 0.8428,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539072192143223e-05,
+      "loss": 0.8962,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535011023723293e-05,
+      "loss": 0.8489,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530946978086882e-05,
+      "loss": 0.7806,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526880056785822e-05,
+      "loss": 0.8391,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752281026137306e-05,
+      "loss": 0.7804,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518737593402633e-05,
+      "loss": 0.8081,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751466205442967e-05,
+      "loss": 0.8307,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75105836460104e-05,
+      "loss": 0.7216,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506502369702154e-05,
+      "loss": 0.8167,
+      "step": 1326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502418227063346e-05,
+      "loss": 0.9146,
+      "step": 1327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498331219653498e-05,
+      "loss": 0.8305,
+      "step": 1328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494241349033212e-05,
+      "loss": 0.8415,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490148616764196e-05,
+      "loss": 0.8231,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748605302440924e-05,
+      "loss": 0.7948,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481954573532233e-05,
+      "loss": 0.9171,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747785326569815e-05,
+      "loss": 0.8493,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473749102473073e-05,
+      "loss": 0.8306,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469642085424148e-05,
+      "loss": 0.7872,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465532216119628e-05,
+      "loss": 0.7293,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7461419496128855e-05,
+      "loss": 0.7778,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457303927022254e-05,
+      "loss": 0.9617,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745318551037134e-05,
+      "loss": 0.8274,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449064247748714e-05,
+      "loss": 0.8559,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744494014072807e-05,
+      "loss": 0.7965,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440813190884177e-05,
+      "loss": 0.7543,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74366833997929e-05,
+      "loss": 0.7725,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432550769031184e-05,
+      "loss": 0.8668,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428415300177053e-05,
+      "loss": 0.821,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424276994809632e-05,
+      "loss": 0.7907,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742013585450911e-05,
+      "loss": 0.7986,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415991880856768e-05,
+      "loss": 0.7892,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411845075434967e-05,
+      "loss": 0.8297,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740769543982715e-05,
+      "loss": 0.8392,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403542975617843e-05,
+      "loss": 0.8466,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399387684392643e-05,
+      "loss": 0.8659,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395229567738234e-05,
+      "loss": 0.7644,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391068627242384e-05,
+      "loss": 0.8268,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386904864493927e-05,
+      "loss": 0.804,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382738281082784e-05,
+      "loss": 0.7998,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737856887859995e-05,
+      "loss": 0.7654,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737439665863749e-05,
+      "loss": 0.7316,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370221622788556e-05,
+      "loss": 0.7162,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736604377264737e-05,
+      "loss": 0.8752,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736186310980923e-05,
+      "loss": 0.8705,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357679635870504e-05,
+      "loss": 0.8188,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353493352428637e-05,
+      "loss": 0.8358,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734930426108215e-05,
+      "loss": 0.7738,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345112363430627e-05,
+      "loss": 0.9006,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340917661074726e-05,
+      "loss": 0.7528,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336720155616186e-05,
+      "loss": 0.7474,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332519848657806e-05,
+      "loss": 0.9279,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328316741803464e-05,
+      "loss": 0.9286,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732411083665809e-05,
+      "loss": 0.7303,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319902134827698e-05,
+      "loss": 0.7993,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731569063791937e-05,
+      "loss": 0.8652,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731147634754125e-05,
+      "loss": 0.7792,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730725926530254e-05,
+      "loss": 0.8186,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730303939281353e-05,
+      "loss": 0.793,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298816731685554e-05,
+      "loss": 0.8251,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294591283531022e-05,
+      "loss": 0.7905,
+      "step": 1377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729036304996341e-05,
+      "loss": 0.8706,
+      "step": 1378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286132032597244e-05,
+      "loss": 0.8206,
+      "step": 1379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281898233048135e-05,
+      "loss": 0.8416,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277661652932734e-05,
+      "loss": 0.7536,
+      "step": 1381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727342229386877e-05,
+      "loss": 0.7664,
+      "step": 1382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269180157475025e-05,
+      "loss": 0.7945,
+      "step": 1383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726493524537134e-05,
+      "loss": 0.9371,
+      "step": 1384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260687559178622e-05,
+      "loss": 0.7378,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256437100518837e-05,
+      "loss": 0.7737,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252183871015002e-05,
+      "loss": 0.8925,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72479278722912e-05,
+      "loss": 0.7914,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724366910597257e-05,
+      "loss": 0.7084,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239407573685304e-05,
+      "loss": 0.8096,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235143277056652e-05,
+      "loss": 0.858,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723087621771492e-05,
+      "loss": 0.893,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226606397289472e-05,
+      "loss": 0.9389,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222333817410724e-05,
+      "loss": 0.7564,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721805847971014e-05,
+      "loss": 0.7294,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213780385820244e-05,
+      "loss": 0.8197,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720949953737461e-05,
+      "loss": 0.8436,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720521593600787e-05,
+      "loss": 0.7682,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72009295833557e-05,
+      "loss": 0.8119,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719664048105482e-05,
+      "loss": 0.8052,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192348630743012e-05,
+      "loss": 0.819,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718805403405911e-05,
+      "loss": 0.7911,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183756692642984e-05,
+      "loss": 0.8261,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179456608135555e-05,
+      "loss": 0.761,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175153782178805e-05,
+      "loss": 0.947,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170848216415742e-05,
+      "loss": 0.7497,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166539912490433e-05,
+      "loss": 0.8426,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716222887204799e-05,
+      "loss": 0.7628,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157915096734573e-05,
+      "loss": 0.7023,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153598588197374e-05,
+      "loss": 0.7817,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714927934808464e-05,
+      "loss": 0.8218,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144957378045656e-05,
+      "loss": 0.7843,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140632679730744e-05,
+      "loss": 0.835,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136305254791286e-05,
+      "loss": 0.8287,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131975104879687e-05,
+      "loss": 0.8856,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127642231649396e-05,
+      "loss": 0.8091,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712330663675491e-05,
+      "loss": 0.9192,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711896832185176e-05,
+      "loss": 0.8468,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114627288596513e-05,
+      "loss": 0.805,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711028353864678e-05,
+      "loss": 0.7549,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105937073661203e-05,
+      "loss": 0.8324,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101587895299463e-05,
+      "loss": 0.9099,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709723600522228e-05,
+      "loss": 0.921,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709288140509141e-05,
+      "loss": 0.7686,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708852409656964e-05,
+      "loss": 0.7265,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084164081320787e-05,
+      "loss": 0.7282,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079801361009713e-05,
+      "loss": 0.7304,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70754359373023e-05,
+      "loss": 0.8092,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.784,
+      "step": 1429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066696986367193e-05,
+      "loss": 0.8075,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062323462476432e-05,
+      "loss": 0.7419,
+      "step": 1431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057947241863207e-05,
+      "loss": 0.7999,
+      "step": 1432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053568326198566e-05,
+      "loss": 0.7196,
+      "step": 1433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049186717154578e-05,
+      "loss": 0.8618,
+      "step": 1434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044802416404347e-05,
+      "loss": 0.8382,
+      "step": 1435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040415425622e-05,
+      "loss": 0.9701,
+      "step": 1436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036025746482692e-05,
+      "loss": 0.8594,
+      "step": 1437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703163338066261e-05,
+      "loss": 0.8336,
+      "step": 1438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702723832983896e-05,
+      "loss": 0.8232,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702284059568998e-05,
+      "loss": 0.8865,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018440179894923e-05,
+      "loss": 0.7794,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7014037084134076e-05,
+      "loss": 0.8097,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009631310088745e-05,
+      "loss": 0.7877,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005222859441256e-05,
+      "loss": 0.7676,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700081173387496e-05,
+      "loss": 0.9246,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996397935074237e-05,
+      "loss": 0.6857,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991981464724468e-05,
+      "loss": 0.8224,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987562324512074e-05,
+      "loss": 0.8077,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983140516124486e-05,
+      "loss": 0.9359,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697871604125015e-05,
+      "loss": 0.8045,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697428890157854e-05,
+      "loss": 0.8921,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696985909880015e-05,
+      "loss": 0.8316,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965426634606476e-05,
+      "loss": 0.8212,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960991510690036e-05,
+      "loss": 0.8125,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695655372874437e-05,
+      "loss": 0.8553,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952113290464032e-05,
+      "loss": 0.8139,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694767019754458e-05,
+      "loss": 0.9618,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69432244516826e-05,
+      "loss": 0.7125,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938776054575673e-05,
+      "loss": 0.9604,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934325007922418e-05,
+      "loss": 0.7845,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929871313422444e-05,
+      "loss": 0.849,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692541497277637e-05,
+      "loss": 0.7332,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920955987685848e-05,
+      "loss": 0.794,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916494359853515e-05,
+      "loss": 0.7561,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912030090983033e-05,
+      "loss": 0.7868,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907563182779064e-05,
+      "loss": 0.9641,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690309363694728e-05,
+      "loss": 0.8285,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898621455194356e-05,
+      "loss": 0.7895,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689414663922799e-05,
+      "loss": 0.7782,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 0.8086,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885189111490687e-05,
+      "loss": 0.7041,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880706403140146e-05,
+      "loss": 0.7674,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687622106741696e-05,
+      "loss": 0.7904,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871733106033827e-05,
+      "loss": 0.8093,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867242520704466e-05,
+      "loss": 0.8275,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686274931314359e-05,
+      "loss": 0.8211,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858253485066908e-05,
+      "loss": 0.7514,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853755038191145e-05,
+      "loss": 0.7018,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684925397423401e-05,
+      "loss": 0.7876,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844750294914217e-05,
+      "loss": 0.8968,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840244001951488e-05,
+      "loss": 0.8131,
+      "step": 1481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835735097066524e-05,
+      "loss": 0.7957,
+      "step": 1482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683122358198104e-05,
+      "loss": 0.804,
+      "step": 1483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682670945841774e-05,
+      "loss": 0.7468,
+      "step": 1484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822192728100324e-05,
+      "loss": 0.869,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681767339275349e-05,
+      "loss": 0.818,
+      "step": 1486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813151454102935e-05,
+      "loss": 0.9284,
+      "step": 1487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808626913875334e-05,
+      "loss": 0.7641,
+      "step": 1488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804099773798378e-05,
+      "loss": 0.8087,
+      "step": 1489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679957003560073e-05,
+      "loss": 0.8401,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679503770101206e-05,
+      "loss": 0.8975,
+      "step": 1491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790502771763018e-05,
+      "loss": 0.9076,
+      "step": 1492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785965249585254e-05,
+      "loss": 0.91,
+      "step": 1493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781425136211402e-05,
+      "loss": 0.8078,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677688243337509e-05,
+      "loss": 0.9268,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772337142810934e-05,
+      "loss": 0.8516,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676778926625453e-05,
+      "loss": 0.7684,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763238805442474e-05,
+      "loss": 0.7446,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675868576211234e-05,
+      "loss": 0.7348,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754130138002692e-05,
+      "loss": 0.7197,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674957193485308e-05,
+      "loss": 0.7734,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745011154404037e-05,
+      "loss": 0.6945,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740447798397076e-05,
+      "loss": 0.7367,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67358818685747e-05,
+      "loss": 0.82,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67313133666804e-05,
+      "loss": 0.8634,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726742294458633e-05,
+      "loss": 0.7951,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722168653654844e-05,
+      "loss": 0.7659,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671759244601547e-05,
+      "loss": 0.6595,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713013673287912e-05,
+      "loss": 0.8227,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708432337220562e-05,
+      "loss": 0.9036,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703848439562787e-05,
+      "loss": 0.8338,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669926198206493e-05,
+      "loss": 0.7907,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669467296647831e-05,
+      "loss": 0.7233,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690081394555227e-05,
+      "loss": 0.8555,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685487268048962e-05,
+      "loss": 0.8383,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668089058871376e-05,
+      "loss": 0.9202,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676291358304847e-05,
+      "loss": 0.7848,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671689578578423e-05,
+      "loss": 0.8341,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666708525129166e-05,
+      "loss": 0.6958,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662478378202703e-05,
+      "loss": 0.9157,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665786896107067e-05,
+      "loss": 0.706,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653257001655652e-05,
+      "loss": 0.8445,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648642501718713e-05,
+      "loss": 0.813,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664402546302187e-05,
+      "loss": 0.6906,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639405887328136e-05,
+      "loss": 0.8738,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634783776401467e-05,
+      "loss": 0.7343,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630159132006812e-05,
+      "loss": 0.8453,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625531955910064e-05,
+      "loss": 0.8373,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620902249878105e-05,
+      "loss": 0.8182,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616270015678756e-05,
+      "loss": 0.8417,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611635255080826e-05,
+      "loss": 0.8259,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606997969854087e-05,
+      "loss": 0.8272,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660235816176926e-05,
+      "loss": 0.8001,
+      "step": 1533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659771583259805e-05,
+      "loss": 0.7669,
+      "step": 1534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593070984113097e-05,
+      "loss": 0.7764,
+      "step": 1535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588423618088034e-05,
+      "loss": 0.8109,
+      "step": 1536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583773736297435e-05,
+      "loss": 0.8408,
+      "step": 1537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579121340516843e-05,
+      "loss": 0.8469,
+      "step": 1538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574466432522754e-05,
+      "loss": 0.7919,
+      "step": 1539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569809014092632e-05,
+      "loss": 0.8268,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565149087004892e-05,
+      "loss": 0.7396,
+      "step": 1541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560486653038916e-05,
+      "loss": 0.8192,
+      "step": 1542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655582171397503e-05,
+      "loss": 0.7054,
+      "step": 1543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551154271594527e-05,
+      "loss": 0.7433,
+      "step": 1544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546484327679654e-05,
+      "loss": 0.7613,
+      "step": 1545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654181188401361e-05,
+      "loss": 0.7953,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537136942380552e-05,
+      "loss": 0.7742,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653245950456559e-05,
+      "loss": 0.8367,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527779572354786e-05,
+      "loss": 0.7179,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652309714753515e-05,
+      "loss": 0.8449,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518412231894657e-05,
+      "loss": 0.7501,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513724827222225e-05,
+      "loss": 0.8445,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509034935307716e-05,
+      "loss": 0.8532,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650434255794195e-05,
+      "loss": 0.8527,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499647696916695e-05,
+      "loss": 0.8362,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494950354024664e-05,
+      "loss": 0.7505,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490250531059527e-05,
+      "loss": 0.7736,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648554822981589e-05,
+      "loss": 0.8982,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480843452089308e-05,
+      "loss": 0.8565,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476136199676286e-05,
+      "loss": 0.7599,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647142647437427e-05,
+      "loss": 0.7489,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466714277981656e-05,
+      "loss": 0.7592,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646199961229777e-05,
+      "loss": 0.8151,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457282479122892e-05,
+      "loss": 0.8804,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645256288025825e-05,
+      "loss": 0.8861,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447840817505997e-05,
+      "loss": 0.7637,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644311629266924e-05,
+      "loss": 0.7908,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643838930755202e-05,
+      "loss": 0.8711,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433659863959318e-05,
+      "loss": 0.7519,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642892796369706e-05,
+      "loss": 0.8121,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424193608572097e-05,
+      "loss": 0.7849,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641945680039223e-05,
+      "loss": 0.8564,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641471754096619e-05,
+      "loss": 0.8075,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640997583210365e-05,
+      "loss": 0.7733,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405231675615212e-05,
+      "loss": 0.7797,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400485073312415e-05,
+      "loss": 0.8275,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395736027007725e-05,
+      "loss": 0.8175,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390984538514558e-05,
+      "loss": 0.7828,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638623060964725e-05,
+      "loss": 0.87,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381474242221066e-05,
+      "loss": 0.8213,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376715438052213e-05,
+      "loss": 0.8935,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371954198957823e-05,
+      "loss": 0.8534,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367190526755954e-05,
+      "loss": 0.8773,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63624244232656e-05,
+      "loss": 0.8053,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635765589030668e-05,
+      "loss": 0.7708,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635288492970004e-05,
+      "loss": 0.8765,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348111543267454e-05,
+      "loss": 0.8142,
+      "step": 1587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634333573283162e-05,
+      "loss": 0.8063,
+      "step": 1588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633855750021617e-05,
+      "loss": 0.9595,
+      "step": 1589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633377684724565e-05,
+      "loss": 0.7415,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328993775745536e-05,
+      "loss": 0.8805,
+      "step": 1591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324208287542228e-05,
+      "loss": 0.829,
+      "step": 1592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631942038446304e-05,
+      "loss": 0.7957,
+      "step": 1593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314630068336226e-05,
+      "loss": 0.8049,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630983734099094e-05,
+      "loss": 0.7652,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630504220425728e-05,
+      "loss": 0.7547,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300244659966243e-05,
+      "loss": 0.8318,
+      "step": 1597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295444709949752e-05,
+      "loss": 0.8575,
+      "step": 1598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6290642356040654e-05,
+      "loss": 0.8672,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285837600072713e-05,
+      "loss": 0.7889,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281030443880603e-05,
+      "loss": 0.7842,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276220889299918e-05,
+      "loss": 0.7802,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271408938167174e-05,
+      "loss": 0.7559,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266594592319794e-05,
+      "loss": 0.876,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261777853596122e-05,
+      "loss": 0.9162,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256958723835406e-05,
+      "loss": 0.9077,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252137204877816e-05,
+      "loss": 0.8521,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247313298564435e-05,
+      "loss": 0.7865,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242487006737246e-05,
+      "loss": 0.8154,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237658331239156e-05,
+      "loss": 0.8175,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232827273913977e-05,
+      "loss": 0.8886,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622799383660643e-05,
+      "loss": 0.8282,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622315802116215e-05,
+      "loss": 0.8022,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218319829427668e-05,
+      "loss": 0.7877,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213479263250433e-05,
+      "loss": 0.7372,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208636324478797e-05,
+      "loss": 0.8396,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620379101496202e-05,
+      "loss": 0.7978,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198943336550265e-05,
+      "loss": 0.8273,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61940932910946e-05,
+      "loss": 0.7891,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189240880446998e-05,
+      "loss": 0.7276,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184386106460334e-05,
+      "loss": 0.8604,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617952897098839e-05,
+      "loss": 0.8272,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174669475885832e-05,
+      "loss": 0.7154,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169807623008252e-05,
+      "loss": 0.7404,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616494341421213e-05,
+      "loss": 0.842,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160076851354844e-05,
+      "loss": 0.8579,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155207936294675e-05,
+      "loss": 0.8686,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150336670890798e-05,
+      "loss": 0.8342,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614546305700329e-05,
+      "loss": 0.7128,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6140587096493123e-05,
+      "loss": 0.8433,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135708791222164e-05,
+      "loss": 0.7377,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130828143053173e-05,
+      "loss": 0.8734,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612594515384981e-05,
+      "loss": 0.907,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 0.7505,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116172159799074e-05,
+      "loss": 0.7874,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611128215868348e-05,
+      "loss": 0.7929,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610638982399708e-05,
+      "loss": 0.9394,
+      "step": 1637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101495157607988e-05,
+      "loss": 0.7583,
+      "step": 1638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609659816138522e-05,
+      "loss": 0.7761,
+      "step": 1639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6091698837198673e-05,
+      "loss": 0.7936,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086797186919136e-05,
+      "loss": 0.7694,
+      "step": 1641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081893212418292e-05,
+      "loss": 0.7668,
+      "step": 1642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6076986915568694e-05,
+      "loss": 0.8306,
+      "step": 1643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607207829824381e-05,
+      "loss": 0.8381,
+      "step": 1644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606716736231796e-05,
+      "loss": 0.8059,
+      "step": 1645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062254109666383e-05,
+      "loss": 0.9713,
+      "step": 1646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057338542165176e-05,
+      "loss": 0.8365,
+      "step": 1647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605242066169133e-05,
+      "loss": 0.8607,
+      "step": 1648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604750047012273e-05,
+      "loss": 0.8286,
+      "step": 1649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042577969338123e-05,
+      "loss": 0.7928,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037653161217152e-05,
+      "loss": 0.8405,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032726047640336e-05,
+      "loss": 0.7128,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027796630489074e-05,
+      "loss": 0.8683,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022864911645654e-05,
+      "loss": 0.7386,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017930892993226e-05,
+      "loss": 0.9012,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012994576415825e-05,
+      "loss": 0.7034,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008055963798374e-05,
+      "loss": 0.783,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003115057026663e-05,
+      "loss": 0.772,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998171857987356e-05,
+      "loss": 0.7382,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993226368567992e-05,
+      "loss": 0.8504,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988278590656996e-05,
+      "loss": 0.792,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983328526143653e-05,
+      "loss": 0.7607,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978376176918132e-05,
+      "loss": 0.7455,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973421544871466e-05,
+      "loss": 0.8463,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968464631895563e-05,
+      "loss": 0.801,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963505439883204e-05,
+      "loss": 0.8988,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595854397072804e-05,
+      "loss": 0.7658,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953580226324582e-05,
+      "loss": 0.7917,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948614208568226e-05,
+      "loss": 0.8562,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943645919355226e-05,
+      "loss": 0.7494,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938675360582704e-05,
+      "loss": 0.7778,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933702534148648e-05,
+      "loss": 0.8589,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928727441951918e-05,
+      "loss": 0.7775,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592375008589223e-05,
+      "loss": 0.7682,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5918770467870174e-05,
+      "loss": 0.7848,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5913788589787195e-05,
+      "loss": 0.8126,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908804453545608e-05,
+      "loss": 0.7948,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903818061048583e-05,
+      "loss": 0.8246,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898829414200162e-05,
+      "loss": 0.8054,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893838514905234e-05,
+      "loss": 0.8219,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888845365069566e-05,
+      "loss": 0.8283,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588384996659976e-05,
+      "loss": 0.9153,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878852321403304e-05,
+      "loss": 0.7613,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873852431388526e-05,
+      "loss": 0.8127,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586885029846461e-05,
+      "loss": 0.8159,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586384592454161e-05,
+      "loss": 0.803,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585883931153043e-05,
+      "loss": 0.8287,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853830461342816e-05,
+      "loss": 0.8543,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848819375891387e-05,
+      "loss": 0.8727,
+      "step": 1689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843806057089603e-05,
+      "loss": 0.7673,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838790506851787e-05,
+      "loss": 0.8501,
+      "step": 1691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583377272709311e-05,
+      "loss": 0.9134,
+      "step": 1692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828752719729583e-05,
+      "loss": 0.8162,
+      "step": 1693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823730486678085e-05,
+      "loss": 0.8015,
+      "step": 1694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818706029856337e-05,
+      "loss": 0.8405,
+      "step": 1695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581367935118291e-05,
+      "loss": 0.7905,
+      "step": 1696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580865045257722e-05,
+      "loss": 0.7202,
+      "step": 1697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580361933595953e-05,
+      "loss": 0.7666,
+      "step": 1698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798586003250963e-05,
+      "loss": 0.8217,
+      "step": 1699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5793550456373475e-05,
+      "loss": 0.7694,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788512697249867e-05,
+      "loss": 0.8906,
+      "step": 1701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783472727803796e-05,
+      "loss": 0.8366,
+      "step": 1702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778430549959753e-05,
+      "loss": 0.7109,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577338616564307e-05,
+      "loss": 0.8159,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5768339576779933e-05,
+      "loss": 0.8241,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576329078529736e-05,
+      "loss": 0.8535,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758239793123218e-05,
+      "loss": 0.7304,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753186602186207e-05,
+      "loss": 0.8123,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574813121441587e-05,
+      "loss": 0.7828,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743073631742588e-05,
+      "loss": 0.7374,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738013856097582e-05,
+      "loss": 0.762,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732951889412905e-05,
+      "loss": 0.7569,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727887733621456e-05,
+      "loss": 0.7882,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572282139065696e-05,
+      "loss": 0.7237,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717752862453984e-05,
+      "loss": 0.7183,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712682150947926e-05,
+      "loss": 0.7619,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707609258075018e-05,
+      "loss": 0.756,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702534185772327e-05,
+      "loss": 0.8764,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569745693597775e-05,
+      "loss": 0.7058,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692377510630017e-05,
+      "loss": 0.8403,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687295911668687e-05,
+      "loss": 0.9072,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682212141034153e-05,
+      "loss": 0.8348,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677126200667632e-05,
+      "loss": 0.821,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567203809251117e-05,
+      "loss": 0.8111,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666947818507645e-05,
+      "loss": 0.8039,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566185538060076e-05,
+      "loss": 0.7954,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565676078073504e-05,
+      "loss": 0.8168,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651664020855845e-05,
+      "loss": 0.9254,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564656510290935e-05,
+      "loss": 0.6974,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564146402884256e-05,
+      "loss": 0.7963,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636360800603303e-05,
+      "loss": 0.7784,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5631255420140225e-05,
+      "loss": 0.8152,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626147889402797e-05,
+      "loss": 0.8649,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621038210341312e-05,
+      "loss": 0.8212,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615926384906884e-05,
+      "loss": 0.7866,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610812415051442e-05,
+      "loss": 0.7373,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605696302727743e-05,
+      "loss": 0.8316,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600578049889344e-05,
+      "loss": 0.7087,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595457658490643e-05,
+      "loss": 0.8597,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590335130486836e-05,
+      "loss": 0.8266,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585210467833944e-05,
+      "loss": 0.7953,
+      "step": 1741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55800836724888e-05,
+      "loss": 0.681,
+      "step": 1742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557495474640905e-05,
+      "loss": 0.806,
+      "step": 1743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569823691553163e-05,
+      "loss": 0.8431,
+      "step": 1744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564690509880403e-05,
+      "loss": 0.7746,
+      "step": 1745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555955520335087e-05,
+      "loss": 0.7352,
+      "step": 1746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554417773925447e-05,
+      "loss": 0.7787,
+      "step": 1747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549278223565853e-05,
+      "loss": 0.7714,
+      "step": 1748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544136554234604e-05,
+      "loss": 0.8479,
+      "step": 1749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538992767895028e-05,
+      "loss": 0.7686,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553384686651126e-05,
+      "loss": 0.8525,
+      "step": 1751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528698852048247e-05,
+      "loss": 0.8176,
+      "step": 1752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523548726471732e-05,
+      "loss": 0.734,
+      "step": 1753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518396491748277e-05,
+      "loss": 0.8037,
+      "step": 1754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513242149845245e-05,
+      "loss": 0.8989,
+      "step": 1755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55080857027308e-05,
+      "loss": 0.7176,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502927152373913e-05,
+      "loss": 0.8128,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497766500744363e-05,
+      "loss": 0.751,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492603749812718e-05,
+      "loss": 0.8686,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487438901550357e-05,
+      "loss": 0.7945,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482271957929462e-05,
+      "loss": 0.8815,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547710292092301e-05,
+      "loss": 0.7599,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5471931792504783e-05,
+      "loss": 0.7945,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466758574649348e-05,
+      "loss": 0.8788,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461583269332087e-05,
+      "loss": 0.7328,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545640587852917e-05,
+      "loss": 0.769,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451226404217563e-05,
+      "loss": 0.8219,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446044848375028e-05,
+      "loss": 0.8155,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440861212980126e-05,
+      "loss": 0.8127,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435675500012212e-05,
+      "loss": 0.8002,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430487711451428e-05,
+      "loss": 0.8166,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425297849278714e-05,
+      "loss": 0.7083,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420105915475793e-05,
+      "loss": 0.7825,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54149119120252e-05,
+      "loss": 0.7968,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540971584091023e-05,
+      "loss": 0.8637,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404517704114997e-05,
+      "loss": 0.8336,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399317503624384e-05,
+      "loss": 0.8714,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394115241424068e-05,
+      "loss": 0.869,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388910919500514e-05,
+      "loss": 0.756,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538370453984098e-05,
+      "loss": 0.8259,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378496104433495e-05,
+      "loss": 0.8419,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373285615266884e-05,
+      "loss": 0.9121,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368073074330755e-05,
+      "loss": 0.8023,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362858483615498e-05,
+      "loss": 0.8155,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357641845112282e-05,
+      "loss": 0.8229,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352423160813063e-05,
+      "loss": 0.7789,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347202432710576e-05,
+      "loss": 0.7212,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341979662798337e-05,
+      "loss": 0.813,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336754853070645e-05,
+      "loss": 0.717,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533152800552257e-05,
+      "loss": 0.7939,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326299122149967e-05,
+      "loss": 0.6964,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321068204949465e-05,
+      "loss": 0.8863,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531583525591847e-05,
+      "loss": 0.8027,
+      "step": 1793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531060027705517e-05,
+      "loss": 0.8421,
+      "step": 1794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305363270358517e-05,
+      "loss": 0.7688,
+      "step": 1795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530012423782824e-05,
+      "loss": 0.7984,
+      "step": 1796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529488318146485e-05,
+      "loss": 0.7437,
+      "step": 1797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 0.8043,
+      "step": 1798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284395005244616e-05,
+      "loss": 0.848,
+      "step": 1799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279147889392644e-05,
+      "loss": 0.7291,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273898757717295e-05,
+      "loss": 0.8409,
+      "step": 1801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526864761222294e-05,
+      "loss": 0.8843,
+      "step": 1802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52633944549147e-05,
+      "loss": 0.9019,
+      "step": 1803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258139287798486e-05,
+      "loss": 0.7721,
+      "step": 1804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252882112880954e-05,
+      "loss": 0.8424,
+      "step": 1805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247622932169543e-05,
+      "loss": 0.8441,
+      "step": 1806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242361747672447e-05,
+      "loss": 0.741,
+      "step": 1807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237098561398638e-05,
+      "loss": 0.8443,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231833375357833e-05,
+      "loss": 0.8132,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226566191560532e-05,
+      "loss": 0.7324,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221297012017994e-05,
+      "loss": 0.8355,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216025838742226e-05,
+      "loss": 0.7424,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521075267374601e-05,
+      "loss": 0.783,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205477519042887e-05,
+      "loss": 0.8,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200200376647155e-05,
+      "loss": 0.7836,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194921248573868e-05,
+      "loss": 0.8921,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189640136838847e-05,
+      "loss": 0.7294,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184357043458664e-05,
+      "loss": 0.7094,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179071970450644e-05,
+      "loss": 0.8594,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517378491983288e-05,
+      "loss": 0.825,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168495893624208e-05,
+      "loss": 0.7991,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163204893844223e-05,
+      "loss": 0.7596,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157911922513279e-05,
+      "loss": 0.8256,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152616981652475e-05,
+      "loss": 0.836,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147320073283664e-05,
+      "loss": 0.8143,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142021199429451e-05,
+      "loss": 0.8787,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136720362113193e-05,
+      "loss": 0.7269,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131417563358995e-05,
+      "loss": 0.9077,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126112805191712e-05,
+      "loss": 0.6605,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120806089636944e-05,
+      "loss": 0.7368,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115497418721044e-05,
+      "loss": 0.7756,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110186794471105e-05,
+      "loss": 0.7401,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104874218914973e-05,
+      "loss": 0.7526,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099559694081236e-05,
+      "loss": 0.8302,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094243221999226e-05,
+      "loss": 0.7938,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088924804699013e-05,
+      "loss": 0.7668,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508360444421142e-05,
+      "loss": 0.7684,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078282142568008e-05,
+      "loss": 0.7802,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072957901801075e-05,
+      "loss": 0.797,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067631723943665e-05,
+      "loss": 0.821,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062303611029561e-05,
+      "loss": 0.8783,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505697356509328e-05,
+      "loss": 0.7439,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051641588170082e-05,
+      "loss": 0.734,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046307682295969e-05,
+      "loss": 0.7533,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040971849507662e-05,
+      "loss": 0.7561,
+      "step": 1845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035634091842637e-05,
+      "loss": 0.6854,
+      "step": 1846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030294411339097e-05,
+      "loss": 0.805,
+      "step": 1847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024952810035976e-05,
+      "loss": 0.7998,
+      "step": 1848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019609289972948e-05,
+      "loss": 0.7177,
+      "step": 1849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014263853190416e-05,
+      "loss": 0.8304,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008916501729516e-05,
+      "loss": 0.7996,
+      "step": 1851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003567237632113e-05,
+      "loss": 0.8651,
+      "step": 1852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998216062940804e-05,
+      "loss": 0.7985,
+      "step": 1853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992862979698915e-05,
+      "loss": 0.8366,
+      "step": 1854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987507989950505e-05,
+      "loss": 0.843,
+      "step": 1855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982151095740356e-05,
+      "loss": 0.7172,
+      "step": 1856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976792299113978e-05,
+      "loss": 0.8074,
+      "step": 1857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971431602117606e-05,
+      "loss": 0.8192,
+      "step": 1858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966069006798203e-05,
+      "loss": 0.8853,
+      "step": 1859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960704515203458e-05,
+      "loss": 0.8175,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955338129381785e-05,
+      "loss": 0.8577,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949969851382315e-05,
+      "loss": 0.7665,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944599683254903e-05,
+      "loss": 0.8015,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939227627050135e-05,
+      "loss": 0.7931,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933853684819307e-05,
+      "loss": 0.7929,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928477858614439e-05,
+      "loss": 0.8278,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923100150488272e-05,
+      "loss": 0.6757,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917720562494263e-05,
+      "loss": 0.8542,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491233909668659e-05,
+      "loss": 0.7541,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906955755120145e-05,
+      "loss": 0.8129,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901570539850538e-05,
+      "loss": 0.7981,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896183452934087e-05,
+      "loss": 0.8157,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890794496427845e-05,
+      "loss": 0.7565,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885403672389553e-05,
+      "loss": 0.765,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880010982877688e-05,
+      "loss": 0.76,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874616429951419e-05,
+      "loss": 0.7964,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486922001567064e-05,
+      "loss": 0.8478,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863821742095955e-05,
+      "loss": 0.8166,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858421611288671e-05,
+      "loss": 0.7025,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853019625310813e-05,
+      "loss": 0.7505,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847615786225102e-05,
+      "loss": 0.7449,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842210096094984e-05,
+      "loss": 0.7736,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836802556984592e-05,
+      "loss": 0.7005,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831393170958785e-05,
+      "loss": 0.7897,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482598194008311e-05,
+      "loss": 0.7601,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820568866423828e-05,
+      "loss": 0.8515,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815153952047902e-05,
+      "loss": 0.7662,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809737199022998e-05,
+      "loss": 0.7911,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804318609417482e-05,
+      "loss": 0.769,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798898185300422e-05,
+      "loss": 0.7933,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793475928741589e-05,
+      "loss": 0.7784,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478805184181145e-05,
+      "loss": 0.8794,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782625926581177e-05,
+      "loss": 0.8317,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477719818512263e-05,
+      "loss": 0.86,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771768619508375e-05,
+      "loss": 0.7521,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766337231811677e-05,
+      "loss": 0.7673,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760904024106482e-05,
+      "loss": 0.8154,
+      "step": 1897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755468998467446e-05,
+      "loss": 0.6431,
+      "step": 1898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750032156969912e-05,
+      "loss": 0.7617,
+      "step": 1899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474459350168992e-05,
+      "loss": 0.8858,
+      "step": 1900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47391530347042e-05,
+      "loss": 0.9354,
+      "step": 1901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733710758090175e-05,
+      "loss": 0.7611,
+      "step": 1902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728266673925955e-05,
+      "loss": 0.7226,
+      "step": 1903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722820784290346e-05,
+      "loss": 0.85,
+      "step": 1904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717373091262843e-05,
+      "loss": 0.8442,
+      "step": 1905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711923596923627e-05,
+      "loss": 0.8643,
+      "step": 1906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706472303353567e-05,
+      "loss": 0.869,
+      "step": 1907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701019212634215e-05,
+      "loss": 0.7481,
+      "step": 1908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695564326847822e-05,
+      "loss": 0.7919,
+      "step": 1909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690107648077311e-05,
+      "loss": 0.7774,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684649178406298e-05,
+      "loss": 0.7675,
+      "step": 1911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679188919919076e-05,
+      "loss": 0.796,
+      "step": 1912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4673726874700629e-05,
+      "loss": 0.8114,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668263044836612e-05,
+      "loss": 0.8804,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662797432413379e-05,
+      "loss": 0.6801,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657330039517948e-05,
+      "loss": 0.7709,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651860868238017e-05,
+      "loss": 0.7875,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646389920661983e-05,
+      "loss": 0.7963,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640917198878896e-05,
+      "loss": 0.7443,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635442704978502e-05,
+      "loss": 0.8164,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629966441051208e-05,
+      "loss": 0.741,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624488409188116e-05,
+      "loss": 0.7213,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619008611480983e-05,
+      "loss": 0.738,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613527050022258e-05,
+      "loss": 0.7927,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460804372690505e-05,
+      "loss": 0.8193,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602558644223146e-05,
+      "loss": 0.7475,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597071804071008e-05,
+      "loss": 0.8507,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591583208543765e-05,
+      "loss": 0.7856,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586092859737214e-05,
+      "loss": 0.7458,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580600759747826e-05,
+      "loss": 0.8388,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575106910672744e-05,
+      "loss": 0.8178,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569611314609767e-05,
+      "loss": 0.876,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4564113973657372e-05,
+      "loss": 0.7657,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558614889914698e-05,
+      "loss": 0.7893,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455311406548155e-05,
+      "loss": 0.8139,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547611502458402e-05,
+      "loss": 0.8601,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454210720294638e-05,
+      "loss": 0.7316,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536601169047288e-05,
+      "loss": 0.8554,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531093402863579e-05,
+      "loss": 0.8068,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525583906498377e-05,
+      "loss": 0.9624,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452007268205546e-05,
+      "loss": 0.7278,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514559731639273e-05,
+      "loss": 0.7893,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509045057354914e-05,
+      "loss": 0.7914,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450352866130814e-05,
+      "loss": 0.8358,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449801054560537e-05,
+      "loss": 0.8022,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449249071235367e-05,
+      "loss": 0.7321,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486969163660778e-05,
+      "loss": 0.8881,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481445901635068e-05,
+      "loss": 0.8212,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475920928385582e-05,
+      "loss": 0.8923,
+      "step": 1949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447039424602201e-05,
+      "loss": 0.8077,
+      "step": 1950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464865856654698e-05,
+      "loss": 0.7638,
+      "step": 1951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459335762394637e-05,
+      "loss": 0.7902,
+      "step": 1952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453803965353474e-05,
+      "loss": 0.8277,
+      "step": 1953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444827046764351e-05,
+      "loss": 0.8018,
+      "step": 1954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442735271377686e-05,
+      "loss": 0.877,
+      "step": 1955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437198378669598e-05,
+      "loss": 0.7602,
+      "step": 1956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431659791633488e-05,
+      "loss": 0.8133,
+      "step": 1957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442611951238425e-05,
+      "loss": 0.8385,
+      "step": 1958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442057754303741e-05,
+      "loss": 0.7262,
+      "step": 1959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415033885709155e-05,
+      "loss": 0.8232,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440948854251631e-05,
+      "loss": 0.7977,
+      "step": 1961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 0.8351,
+      "step": 1962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398392807007372e-05,
+      "loss": 0.7354,
+      "step": 1963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392842418928137e-05,
+      "loss": 0.8012,
+      "step": 1964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387290353458048e-05,
+      "loss": 0.812,
+      "step": 1965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438173661271713e-05,
+      "loss": 0.7706,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376181198826067e-05,
+      "loss": 0.7977,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370624113906168e-05,
+      "loss": 0.7732,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365065360079391e-05,
+      "loss": 0.7072,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435950493946832e-05,
+      "loss": 0.8607,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353942854196186e-05,
+      "loss": 0.7683,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434837910638685e-05,
+      "loss": 0.7425,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342813698164804e-05,
+      "loss": 0.7851,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337246631655186e-05,
+      "loss": 0.7937,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331677908983752e-05,
+      "loss": 0.717,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326107532276908e-05,
+      "loss": 0.8876,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320535503661674e-05,
+      "loss": 0.783,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314961825265714e-05,
+      "loss": 0.7837,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309386499217315e-05,
+      "loss": 0.7212,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303809527645393e-05,
+      "loss": 0.8191,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298230912679497e-05,
+      "loss": 0.7554,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42926506564498e-05,
+      "loss": 0.754,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287068761087103e-05,
+      "loss": 0.8113,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281485228722827e-05,
+      "loss": 0.7381,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275900061489032e-05,
+      "loss": 0.7538,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270313261518389e-05,
+      "loss": 0.739,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264724830944198e-05,
+      "loss": 0.7323,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259134771900383e-05,
+      "loss": 0.7965,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253543086521482e-05,
+      "loss": 0.7514,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247949776942665e-05,
+      "loss": 0.8373,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242354845299713e-05,
+      "loss": 0.8225,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236758293729034e-05,
+      "loss": 0.8095,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423116012436765e-05,
+      "loss": 0.7239,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42255603393532e-05,
+      "loss": 0.7375,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219958940823941e-05,
+      "loss": 0.8431,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214355930918751e-05,
+      "loss": 0.7974,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208751311777112e-05,
+      "loss": 0.7153,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203145085539131e-05,
+      "loss": 0.7208,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197537254345526e-05,
+      "loss": 0.7795,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191927820337627e-05,
+      "loss": 0.7615,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186316785657374e-05,
+      "loss": 0.7684,
+      "step": 2001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4180704152447322e-05,
+      "loss": 0.8005,
+      "step": 2002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175089922850633e-05,
+      "loss": 0.7614,
+      "step": 2003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416947409901108e-05,
+      "loss": 0.8663,
+      "step": 2004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163856683073047e-05,
+      "loss": 0.753,
+      "step": 2005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415823767718152e-05,
+      "loss": 0.9051,
+      "step": 2006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41526170834821e-05,
+      "loss": 0.8081,
+      "step": 2007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146994904120988e-05,
+      "loss": 0.834,
+      "step": 2008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141371141244994e-05,
+      "loss": 0.7413,
+      "step": 2009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413574579700153e-05,
+      "loss": 0.8876,
+      "step": 2010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130118873538612e-05,
+      "loss": 0.8811,
+      "step": 2011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124490373004864e-05,
+      "loss": 0.8244,
+      "step": 2012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118860297549507e-05,
+      "loss": 0.7421,
+      "step": 2013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113228649322361e-05,
+      "loss": 0.9715,
+      "step": 2014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107595430473855e-05,
+      "loss": 0.7792,
+      "step": 2015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410196064315501e-05,
+      "loss": 0.9228,
+      "step": 2016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096324289517452e-05,
+      "loss": 0.6775,
+      "step": 2017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090686371713403e-05,
+      "loss": 0.8166,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085046891895674e-05,
+      "loss": 0.875,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079405852217685e-05,
+      "loss": 0.75,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407376325483345e-05,
+      "loss": 0.8341,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068119101897568e-05,
+      "loss": 0.7966,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062473395565244e-05,
+      "loss": 0.7488,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405682613799226e-05,
+      "loss": 0.7353,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051177331335014e-05,
+      "loss": 0.7367,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045526977750472e-05,
+      "loss": 0.8344,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039875079396207e-05,
+      "loss": 0.7721,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034221638430372e-05,
+      "loss": 0.7588,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4028566657011716e-05,
+      "loss": 0.7186,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022910137299572e-05,
+      "loss": 0.8062,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017252081453859e-05,
+      "loss": 0.8189,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011592491635088e-05,
+      "loss": 0.7942,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005931370004351e-05,
+      "loss": 0.768,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000268718723327e-05,
+      "loss": 0.7494,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399460453995428e-05,
+      "loss": 0.8789,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988938835860056e-05,
+      "loss": 0.8605,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983271608604078e-05,
+      "loss": 0.9434,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977602860350364e-05,
+      "loss": 0.8241,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39719325932635e-05,
+      "loss": 0.7815,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396626080950866e-05,
+      "loss": 0.7207,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960587511251592e-05,
+      "loss": 0.6874,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954912700658626e-05,
+      "loss": 0.7301,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394923637989666e-05,
+      "loss": 0.7723,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943558551133186e-05,
+      "loss": 0.8494,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937879216536262e-05,
+      "loss": 0.8122,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932198378274512e-05,
+      "loss": 0.8241,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926516038517153e-05,
+      "loss": 0.8515,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392083219943396e-05,
+      "loss": 0.7793,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915146863195292e-05,
+      "loss": 0.7714,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909460031972067e-05,
+      "loss": 0.7521,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903771707935785e-05,
+      "loss": 0.8973,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389808189325851e-05,
+      "loss": 0.8335,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892390590112876e-05,
+      "loss": 0.7715,
+      "step": 2053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886697800672094e-05,
+      "loss": 0.8082,
+      "step": 2054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881003527109922e-05,
+      "loss": 0.7824,
+      "step": 2055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875307771600706e-05,
+      "loss": 0.8639,
+      "step": 2056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869610536319349e-05,
+      "loss": 0.751,
+      "step": 2057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386391182344132e-05,
+      "loss": 0.7602,
+      "step": 2058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858211635142649e-05,
+      "loss": 0.7313,
+      "step": 2059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852509973599935e-05,
+      "loss": 0.7439,
+      "step": 2060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846806840990335e-05,
+      "loss": 0.7869,
+      "step": 2061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841102239491567e-05,
+      "loss": 0.778,
+      "step": 2062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835396171281916e-05,
+      "loss": 0.7997,
+      "step": 2063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829688638540222e-05,
+      "loss": 0.8112,
+      "step": 2064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823979643445884e-05,
+      "loss": 0.7747,
+      "step": 2065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818269188178865e-05,
+      "loss": 0.8425,
+      "step": 2066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381255727491968e-05,
+      "loss": 0.735,
+      "step": 2067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806843905849398e-05,
+      "loss": 0.7504,
+      "step": 2068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801129083149652e-05,
+      "loss": 0.6597,
+      "step": 2069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795412809002628e-05,
+      "loss": 0.7856,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789695085591062e-05,
+      "loss": 0.7667,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783975915098244e-05,
+      "loss": 0.8154,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778255299708021e-05,
+      "loss": 0.7921,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377253324160479e-05,
+      "loss": 0.8102,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766809742973494e-05,
+      "loss": 0.6984,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761084805999636e-05,
+      "loss": 0.829,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755358432869256e-05,
+      "loss": 0.7998,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749630625768953e-05,
+      "loss": 0.7507,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374390138688587e-05,
+      "loss": 0.7937,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738170718407689e-05,
+      "loss": 0.7887,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732438622522655e-05,
+      "loss": 0.7347,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726705101419538e-05,
+      "loss": 0.8409,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720970157287668e-05,
+      "loss": 0.913,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715233792316912e-05,
+      "loss": 0.8306,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370949600869768e-05,
+      "loss": 0.7268,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703756808620923e-05,
+      "loss": 0.8812,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369801619427813e-05,
+      "loss": 0.7618,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692274167861342e-05,
+      "loss": 0.8338,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686530731563123e-05,
+      "loss": 0.7528,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680785887576591e-05,
+      "loss": 0.848,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675039638095386e-05,
+      "loss": 0.7904,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669291985313695e-05,
+      "loss": 0.728,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663542931426239e-05,
+      "loss": 0.8383,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657792478628274e-05,
+      "loss": 0.6759,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3652040629115588e-05,
+      "loss": 0.81,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646287385084503e-05,
+      "loss": 0.8255,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640532748731877e-05,
+      "loss": 0.8176,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634776722255097e-05,
+      "loss": 0.8573,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629019307852076e-05,
+      "loss": 0.7165,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623260507721267e-05,
+      "loss": 0.8128,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617500324061644e-05,
+      "loss": 0.7416,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611738759072712e-05,
+      "loss": 0.7807,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605975814954507e-05,
+      "loss": 0.7335,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3600211493907582e-05,
+      "loss": 0.7198,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594445798133027e-05,
+      "loss": 0.7169,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358867872983245e-05,
+      "loss": 0.7495,
+      "step": 2106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582910291207988e-05,
+      "loss": 0.8158,
+      "step": 2107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577140484462295e-05,
+      "loss": 0.6954,
+      "step": 2108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357136931179855e-05,
+      "loss": 0.814,
+      "step": 2109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565596775420455e-05,
+      "loss": 0.773,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559822877532234e-05,
+      "loss": 0.7638,
+      "step": 2111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554047620338629e-05,
+      "loss": 0.8246,
+      "step": 2112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548271006044896e-05,
+      "loss": 0.7729,
+      "step": 2113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542493036856818e-05,
+      "loss": 0.784,
+      "step": 2114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536713714980691e-05,
+      "loss": 0.8287,
+      "step": 2115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530933042623331e-05,
+      "loss": 0.7484,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352515102199206e-05,
+      "loss": 0.7141,
+      "step": 2117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519367655294728e-05,
+      "loss": 0.8186,
+      "step": 2118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513582944739686e-05,
+      "loss": 0.8277,
+      "step": 2119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507796892535812e-05,
+      "loss": 0.8201,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502009500892483e-05,
+      "loss": 0.7376,
+      "step": 2121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496220772019597e-05,
+      "loss": 0.8228,
+      "step": 2122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3490430708127556e-05,
+      "loss": 0.7985,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484639311427275e-05,
+      "loss": 0.7816,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478846584130181e-05,
+      "loss": 0.6664,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 0.7654,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467257146593784e-05,
+      "loss": 0.6889,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461460440779863e-05,
+      "loss": 0.7526,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455662413219894e-05,
+      "loss": 0.8042,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449863066127835e-05,
+      "loss": 0.7769,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444062401718143e-05,
+      "loss": 0.7749,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438260422205779e-05,
+      "loss": 0.7496,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432457129806211e-05,
+      "loss": 0.8481,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426652526735404e-05,
+      "loss": 0.7657,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420846615209824e-05,
+      "loss": 0.7092,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415039397446439e-05,
+      "loss": 0.7877,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340923087566271e-05,
+      "loss": 0.9575,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403421052076604e-05,
+      "loss": 0.8082,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3397609928906581e-05,
+      "loss": 0.8053,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391797508371592e-05,
+      "loss": 0.8323,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385983792691093e-05,
+      "loss": 0.8131,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380168784085028e-05,
+      "loss": 0.8071,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374352484773841e-05,
+      "loss": 0.7985,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336853489697846e-05,
+      "loss": 0.8472,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362716022920312e-05,
+      "loss": 0.7994,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335689586482131e-05,
+      "loss": 0.8082,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351074424903863e-05,
+      "loss": 0.6329,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334525170539087e-05,
+      "loss": 0.811,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339427708505705e-05,
+      "loss": 0.7296,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333602436472248e-05,
+      "loss": 0.8342,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327775891514855e-05,
+      "loss": 0.7984,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321948075858377e-05,
+      "loss": 0.7952,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316118991728135e-05,
+      "loss": 0.7202,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310288641349952e-05,
+      "loss": 0.8011,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304457026950115e-05,
+      "loss": 0.748,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298624150755414e-05,
+      "loss": 0.7725,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292790014993112e-05,
+      "loss": 0.7241,
+      "step": 2157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286954621890947e-05,
+      "loss": 0.7774,
+      "step": 2158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281117973677145e-05,
+      "loss": 0.8405,
+      "step": 2159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275280072580405e-05,
+      "loss": 0.7698,
+      "step": 2160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269440920829918e-05,
+      "loss": 0.8004,
+      "step": 2161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263600520655333e-05,
+      "loss": 0.7992,
+      "step": 2162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257758874286792e-05,
+      "loss": 0.7923,
+      "step": 2163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32519159839549e-05,
+      "loss": 0.7507,
+      "step": 2164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246071851890751e-05,
+      "loss": 0.7598,
+      "step": 2165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32402264803259e-05,
+      "loss": 0.8315,
+      "step": 2166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234379871492381e-05,
+      "loss": 0.7809,
+      "step": 2167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228532027622705e-05,
+      "loss": 0.7505,
+      "step": 2168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222682950949843e-05,
+      "loss": 0.722,
+      "step": 2169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216832643707252e-05,
+      "loss": 0.8491,
+      "step": 2170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321098110812884e-05,
+      "loss": 0.685,
+      "step": 2171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205128346449003e-05,
+      "loss": 0.7479,
+      "step": 2172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319927436090259e-05,
+      "loss": 0.7277,
+      "step": 2173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193419153724933e-05,
+      "loss": 0.7808,
+      "step": 2174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187562727151812e-05,
+      "loss": 0.7766,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318170508341949e-05,
+      "loss": 0.8315,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175846224764683e-05,
+      "loss": 0.8042,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3169986153424575e-05,
+      "loss": 0.7293,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164124871636812e-05,
+      "loss": 0.7482,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158262381639504e-05,
+      "loss": 0.8223,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152398685671223e-05,
+      "loss": 0.8503,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146533785970997e-05,
+      "loss": 0.8029,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314066768477832e-05,
+      "loss": 0.7477,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134800384333138e-05,
+      "loss": 0.8497,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312893188687586e-05,
+      "loss": 0.7188,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123062194647349e-05,
+      "loss": 0.7132,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117191309888927e-05,
+      "loss": 0.7862,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111319234842373e-05,
+      "loss": 0.8053,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310544597174991e-05,
+      "loss": 0.8443,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099571522854229e-05,
+      "loss": 0.7397,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093695890398462e-05,
+      "loss": 0.8712,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087819076626201e-05,
+      "loss": 0.7233,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081941083781482e-05,
+      "loss": 0.8695,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076061914108802e-05,
+      "loss": 0.8458,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070181569853093e-05,
+      "loss": 0.8376,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064300053259749e-05,
+      "loss": 0.7971,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058417366574602e-05,
+      "loss": 0.8162,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052533512043935e-05,
+      "loss": 0.6903,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304664849191448e-05,
+      "loss": 0.7685,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040762308433402e-05,
+      "loss": 0.7201,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034874963848327e-05,
+      "loss": 0.7434,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028986460407312e-05,
+      "loss": 0.8348,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023096800358865e-05,
+      "loss": 0.8338,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017205985951926e-05,
+      "loss": 0.7814,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011314019435881e-05,
+      "loss": 0.8775,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005420903060562e-05,
+      "loss": 0.7811,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999526639076226e-05,
+      "loss": 0.6877,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993631229733584e-05,
+      "loss": 0.8416,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987734677283768e-05,
+      "loss": 0.8128,
+      "step": 2209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981836983978363e-05,
+      "loss": 0.6808,
+      "step": 2210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2975938152069376e-05,
+      "loss": 0.839,
+      "step": 2211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297003818380926e-05,
+      "loss": 0.6736,
+      "step": 2212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964137081450888e-05,
+      "loss": 0.8658,
+      "step": 2213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958234847247584e-05,
+      "loss": 0.7904,
+      "step": 2214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952331483453092e-05,
+      "loss": 0.7713,
+      "step": 2215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946426992321583e-05,
+      "loss": 0.8565,
+      "step": 2216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940521376107675e-05,
+      "loss": 0.771,
+      "step": 2217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934614637066397e-05,
+      "loss": 0.7422,
+      "step": 2218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928706777453224e-05,
+      "loss": 0.8373,
+      "step": 2219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922797799524044e-05,
+      "loss": 0.7911,
+      "step": 2220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916887705535186e-05,
+      "loss": 0.7736,
+      "step": 2221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910976497743389e-05,
+      "loss": 0.7008,
+      "step": 2222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905064178405834e-05,
+      "loss": 0.7142,
+      "step": 2223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899150749780114e-05,
+      "loss": 0.6571,
+      "step": 2224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893236214124251e-05,
+      "loss": 0.8102,
+      "step": 2225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887320573696692e-05,
+      "loss": 0.9057,
+      "step": 2226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881403830756296e-05,
+      "loss": 0.7193,
+      "step": 2227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2875485987562357e-05,
+      "loss": 0.8111,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869567046374579e-05,
+      "loss": 0.725,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863647009453086e-05,
+      "loss": 0.7593,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285772587905842e-05,
+      "loss": 0.7936,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851803657451554e-05,
+      "loss": 0.9043,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845880346893855e-05,
+      "loss": 0.6386,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839955949647127e-05,
+      "loss": 0.7974,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834030467973572e-05,
+      "loss": 0.8377,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828103904135817e-05,
+      "loss": 0.7533,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822176260396902e-05,
+      "loss": 0.7866,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816247539020268e-05,
+      "loss": 0.6591,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810317742269783e-05,
+      "loss": 0.8492,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804386872409711e-05,
+      "loss": 0.6962,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798454931704743e-05,
+      "loss": 0.8862,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792521922419958e-05,
+      "loss": 0.8253,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786587846820865e-05,
+      "loss": 0.7593,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278065270717336e-05,
+      "loss": 0.812,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774716505743758e-05,
+      "loss": 0.7718,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768779244798778e-05,
+      "loss": 0.8116,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762840926605535e-05,
+      "loss": 0.8745,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756901553431564e-05,
+      "loss": 0.8085,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750961127544782e-05,
+      "loss": 0.7284,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745019651213525e-05,
+      "loss": 0.8001,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739077126706527e-05,
+      "loss": 0.803,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733133556292914e-05,
+      "loss": 0.7178,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727188942242216e-05,
+      "loss": 0.7695,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721243286824367e-05,
+      "loss": 0.7698,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715296592309693e-05,
+      "loss": 0.7483,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709348860968916e-05,
+      "loss": 0.8035,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703400095073153e-05,
+      "loss": 0.7196,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697450296893918e-05,
+      "loss": 0.7979,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269149946870313e-05,
+      "loss": 0.7584,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268554761277308e-05,
+      "loss": 0.9518,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679594731376468e-05,
+      "loss": 0.8969,
+      "step": 2261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673640826786378e-05,
+      "loss": 0.6891,
+      "step": 2262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667685901276285e-05,
+      "loss": 0.7578,
+      "step": 2263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661729957120061e-05,
+      "loss": 0.7765,
+      "step": 2264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655772996591957e-05,
+      "loss": 0.8327,
+      "step": 2265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264981502196662e-05,
+      "loss": 0.8017,
+      "step": 2266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643856035519075e-05,
+      "loss": 0.7715,
+      "step": 2267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637896039524747e-05,
+      "loss": 0.8966,
+      "step": 2268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263193503625943e-05,
+      "loss": 0.7773,
+      "step": 2269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625973027999323e-05,
+      "loss": 0.7388,
+      "step": 2270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620010017020985e-05,
+      "loss": 0.7722,
+      "step": 2271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614046005601377e-05,
+      "loss": 0.7243,
+      "step": 2272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608080996017836e-05,
+      "loss": 0.7182,
+      "step": 2273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602114990548073e-05,
+      "loss": 0.7352,
+      "step": 2274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596147991470191e-05,
+      "loss": 0.7429,
+      "step": 2275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590180001062663e-05,
+      "loss": 0.7548,
+      "step": 2276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584211021604347e-05,
+      "loss": 0.7647,
+      "step": 2277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578241055374475e-05,
+      "loss": 0.793,
+      "step": 2278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257227010465266e-05,
+      "loss": 0.6282,
+      "step": 2279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256629817171888e-05,
+      "loss": 0.6514,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560325258853503e-05,
+      "loss": 0.7881,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554351368337262e-05,
+      "loss": 0.7307,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548376502451265e-05,
+      "loss": 0.7475,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542400663476993e-05,
+      "loss": 0.8145,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536423853696295e-05,
+      "loss": 0.7394,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530446075391403e-05,
+      "loss": 0.7916,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524467330844898e-05,
+      "loss": 0.7621,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518487622339752e-05,
+      "loss": 0.757,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251250695215929e-05,
+      "loss": 0.7875,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 0.8205,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2500542735907571e-05,
+      "loss": 0.7648,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494559194404809e-05,
+      "loss": 0.7979,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488574700363714e-05,
+      "loss": 0.7636,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482589256069444e-05,
+      "loss": 0.7809,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476602863807519e-05,
+      "loss": 0.8044,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247061552586382e-05,
+      "loss": 0.8293,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464627244524595e-05,
+      "loss": 0.803,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458638022076438e-05,
+      "loss": 0.7962,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245264786080632e-05,
+      "loss": 0.8488,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446656763001558e-05,
+      "loss": 0.791,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440664730949836e-05,
+      "loss": 0.6538,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434671766939184e-05,
+      "loss": 0.8449,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428677873257992e-05,
+      "loss": 0.7957,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422683052195015e-05,
+      "loss": 0.7589,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416687306039345e-05,
+      "loss": 0.7724,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410690637080443e-05,
+      "loss": 0.7104,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404693047608108e-05,
+      "loss": 0.7913,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398694539912501e-05,
+      "loss": 0.7674,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392695116284131e-05,
+      "loss": 0.8295,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386694779013855e-05,
+      "loss": 0.7049,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238069353039288e-05,
+      "loss": 0.6931,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374691372712761e-05,
+      "loss": 0.843,
+      "step": 2312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368688308265404e-05,
+      "loss": 0.8527,
+      "step": 2313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236268433934305e-05,
+      "loss": 0.7926,
+      "step": 2314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356679468238298e-05,
+      "loss": 0.8027,
+      "step": 2315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350673697244083e-05,
+      "loss": 0.7734,
+      "step": 2316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344667028653687e-05,
+      "loss": 0.7682,
+      "step": 2317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338659464760736e-05,
+      "loss": 0.8556,
+      "step": 2318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332651007859198e-05,
+      "loss": 0.7748,
+      "step": 2319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326641660243375e-05,
+      "loss": 0.7655,
+      "step": 2320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320631424207919e-05,
+      "loss": 0.7152,
+      "step": 2321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314620302047818e-05,
+      "loss": 0.8232,
+      "step": 2322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308608296058391e-05,
+      "loss": 0.744,
+      "step": 2323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302595408535308e-05,
+      "loss": 0.7546,
+      "step": 2324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296581641774557e-05,
+      "loss": 0.6829,
+      "step": 2325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2290566998072487e-05,
+      "loss": 0.8158,
+      "step": 2326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284551479725756e-05,
+      "loss": 0.7633,
+      "step": 2327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278535089031377e-05,
+      "loss": 0.7695,
+      "step": 2328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2272517828286681e-05,
+      "loss": 0.7901,
+      "step": 2329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266499699789338e-05,
+      "loss": 0.7413,
+      "step": 2330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226048070583735e-05,
+      "loss": 0.7165,
+      "step": 2331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254460848729046e-05,
+      "loss": 0.8009,
+      "step": 2332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.224844013076309e-05,
+      "loss": 0.8129,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2242418554238463e-05,
+      "loss": 0.8991,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236396121454492e-05,
+      "loss": 0.7545,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230372834710816e-05,
+      "loss": 0.8156,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2224348696307406e-05,
+      "loss": 0.8627,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218323708544555e-05,
+      "loss": 0.7934,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212297873722884e-05,
+      "loss": 0.7739,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220627119414334e-05,
+      "loss": 0.7079,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200243672107186e-05,
+      "loss": 0.7808,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194215309916005e-05,
+      "loss": 0.7739,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218818610987171e-05,
+      "loss": 0.7659,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182156074276527e-05,
+      "loss": 0.8137,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176125205433005e-05,
+      "loss": 0.7221,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170093505644009e-05,
+      "loss": 0.7898,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164060977212719e-05,
+      "loss": 0.7507,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158027622442636e-05,
+      "loss": 0.6975,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151993443637576e-05,
+      "loss": 0.8563,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145958443101665e-05,
+      "loss": 0.8121,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139922623139346e-05,
+      "loss": 0.8335,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133885986055379e-05,
+      "loss": 0.6928,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127848534154827e-05,
+      "loss": 0.7574,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212181026974307e-05,
+      "loss": 0.8726,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115771195125802e-05,
+      "loss": 0.6967,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109731312609017e-05,
+      "loss": 0.7705,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210369062449902e-05,
+      "loss": 0.6906,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097649133102429e-05,
+      "loss": 0.8814,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209160684072617e-05,
+      "loss": 0.8589,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208556374967746e-05,
+      "loss": 0.7516,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207951986226384e-05,
+      "loss": 0.8152,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073475180793144e-05,
+      "loss": 0.6922,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067429707573508e-05,
+      "loss": 0.7997,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206138344491338e-05,
+      "loss": 0.7967,
+      "step": 2364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055336395121495e-05,
+      "loss": 0.7261,
+      "step": 2365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2049288560506904e-05,
+      "loss": 0.7019,
+      "step": 2366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204323994337895e-05,
+      "loss": 0.6897,
+      "step": 2367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037190546047275e-05,
+      "loss": 0.7024,
+      "step": 2368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031140370821814e-05,
+      "loss": 0.844,
+      "step": 2369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025089420012813e-05,
+      "loss": 0.832,
+      "step": 2370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20190376959308e-05,
+      "loss": 0.786,
+      "step": 2371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012985200886602e-05,
+      "loss": 0.7175,
+      "step": 2372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006931937191348e-05,
+      "loss": 0.7885,
+      "step": 2373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200087790715645e-05,
+      "loss": 0.77,
+      "step": 2374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994823113093619e-05,
+      "loss": 0.7349,
+      "step": 2375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988767557314856e-05,
+      "loss": 0.8029,
+      "step": 2376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982711242132457e-05,
+      "loss": 0.8634,
+      "step": 2377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976654169858994e-05,
+      "loss": 0.7568,
+      "step": 2378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970596342807345e-05,
+      "loss": 0.8275,
+      "step": 2379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196453776329067e-05,
+      "loss": 0.8329,
+      "step": 2380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958478433622414e-05,
+      "loss": 0.8973,
+      "step": 2381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952418356116309e-05,
+      "loss": 0.8049,
+      "step": 2382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194635753308637e-05,
+      "loss": 0.7908,
+      "step": 2383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940295966846906e-05,
+      "loss": 0.7575,
+      "step": 2384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934233659712504e-05,
+      "loss": 0.8521,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192817061399803e-05,
+      "loss": 0.7764,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922106832018638e-05,
+      "loss": 0.7264,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916042316089762e-05,
+      "loss": 0.8259,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909977068527115e-05,
+      "loss": 0.7455,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903911091646684e-05,
+      "loss": 0.8092,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189784438776475e-05,
+      "loss": 0.8372,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891776959197854e-05,
+      "loss": 0.867,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885708808262823e-05,
+      "loss": 0.8051,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1879639937276761e-05,
+      "loss": 0.7897,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873570348557047e-05,
+      "loss": 0.7959,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867500044421321e-05,
+      "loss": 0.7759,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861429027187512e-05,
+      "loss": 0.7763,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185535729917382e-05,
+      "loss": 0.7542,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184928486269871e-05,
+      "loss": 0.8412,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843211720080913e-05,
+      "loss": 0.8072,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837137873639447e-05,
+      "loss": 0.7836,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831063325693578e-05,
+      "loss": 0.7764,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824988078562856e-05,
+      "loss": 0.843,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818912134567093e-05,
+      "loss": 0.8327,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181283549602636e-05,
+      "loss": 0.8632,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806758165261006e-05,
+      "loss": 0.8165,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800680144591632e-05,
+      "loss": 0.6987,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794601436339114e-05,
+      "loss": 0.7256,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788522042824579e-05,
+      "loss": 0.763,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782441966369427e-05,
+      "loss": 0.7609,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776361209295303e-05,
+      "loss": 0.7857,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770279773924133e-05,
+      "loss": 0.7686,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764197662578087e-05,
+      "loss": 0.7097,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758114877579593e-05,
+      "loss": 0.7738,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752031421251347e-05,
+      "loss": 0.8133,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745947295916287e-05,
+      "loss": 0.7957,
+      "step": 2416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739862503897624e-05,
+      "loss": 0.883,
+      "step": 2417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733777047518804e-05,
+      "loss": 0.7015,
+      "step": 2418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172769092910354e-05,
+      "loss": 0.7018,
+      "step": 2419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1721604150975795e-05,
+      "loss": 0.7641,
+      "step": 2420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715516715459784e-05,
+      "loss": 0.8303,
+      "step": 2421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709428624879971e-05,
+      "loss": 0.8055,
+      "step": 2422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703339881561068e-05,
+      "loss": 0.6736,
+      "step": 2423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697250487828042e-05,
+      "loss": 0.7939,
+      "step": 2424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691160446006106e-05,
+      "loss": 0.7164,
+      "step": 2425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685069758420724e-05,
+      "loss": 0.755,
+      "step": 2426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678978427397597e-05,
+      "loss": 0.7571,
+      "step": 2427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167288645526268e-05,
+      "loss": 0.7592,
+      "step": 2428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666793844342168e-05,
+      "loss": 0.6936,
+      "step": 2429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660700596962506e-05,
+      "loss": 0.7515,
+      "step": 2430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654606715450377e-05,
+      "loss": 0.7451,
+      "step": 2431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648512202132705e-05,
+      "loss": 0.8547,
+      "step": 2432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1642417059336659e-05,
+      "loss": 0.825,
+      "step": 2433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163632128938965e-05,
+      "loss": 0.8019,
+      "step": 2434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630224894619322e-05,
+      "loss": 0.711,
+      "step": 2435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624127877353562e-05,
+      "loss": 0.739,
+      "step": 2436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618030239920497e-05,
+      "loss": 0.8201,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161193198464848e-05,
+      "loss": 0.6732,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605833113866116e-05,
+      "loss": 0.8031,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1599733629902232e-05,
+      "loss": 0.7408,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593633535085894e-05,
+      "loss": 0.7912,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15875328317464e-05,
+      "loss": 0.7866,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581431522213285e-05,
+      "loss": 0.7588,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157532960881631e-05,
+      "loss": 0.6848,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569227093885467e-05,
+      "loss": 0.7682,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563123979750981e-05,
+      "loss": 0.749,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557020268743303e-05,
+      "loss": 0.795,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155091596319311e-05,
+      "loss": 0.7933,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544811065431313e-05,
+      "loss": 0.7492,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538705577789045e-05,
+      "loss": 0.8083,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532599502597661e-05,
+      "loss": 0.8191,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526492842188746e-05,
+      "loss": 0.8312,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1520385598894106e-05,
+      "loss": 0.7174,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 0.8416,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508169372975986e-05,
+      "loss": 0.8366,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150206039501722e-05,
+      "loss": 0.8559,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495950843502176e-05,
+      "loss": 0.9533,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489840720763758e-05,
+      "loss": 0.8631,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483730029135091e-05,
+      "loss": 0.7142,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477618770949522e-05,
+      "loss": 0.7561,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471506948540613e-05,
+      "loss": 0.8013,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465394564242142e-05,
+      "loss": 0.8148,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459281620388102e-05,
+      "loss": 0.9193,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453168119312695e-05,
+      "loss": 0.7228,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447054063350336e-05,
+      "loss": 0.7339,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440939454835665e-05,
+      "loss": 0.6639,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434824296103514e-05,
+      "loss": 0.7551,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428708589488942e-05,
+      "loss": 0.7499,
+      "step": 2468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422592337327198e-05,
+      "loss": 0.7112,
+      "step": 2469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141647554195376e-05,
+      "loss": 0.7065,
+      "step": 2470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410358205704304e-05,
+      "loss": 0.6921,
+      "step": 2471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404240330914706e-05,
+      "loss": 0.8077,
+      "step": 2472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398121919921062e-05,
+      "loss": 0.832,
+      "step": 2473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392002975059656e-05,
+      "loss": 0.8041,
+      "step": 2474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385883498666992e-05,
+      "loss": 0.8235,
+      "step": 2475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379763493079767e-05,
+      "loss": 0.8749,
+      "step": 2476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373642960634886e-05,
+      "loss": 0.7667,
+      "step": 2477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367521903669444e-05,
+      "loss": 0.7388,
+      "step": 2478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361400324520753e-05,
+      "loss": 0.7292,
+      "step": 2479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355278225526311e-05,
+      "loss": 0.7675,
+      "step": 2480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349155609023821e-05,
+      "loss": 0.833,
+      "step": 2481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343032477351183e-05,
+      "loss": 0.7814,
+      "step": 2482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336908832846485e-05,
+      "loss": 0.8184,
+      "step": 2483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330784677848028e-05,
+      "loss": 0.7595,
+      "step": 2484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324660014694296e-05,
+      "loss": 0.7595,
+      "step": 2485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318534845723966e-05,
+      "loss": 0.7031,
+      "step": 2486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131240917327591e-05,
+      "loss": 0.7022,
+      "step": 2487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306282999689197e-05,
+      "loss": 0.8154,
+      "step": 2488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300156327303089e-05,
+      "loss": 0.7902,
+      "step": 2489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1294029158457023e-05,
+      "loss": 0.7404,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287901495490646e-05,
+      "loss": 0.6433,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128177334074377e-05,
+      "loss": 0.7431,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275644696556427e-05,
+      "loss": 0.7951,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269515565268805e-05,
+      "loss": 0.8097,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263385949221294e-05,
+      "loss": 0.8665,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257255850754464e-05,
+      "loss": 0.6919,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125112527220907e-05,
+      "loss": 0.7452,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244994215926058e-05,
+      "loss": 0.8139,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238862684246538e-05,
+      "loss": 0.615,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232730679511825e-05,
+      "loss": 0.7041,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226598204063393e-05,
+      "loss": 0.8946,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122046526024291e-05,
+      "loss": 0.8072,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214331850392219e-05,
+      "loss": 0.8792,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208197976853339e-05,
+      "loss": 0.8092,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202063641968466e-05,
+      "loss": 0.7359,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195928848079974e-05,
+      "loss": 0.8347,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189793597530412e-05,
+      "loss": 0.834,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183657892662502e-05,
+      "loss": 0.7812,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177521735819146e-05,
+      "loss": 0.702,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171385129343403e-05,
+      "loss": 0.7393,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165248075578525e-05,
+      "loss": 0.8518,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159110576867915e-05,
+      "loss": 0.7541,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115297263555516e-05,
+      "loss": 0.7524,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146834253984008e-05,
+      "loss": 0.8036,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140695434498374e-05,
+      "loss": 0.8,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134556179442352e-05,
+      "loss": 0.8572,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128416491160188e-05,
+      "loss": 0.6987,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122276371996304e-05,
+      "loss": 0.8657,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116135824295274e-05,
+      "loss": 0.7286,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1109994850401854e-05,
+      "loss": 0.7509,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103853452660945e-05,
+      "loss": 0.8228,
+      "step": 2521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097711633417623e-05,
+      "loss": 0.7312,
+      "step": 2522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109156939501711e-05,
+      "loss": 0.6756,
+      "step": 2523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108542673980481e-05,
+      "loss": 0.7277,
+      "step": 2524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1079283670126267e-05,
+      "loss": 0.7353,
+      "step": 2525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107314018832719e-05,
+      "loss": 0.7384,
+      "step": 2526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066996296753441e-05,
+      "loss": 0.7751,
+      "step": 2527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106085199775105e-05,
+      "loss": 0.866,
+      "step": 2528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054707293666191e-05,
+      "loss": 0.7797,
+      "step": 2529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048562186845198e-05,
+      "loss": 0.7653,
+      "step": 2530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042416679634558e-05,
+      "loss": 0.7821,
+      "step": 2531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036270774380906e-05,
+      "loss": 0.8204,
+      "step": 2532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030124473431043e-05,
+      "loss": 0.8016,
+      "step": 2533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023977779131903e-05,
+      "loss": 0.8081,
+      "step": 2534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017830693830584e-05,
+      "loss": 0.7415,
+      "step": 2535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011683219874324e-05,
+      "loss": 0.7849,
+      "step": 2536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005535359610513e-05,
+      "loss": 0.7694,
+      "step": 2537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999387115386697e-05,
+      "loss": 0.7485,
+      "step": 2538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993238489550551e-05,
+      "loss": 0.7854,
+      "step": 2539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987089484449911e-05,
+      "loss": 0.7895,
+      "step": 2540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098094010243275e-05,
+      "loss": 0.8483,
+      "step": 2541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974790345847187e-05,
+      "loss": 0.7333,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968640217041486e-05,
+      "loss": 0.8755,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096248971836405e-05,
+      "loss": 0.8027,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956338852163424e-05,
+      "loss": 0.7956,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950187620788296e-05,
+      "loss": 0.7776,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944036026587487e-05,
+      "loss": 0.7096,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937884071909965e-05,
+      "loss": 0.7325,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931731759104831e-05,
+      "loss": 0.8632,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925579090521319e-05,
+      "loss": 0.7758,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091942606850881e-05,
+      "loss": 0.7782,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913272695416807e-05,
+      "loss": 0.7844,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0907118973594957e-05,
+      "loss": 0.8007,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900964905393032e-05,
+      "loss": 0.766,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894810493160946e-05,
+      "loss": 0.7559,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888655739248735e-05,
+      "loss": 0.8213,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882500646006568e-05,
+      "loss": 0.8016,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876345215784751e-05,
+      "loss": 0.8542,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870189450933703e-05,
+      "loss": 0.7141,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864033353803993e-05,
+      "loss": 0.7636,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857876926746294e-05,
+      "loss": 0.7047,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085172017211142e-05,
+      "loss": 0.8785,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845563092250301e-05,
+      "loss": 0.8934,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839405689514e-05,
+      "loss": 0.7944,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833247966253699e-05,
+      "loss": 0.7692,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827089924820699e-05,
+      "loss": 0.8013,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820931567566429e-05,
+      "loss": 0.7722,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814772896842427e-05,
+      "loss": 0.7425,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808613915000373e-05,
+      "loss": 0.7084,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802454624392037e-05,
+      "loss": 0.7772,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796295027369335e-05,
+      "loss": 0.8325,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790135126284275e-05,
+      "loss": 0.7862,
+      "step": 2572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783974923489002e-05,
+      "loss": 0.6932,
+      "step": 2573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777814421335764e-05,
+      "loss": 0.7508,
+      "step": 2574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077165362217692e-05,
+      "loss": 0.849,
+      "step": 2575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076549252836496e-05,
+      "loss": 0.8844,
+      "step": 2576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759331142252463e-05,
+      "loss": 0.8257,
+      "step": 2577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753169466192144e-05,
+      "loss": 0.7312,
+      "step": 2578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747007502536806e-05,
+      "loss": 0.8136,
+      "step": 2579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074084525363938e-05,
+      "loss": 0.8227,
+      "step": 2580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734682721852892e-05,
+      "loss": 0.7632,
+      "step": 2581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072851990953049e-05,
+      "loss": 0.7983,
+      "step": 2582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722356819025415e-05,
+      "loss": 0.7426,
+      "step": 2583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716193452691024e-05,
+      "loss": 0.768,
+      "step": 2584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071002981288077e-05,
+      "loss": 0.76,
+      "step": 2585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703865901948225e-05,
+      "loss": 0.7161,
+      "step": 2586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697701722247051e-05,
+      "loss": 0.8528,
+      "step": 2587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691537276131018e-05,
+      "loss": 0.8028,
+      "step": 2588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685372565954e-05,
+      "loss": 0.7449,
+      "step": 2589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679207594069963e-05,
+      "loss": 0.8566,
+      "step": 2590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673042362832984e-05,
+      "loss": 0.8057,
+      "step": 2591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666876874597235e-05,
+      "loss": 0.768,
+      "step": 2592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066071113171698e-05,
+      "loss": 0.7323,
+      "step": 2593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065454513654659e-05,
+      "loss": 0.8751,
+      "step": 2594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648378891440528e-05,
+      "loss": 0.7385,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064221239875335e-05,
+      "loss": 0.8452,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636045660839707e-05,
+      "loss": 0.7639,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629878680054348e-05,
+      "loss": 0.7016,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062371145875211e-05,
+      "loss": 0.8005,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061754399928793e-05,
+      "loss": 0.726,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611376304016822e-05,
+      "loss": 0.8299,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605208375293905e-05,
+      "loss": 0.77,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599040215474374e-05,
+      "loss": 0.76,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592871826913522e-05,
+      "loss": 0.7503,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586703211966727e-05,
+      "loss": 0.7012,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580534372989447e-05,
+      "loss": 0.6823,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574365312337235e-05,
+      "loss": 0.7405,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0568196032365719e-05,
+      "loss": 0.7284,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562026535430625e-05,
+      "loss": 0.6351,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555856823887745e-05,
+      "loss": 0.7801,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549686900092963e-05,
+      "loss": 0.7042,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543516766402245e-05,
+      "loss": 0.7069,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053734642517163e-05,
+      "loss": 0.7218,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531175878757243e-05,
+      "loss": 0.8196,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525005129515282e-05,
+      "loss": 0.7526,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051883417980203e-05,
+      "loss": 0.7483,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051266303197383e-05,
+      "loss": 0.6823,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 0.8459,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050032015139842e-05,
+      "loss": 0.8712,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494148423364287e-05,
+      "loss": 0.8746,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048797650664138e-05,
+      "loss": 0.7098,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481804403586421e-05,
+      "loss": 0.7452,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475632116556212e-05,
+      "loss": 0.6967,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469459647907609e-05,
+      "loss": 0.7302,
+      "step": 2624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463286999997555e-05,
+      "loss": 0.6944,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457114175183043e-05,
+      "loss": 0.7533,
+      "step": 2626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450941175821157e-05,
+      "loss": 0.7924,
+      "step": 2627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444768004269024e-05,
+      "loss": 0.7447,
+      "step": 2628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438594662883852e-05,
+      "loss": 0.7322,
+      "step": 2629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432421154022905e-05,
+      "loss": 0.8203,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042624748004352e-05,
+      "loss": 0.8462,
+      "step": 2631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420073643303085e-05,
+      "loss": 0.7988,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413899646159064e-05,
+      "loss": 0.8127,
+      "step": 2633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407725490968974e-05,
+      "loss": 0.7312,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0401551180090385e-05,
+      "loss": 0.8753,
+      "step": 2635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395376715880949e-05,
+      "loss": 0.7097,
+      "step": 2636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389202100698352e-05,
+      "loss": 0.6456,
+      "step": 2637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383027336900356e-05,
+      "loss": 0.7732,
+      "step": 2638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376852426844763e-05,
+      "loss": 0.8567,
+      "step": 2639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370677372889449e-05,
+      "loss": 0.7877,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0364502177392329e-05,
+      "loss": 0.8587,
+      "step": 2641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358326842711383e-05,
+      "loss": 0.711,
+      "step": 2642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035215137120464e-05,
+      "loss": 0.8031,
+      "step": 2643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345975765230177e-05,
+      "loss": 0.8413,
+      "step": 2644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339800027146133e-05,
+      "loss": 0.7812,
+      "step": 2645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333624159310685e-05,
+      "loss": 0.834,
+      "step": 2646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327448164082068e-05,
+      "loss": 0.7186,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321272043818565e-05,
+      "loss": 0.717,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.03150958008785e-05,
+      "loss": 0.8068,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308919437620257e-05,
+      "loss": 0.6573,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302742956402252e-05,
+      "loss": 0.8248,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296566359582951e-05,
+      "loss": 0.7768,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290389649520868e-05,
+      "loss": 0.7528,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028421282857456e-05,
+      "loss": 0.6241,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278035899102613e-05,
+      "loss": 0.7879,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271858863463678e-05,
+      "loss": 0.9554,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265681724016423e-05,
+      "loss": 0.7872,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259504483119571e-05,
+      "loss": 0.7346,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025332714313188e-05,
+      "loss": 0.8543,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247149706412144e-05,
+      "loss": 0.7795,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024097217531919e-05,
+      "loss": 0.8153,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023479455221189e-05,
+      "loss": 0.7048,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022861683944915e-05,
+      "loss": 0.812,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222439039389902e-05,
+      "loss": 0.7053,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021626115439312e-05,
+      "loss": 0.7636,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210083186817801e-05,
+      "loss": 0.8121,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203905139022983e-05,
+      "loss": 0.8552,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197727013367731e-05,
+      "loss": 0.7939,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191548812211143e-05,
+      "loss": 0.8083,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185370537912338e-05,
+      "loss": 0.8296,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179192192830468e-05,
+      "loss": 0.781,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0173013779324714e-05,
+      "loss": 0.799,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166835299754279e-05,
+      "loss": 0.8394,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160656756478393e-05,
+      "loss": 0.7589,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154478151856308e-05,
+      "loss": 0.7807,
+      "step": 2675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148299488247307e-05,
+      "loss": 0.8446,
+      "step": 2676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142120768010686e-05,
+      "loss": 0.7636,
+      "step": 2677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135941993505768e-05,
+      "loss": 0.6806,
+      "step": 2678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129763167091895e-05,
+      "loss": 0.7972,
+      "step": 2679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012358429112843e-05,
+      "loss": 0.8162,
+      "step": 2680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117405367974757e-05,
+      "loss": 0.6718,
+      "step": 2681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111226399990267e-05,
+      "loss": 0.7899,
+      "step": 2682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105047389534389e-05,
+      "loss": 0.7936,
+      "step": 2683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098868338966543e-05,
+      "loss": 0.71,
+      "step": 2684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009268925064618e-05,
+      "loss": 0.7192,
+      "step": 2685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008651012693277e-05,
+      "loss": 0.8249,
+      "step": 2686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008033097018578e-05,
+      "loss": 0.8089,
+      "step": 2687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074151782764705e-05,
+      "loss": 0.7961,
+      "step": 2688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067972567029038e-05,
+      "loss": 0.7185,
+      "step": 2689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061793325338298e-05,
+      "loss": 0.8103,
+      "step": 2690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055614060052e-05,
+      "loss": 0.7728,
+      "step": 2691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049434773529678e-05,
+      "loss": 0.719,
+      "step": 2692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043255468130866e-05,
+      "loss": 0.7691,
+      "step": 2693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037076146215112e-05,
+      "loss": 0.8492,
+      "step": 2694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030896810141965e-05,
+      "loss": 0.8079,
+      "step": 2695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024717462270988e-05,
+      "loss": 0.7448,
+      "step": 2696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018538104961732e-05,
+      "loss": 0.756,
+      "step": 2697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012358740573772e-05,
+      "loss": 0.7351,
+      "step": 2698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006179371466672e-05,
+      "loss": 0.8238,
+      "step": 2699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.8987,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99382062853333e-06,
+      "loss": 0.7673,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98764125942623e-06,
+      "loss": 0.8929,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98146189503827e-06,
+      "loss": 0.7794,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975282537729019e-06,
+      "loss": 0.8011,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969103189858038e-06,
+      "loss": 0.8101,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96292385378489e-06,
+      "loss": 0.8069,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95674453186914e-06,
+      "loss": 0.7302,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950565226470325e-06,
+      "loss": 0.7576,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944385939948001e-06,
+      "loss": 0.6527,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938206674661704e-06,
+      "loss": 0.7867,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932027432970962e-06,
+      "loss": 0.7508,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9258482172353e-06,
+      "loss": 0.7381,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919669029814223e-06,
+      "loss": 0.8316,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91348987306723e-06,
+      "loss": 0.7988,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90731074935382e-06,
+      "loss": 0.7793,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90113166103346e-06,
+      "loss": 0.7258,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894952610465617e-06,
+      "loss": 0.7824,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888773600009734e-06,
+      "loss": 0.7103,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882594632025246e-06,
+      "loss": 0.7161,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876415708871572e-06,
+      "loss": 0.7542,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870236832908106e-06,
+      "loss": 0.752,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864058006494237e-06,
+      "loss": 0.7673,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857879231989317e-06,
+      "loss": 0.7994,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851700511752695e-06,
+      "loss": 0.8015,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845521848143695e-06,
+      "loss": 0.792,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83934324352161e-06,
+      "loss": 0.7647,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833164700245723e-06,
+      "loss": 0.7914,
+      "step": 2727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826986220675289e-06,
+      "loss": 0.7943,
+      "step": 2728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820807807169534e-06,
+      "loss": 0.6888,
+      "step": 2729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814629462087666e-06,
+      "loss": 0.7188,
+      "step": 2730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80845118778886e-06,
+      "loss": 0.7404,
+      "step": 2731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80227298663227e-06,
+      "loss": 0.7165,
+      "step": 2732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79609486097702e-06,
+      "loss": 0.7682,
+      "step": 2733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789916813182204e-06,
+      "loss": 0.6812,
+      "step": 2734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783738845606886e-06,
+      "loss": 0.6825,
+      "step": 2735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777560960610101e-06,
+      "loss": 0.9305,
+      "step": 2736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771383160550851e-06,
+      "loss": 0.7393,
+      "step": 2737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.765205447788111e-06,
+      "loss": 0.7641,
+      "step": 2738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759027824680813e-06,
+      "loss": 0.7017,
+      "step": 2739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752850293587861e-06,
+      "loss": 0.7055,
+      "step": 2740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746672856868124e-06,
+      "loss": 0.7286,
+      "step": 2741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740495516880428e-06,
+      "loss": 0.7248,
+      "step": 2742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73431827598358e-06,
+      "loss": 0.7232,
+      "step": 2743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728141136536324e-06,
+      "loss": 0.8078,
+      "step": 2744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721964100897387e-06,
+      "loss": 0.7744,
+      "step": 2745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715787171425444e-06,
+      "loss": 0.7845,
+      "step": 2746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709610350479132e-06,
+      "loss": 0.7179,
+      "step": 2747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70343364041705e-06,
+      "loss": 0.7658,
+      "step": 2748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697257043597752e-06,
+      "loss": 0.799,
+      "step": 2749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691080562379744e-06,
+      "loss": 0.7896,
+      "step": 2750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684904199121502e-06,
+      "loss": 0.757,
+      "step": 2751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678727956181438e-06,
+      "loss": 0.7457,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672551835917936e-06,
+      "loss": 0.816,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666375840689319e-06,
+      "loss": 0.7727,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660199972853869e-06,
+      "loss": 0.6715,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654024234769827e-06,
+      "loss": 0.7501,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647848628795365e-06,
+      "loss": 0.6623,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641673157288622e-06,
+      "loss": 0.7319,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.635497822607674e-06,
+      "loss": 0.7094,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629322627110553e-06,
+      "loss": 0.7699,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62314757315524e-06,
+      "loss": 0.8456,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616972663099648e-06,
+      "loss": 0.7521,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610797899301648e-06,
+      "loss": 0.7115,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604623284119053e-06,
+      "loss": 0.7683,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598448819909615e-06,
+      "loss": 0.7738,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592274509031031e-06,
+      "loss": 0.7755,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586100353840939e-06,
+      "loss": 0.8156,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579926356696917e-06,
+      "loss": 0.7483,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573752519956486e-06,
+      "loss": 0.6224,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567578845977096e-06,
+      "loss": 0.8576,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561405337116153e-06,
+      "loss": 0.7906,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55523199573098e-06,
+      "loss": 0.856,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549058824178845e-06,
+      "loss": 0.6128,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542885824816959e-06,
+      "loss": 0.8296,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536713000002448e-06,
+      "loss": 0.7126,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530540352092396e-06,
+      "loss": 0.701,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524367883443793e-06,
+      "loss": 0.8389,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51819559641358e-06,
+      "loss": 0.749,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512023493358624e-06,
+      "loss": 0.6796,
+      "step": 2779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505851576635715e-06,
+      "loss": 0.7508,
+      "step": 2780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49967984860158e-06,
+      "loss": 0.6713,
+      "step": 2781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 0.8039,
+      "step": 2782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487336968026169e-06,
+      "loss": 0.8133,
+      "step": 2783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481165820197977e-06,
+      "loss": 0.7867,
+      "step": 2784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474994870484722e-06,
+      "loss": 0.6798,
+      "step": 2785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46882412124276e-06,
+      "loss": 0.8134,
+      "step": 2786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462653574828374e-06,
+      "loss": 0.8076,
+      "step": 2787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456483233597758e-06,
+      "loss": 0.7748,
+      "step": 2788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45031309990704e-06,
+      "loss": 0.6862,
+      "step": 2789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444143176112259e-06,
+      "loss": 0.7145,
+      "step": 2790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437973464569377e-06,
+      "loss": 0.6965,
+      "step": 2791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431803967634284e-06,
+      "loss": 0.637,
+      "step": 2792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425634687662768e-06,
+      "loss": 0.8045,
+      "step": 2793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419465627010558e-06,
+      "loss": 0.7508,
+      "step": 2794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413296788033276e-06,
+      "loss": 0.7938,
+      "step": 2795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407128173086478e-06,
+      "loss": 0.7379,
+      "step": 2796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400959784525627e-06,
+      "loss": 0.7481,
+      "step": 2797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394791624706096e-06,
+      "loss": 0.7811,
+      "step": 2798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388623695983178e-06,
+      "loss": 0.9082,
+      "step": 2799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382456000712072e-06,
+      "loss": 0.7771,
+      "step": 2800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37628854124789e-06,
+      "loss": 0.7386,
+      "step": 2801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370121319945657e-06,
+      "loss": 0.7576,
+      "step": 2802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363954339160298e-06,
+      "loss": 0.7756,
+      "step": 2803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357787601246654e-06,
+      "loss": 0.7772,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351621108559477e-06,
+      "loss": 0.6947,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345454863453412e-06,
+      "loss": 0.7118,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339288868283025e-06,
+      "loss": 0.7726,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33312312540277e-06,
+      "loss": 0.6859,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326957637167016e-06,
+      "loss": 0.817,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320792405930042e-06,
+      "loss": 0.6512,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314627434046002e-06,
+      "loss": 0.7544,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308462723868987e-06,
+      "loss": 0.8193,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30229827775295e-06,
+      "loss": 0.7748,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296134098051775e-06,
+      "loss": 0.7215,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289970187119231e-06,
+      "loss": 0.8123,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28380654730898e-06,
+      "loss": 0.8189,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277643180974586e-06,
+      "loss": 0.7361,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271480090469512e-06,
+      "loss": 0.8231,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265317278147108e-06,
+      "loss": 0.7894,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259154746360624e-06,
+      "loss": 0.7546,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252992497463196e-06,
+      "loss": 0.7945,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246830533807857e-06,
+      "loss": 0.8087,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24066885774754e-06,
+      "loss": 0.8143,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234507471635043e-06,
+      "loss": 0.8016,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228346377823085e-06,
+      "loss": 0.8228,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22218557866424e-06,
+      "loss": 0.8023,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216025076511e-06,
+      "loss": 0.657,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209864873715726e-06,
+      "loss": 0.8471,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203704972630668e-06,
+      "loss": 0.7652,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197545375607964e-06,
+      "loss": 0.7759,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.191386084999632e-06,
+      "loss": 0.7041,
+      "step": 2831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185227103157573e-06,
+      "loss": 0.7944,
+      "step": 2832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179068432433576e-06,
+      "loss": 0.8066,
+      "step": 2833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172910075179304e-06,
+      "loss": 0.7285,
+      "step": 2834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166752033746303e-06,
+      "loss": 0.6658,
+      "step": 2835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160594310486002e-06,
+      "loss": 0.7178,
+      "step": 2836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1544369077497e-06,
+      "loss": 0.8095,
+      "step": 2837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148279827888586e-06,
+      "loss": 0.7388,
+      "step": 2838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14212307325371e-06,
+      "loss": 0.6926,
+      "step": 2839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135966646196008e-06,
+      "loss": 0.7283,
+      "step": 2840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129810549066298e-06,
+      "loss": 0.6678,
+      "step": 2841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12365478421525e-06,
+      "loss": 0.6884,
+      "step": 2842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117499353993437e-06,
+      "loss": 0.6851,
+      "step": 2843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111344260751269e-06,
+      "loss": 0.8087,
+      "step": 2844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.105189506839056e-06,
+      "loss": 0.6461,
+      "step": 2845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09903509460697e-06,
+      "loss": 0.8736,
+      "step": 2846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092881026405046e-06,
+      "loss": 0.7015,
+      "step": 2847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086727304583196e-06,
+      "loss": 0.6957,
+      "step": 2848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080573931491194e-06,
+      "loss": 0.7404,
+      "step": 2849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074420909478681e-06,
+      "loss": 0.672,
+      "step": 2850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068268240895172e-06,
+      "loss": 0.7753,
+      "step": 2851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062115928090036e-06,
+      "loss": 0.8453,
+      "step": 2852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055963973412514e-06,
+      "loss": 0.8055,
+      "step": 2853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04981237921171e-06,
+      "loss": 0.7148,
+      "step": 2854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043661147836578e-06,
+      "loss": 0.815,
+      "step": 2855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037510281635955e-06,
+      "loss": 0.8785,
+      "step": 2856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031359782958517e-06,
+      "loss": 0.8102,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025209654152814e-06,
+      "loss": 0.8349,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019059897567256e-06,
+      "loss": 0.7316,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01291051555009e-06,
+      "loss": 0.6752,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006761510449454e-06,
+      "loss": 0.8206,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000612884613306e-06,
+      "loss": 0.7866,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994464640389487e-06,
+      "loss": 0.7817,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98831678012568e-06,
+      "loss": 0.7897,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982169306169419e-06,
+      "loss": 0.7945,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9760222208681e-06,
+      "loss": 0.788,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96987552656896e-06,
+      "loss": 0.643,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963729225619094e-06,
+      "loss": 0.6459,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957583320365446e-06,
+      "loss": 0.7807,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951437813154806e-06,
+      "loss": 0.725,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94529270633381e-06,
+      "loss": 0.7783,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939148002248954e-06,
+      "loss": 0.7771,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93300370324656e-06,
+      "loss": 0.8047,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926859811672817e-06,
+      "loss": 0.7283,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920716329873737e-06,
+      "loss": 0.7881,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91457326019519e-06,
+      "loss": 0.7056,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908430604982891e-06,
+      "loss": 0.7719,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902288366582382e-06,
+      "loss": 0.7231,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89614654733906e-06,
+      "loss": 0.6944,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89000514959815e-06,
+      "loss": 0.6882,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883864175704728e-06,
+      "loss": 0.6727,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877723628003703e-06,
+      "loss": 0.6756,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871583508839815e-06,
+      "loss": 0.7059,
+      "step": 2883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865443820557653e-06,
+      "loss": 0.6879,
+      "step": 2884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859304565501628e-06,
+      "loss": 0.7995,
+      "step": 2885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853165746015997e-06,
+      "loss": 0.7986,
+      "step": 2886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847027364444844e-06,
+      "loss": 0.8347,
+      "step": 2887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.840889423132089e-06,
+      "loss": 0.7945,
+      "step": 2888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834751924421477e-06,
+      "loss": 0.6941,
+      "step": 2889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8286148706566e-06,
+      "loss": 0.742,
+      "step": 2890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822478264180856e-06,
+      "loss": 0.6575,
+      "step": 2891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816342107337501e-06,
+      "loss": 0.8178,
+      "step": 2892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81020640246959e-06,
+      "loss": 0.7238,
+      "step": 2893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804071151920028e-06,
+      "loss": 0.8439,
+      "step": 2894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797936358031538e-06,
+      "loss": 0.7445,
+      "step": 2895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791802023146664e-06,
+      "loss": 0.7945,
+      "step": 2896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785668149607784e-06,
+      "loss": 0.74,
+      "step": 2897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779534739757091e-06,
+      "loss": 0.751,
+      "step": 2898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773401795936609e-06,
+      "loss": 0.8316,
+      "step": 2899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76726932048818e-06,
+      "loss": 0.7608,
+      "step": 2900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761137315753464e-06,
+      "loss": 0.7937,
+      "step": 2901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755005784073948e-06,
+      "loss": 0.737,
+      "step": 2902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748874727790933e-06,
+      "loss": 0.7884,
+      "step": 2903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742744149245539e-06,
+      "loss": 0.8543,
+      "step": 2904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73661405077871e-06,
+      "loss": 0.839,
+      "step": 2905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730484434731198e-06,
+      "loss": 0.8148,
+      "step": 2906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724355303443573e-06,
+      "loss": 0.7667,
+      "step": 2907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718226659256231e-06,
+      "loss": 0.7308,
+      "step": 2908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712098504509358e-06,
+      "loss": 0.7211,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705970841542982e-06,
+      "loss": 0.7709,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699843672696914e-06,
+      "loss": 0.6819,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693717000310801e-06,
+      "loss": 0.775,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687590826724093e-06,
+      "loss": 0.7685,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681465154276037e-06,
+      "loss": 0.7799,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675339985305709e-06,
+      "loss": 0.8248,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669215322151974e-06,
+      "loss": 0.8223,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663091167153516e-06,
+      "loss": 0.7759,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656967522648822e-06,
+      "loss": 0.6813,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650844390976182e-06,
+      "loss": 0.7527,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644721774473694e-06,
+      "loss": 0.7523,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63859967547925e-06,
+      "loss": 0.8492,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632478096330559e-06,
+      "loss": 0.6965,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626357039365121e-06,
+      "loss": 0.8024,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620236506920236e-06,
+      "loss": 0.792,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61411650133301e-06,
+      "loss": 0.7295,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60799702494035e-06,
+      "loss": 0.9121,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601878080078942e-06,
+      "loss": 0.764,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595759669085299e-06,
+      "loss": 0.8068,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589641794295699e-06,
+      "loss": 0.7956,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58352445804624e-06,
+      "loss": 0.7439,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577407662672804e-06,
+      "loss": 0.8283,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571291410511063e-06,
+      "loss": 0.7145,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565175703896489e-06,
+      "loss": 0.6634,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559060545164338e-06,
+      "loss": 0.8053,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552945936649664e-06,
+      "loss": 0.7632,
+      "step": 2935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54683188068731e-06,
+      "loss": 0.7386,
+      "step": 2936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540718379611902e-06,
+      "loss": 0.6795,
+      "step": 2937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534605435757861e-06,
+      "loss": 0.7877,
+      "step": 2938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52849305145939e-06,
+      "loss": 0.7799,
+      "step": 2939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522381229050478e-06,
+      "loss": 0.7719,
+      "step": 2940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516269970864914e-06,
+      "loss": 0.7395,
+      "step": 2941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510159279236244e-06,
+      "loss": 0.8439,
+      "step": 2942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504049156497824e-06,
+      "loss": 0.6981,
+      "step": 2943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497939604982781e-06,
+      "loss": 0.8217,
+      "step": 2944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491830627024018e-06,
+      "loss": 0.7909,
+      "step": 2945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 0.7691,
+      "step": 2946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479614401105897e-06,
+      "loss": 0.7672,
+      "step": 2947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473507157811254e-06,
+      "loss": 0.7423,
+      "step": 2948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467400497402342e-06,
+      "loss": 0.7317,
+      "step": 2949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461294422210957e-06,
+      "loss": 0.6866,
+      "step": 2950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45518893456869e-06,
+      "loss": 0.7834,
+      "step": 2951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449084036806893e-06,
+      "loss": 0.8043,
+      "step": 2952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442979731256702e-06,
+      "loss": 0.7812,
+      "step": 2953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436876020249024e-06,
+      "loss": 0.728,
+      "step": 2954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430772906114538e-06,
+      "loss": 0.6929,
+      "step": 2955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424670391183692e-06,
+      "loss": 0.7449,
+      "step": 2956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418568477786718e-06,
+      "loss": 0.8772,
+      "step": 2957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412467168253599e-06,
+      "loss": 0.8134,
+      "step": 2958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406366464914111e-06,
+      "loss": 0.8252,
+      "step": 2959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40026637009777e-06,
+      "loss": 0.7859,
+      "step": 2960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394166886133886e-06,
+      "loss": 0.7853,
+      "step": 2961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388068015351521e-06,
+      "loss": 0.6515,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381969760079507e-06,
+      "loss": 0.6991,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37587212264644e-06,
+      "loss": 0.7649,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36977510538068e-06,
+      "loss": 0.7161,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363678710610352e-06,
+      "loss": 0.8144,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.357582940663344e-06,
+      "loss": 0.8367,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351487797867298e-06,
+      "loss": 0.7932,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345393284549628e-06,
+      "loss": 0.6852,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339299403037498e-06,
+      "loss": 0.7967,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333206155657833e-06,
+      "loss": 0.8258,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327113544737325e-06,
+      "loss": 0.735,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321021572602405e-06,
+      "loss": 0.6447,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314930241579276e-06,
+      "loss": 0.8735,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308839553993896e-06,
+      "loss": 0.7375,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302749512171956e-06,
+      "loss": 0.7251,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296660118438937e-06,
+      "loss": 0.7846,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290571375120032e-06,
+      "loss": 0.7763,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284483284540217e-06,
+      "loss": 0.7259,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278395849024206e-06,
+      "loss": 0.7411,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272309070896461e-06,
+      "loss": 0.7016,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2662229524812e-06,
+      "loss": 0.6801,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260137496102381e-06,
+      "loss": 0.774,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254052704083713e-06,
+      "loss": 0.738,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247968578748657e-06,
+      "loss": 0.8299,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241885122420409e-06,
+      "loss": 0.7359,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23580233742192e-06,
+      "loss": 0.8845,
+      "step": 2987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22972022607587e-06,
+      "loss": 0.7422,
+      "step": 2988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223638790704698e-06,
+      "loss": 0.8073,
+      "step": 2989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21755803363058e-06,
+      "loss": 0.7768,
+      "step": 2990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211477957175423e-06,
+      "loss": 0.8183,
+      "step": 2991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205398563660886e-06,
+      "loss": 0.7071,
+      "step": 2992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19931985540837e-06,
+      "loss": 0.6878,
+      "step": 2993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193241834738994e-06,
+      "loss": 0.6598,
+      "step": 2994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.187164503973643e-06,
+      "loss": 0.6226,
+      "step": 2995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18108786543291e-06,
+      "loss": 0.7467,
+      "step": 2996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175011921437145e-06,
+      "loss": 0.8077,
+      "step": 2997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168936674306425e-06,
+      "loss": 0.7552,
+      "step": 2998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162862126360557e-06,
+      "loss": 0.767,
+      "step": 2999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15678827991909e-06,
+      "loss": 0.8068,
+      "step": 3000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150715137301295e-06,
+      "loss": 0.815,
+      "step": 3001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144642700826182e-06,
+      "loss": 0.789,
+      "step": 3002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13857097281249e-06,
+      "loss": 0.7356,
+      "step": 3003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132499955578682e-06,
+      "loss": 0.6956,
+      "step": 3004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12642965144296e-06,
+      "loss": 0.6948,
+      "step": 3005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12036006272324e-06,
+      "loss": 0.7219,
+      "step": 3006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114291191737176e-06,
+      "loss": 0.7931,
+      "step": 3007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10822304080215e-06,
+      "loss": 0.6981,
+      "step": 3008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102155612235252e-06,
+      "loss": 0.8263,
+      "step": 3009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096088908353316e-06,
+      "loss": 0.7769,
+      "step": 3010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09002293147289e-06,
+      "loss": 0.7779,
+      "step": 3011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08395768391024e-06,
+      "loss": 0.6708,
+      "step": 3012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077893167981364e-06,
+      "loss": 0.8418,
+      "step": 3013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071829386001971e-06,
+      "loss": 0.9258,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065766340287498e-06,
+      "loss": 0.8349,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059704033153096e-06,
+      "loss": 0.7894,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053642466913631e-06,
+      "loss": 0.8609,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047581643883698e-06,
+      "loss": 0.7937,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041521566377591e-06,
+      "loss": 0.7481,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035462236709331e-06,
+      "loss": 0.7681,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029403657192657e-06,
+      "loss": 0.7327,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02334583014101e-06,
+      "loss": 0.7807,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01728875786755e-06,
+      "loss": 0.7607,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011232442685146e-06,
+      "loss": 0.7424,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005176886906381e-06,
+      "loss": 0.8471,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999122092843554e-06,
+      "loss": 0.7898,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993068062808654e-06,
+      "loss": 0.7694,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987014799113398e-06,
+      "loss": 0.8059,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980962304069203e-06,
+      "loss": 0.794,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974910579987189e-06,
+      "loss": 0.7843,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968859629178188e-06,
+      "loss": 0.7558,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96280945395273e-06,
+      "loss": 0.7742,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956760056621051e-06,
+      "loss": 0.633,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950711439493097e-06,
+      "loss": 0.8014,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944663604878507e-06,
+      "loss": 0.7401,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938616555086627e-06,
+      "loss": 0.6708,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932570292426496e-06,
+      "loss": 0.7716,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92652481920686e-06,
+      "loss": 0.7036,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920480137736163e-06,
+      "loss": 0.7824,
+      "step": 3039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91443625032254e-06,
+      "loss": 0.8062,
+      "step": 3040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908393159273835e-06,
+      "loss": 0.7278,
+      "step": 3041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902350866897573e-06,
+      "loss": 0.7919,
+      "step": 3042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896309375500982e-06,
+      "loss": 0.7303,
+      "step": 3043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89026868739099e-06,
+      "loss": 0.726,
+      "step": 3044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8842288048742e-06,
+      "loss": 0.7927,
+      "step": 3045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878189730256928e-06,
+      "loss": 0.7477,
+      "step": 3046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872151465845175e-06,
+      "loss": 0.7148,
+      "step": 3047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866114013944623e-06,
+      "loss": 0.7582,
+      "step": 3048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860077376860657e-06,
+      "loss": 0.7972,
+      "step": 3049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854041556898337e-06,
+      "loss": 0.6722,
+      "step": 3050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848006556362426e-06,
+      "loss": 0.7539,
+      "step": 3051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841972377557366e-06,
+      "loss": 0.7184,
+      "step": 3052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835939022787283e-06,
+      "loss": 0.7024,
+      "step": 3053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829906494355995e-06,
+      "loss": 0.8663,
+      "step": 3054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823874794566997e-06,
+      "loss": 0.7268,
+      "step": 3055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817843925723474e-06,
+      "loss": 0.8056,
+      "step": 3056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.811813890128294e-06,
+      "loss": 0.8071,
+      "step": 3057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805784690083999e-06,
+      "loss": 0.6591,
+      "step": 3058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799756327892821e-06,
+      "loss": 0.7895,
+      "step": 3059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793728805856663e-06,
+      "loss": 0.7951,
+      "step": 3060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787702126277114e-06,
+      "loss": 0.7195,
+      "step": 3061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78167629145545e-06,
+      "loss": 0.7995,
+      "step": 3062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.775651303692598e-06,
+      "loss": 0.6606,
+      "step": 3063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769627165289186e-06,
+      "loss": 0.8197,
+      "step": 3064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76360387854551e-06,
+      "loss": 0.6868,
+      "step": 3065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757581445761537e-06,
+      "loss": 0.6907,
+      "step": 3066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.751559869236915e-06,
+      "loss": 0.7761,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745539151270957e-06,
+      "loss": 0.7918,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739519294162652e-06,
+      "loss": 0.7826,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733500300210665e-06,
+      "loss": 0.7347,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727482171713322e-06,
+      "loss": 0.8481,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721464910968628e-06,
+      "loss": 0.7319,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715448520274245e-06,
+      "loss": 0.7328,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709433001927515e-06,
+      "loss": 0.7014,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703418358225445e-06,
+      "loss": 0.7953,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697404591464697e-06,
+      "loss": 0.7223,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691391703941614e-06,
+      "loss": 0.6848,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685379697952186e-06,
+      "loss": 0.746,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679368575792081e-06,
+      "loss": 0.8138,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673358339756628e-06,
+      "loss": 0.7969,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667348992140806e-06,
+      "loss": 0.6486,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661340535239266e-06,
+      "loss": 0.7932,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655332971346317e-06,
+      "loss": 0.8312,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64932630275592e-06,
+      "loss": 0.7256,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643320531761707e-06,
+      "loss": 0.8001,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637315660656954e-06,
+      "loss": 0.6583,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631311691734599e-06,
+      "loss": 0.6921,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6253086272872404e-06,
+      "loss": 0.8021,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6193064696071216e-06,
+      "loss": 0.8336,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613305220986149e-06,
+      "loss": 0.7349,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6073048837158725e-06,
+      "loss": 0.6869,
+      "step": 3091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6013054600875005e-06,
+      "loss": 0.8259,
+      "step": 3092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595306952391897e-06,
+      "loss": 0.8208,
+      "step": 3093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58930936291956e-06,
+      "loss": 0.8006,
+      "step": 3094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583312693960659e-06,
+      "loss": 0.7772,
+      "step": 3095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577316947804988e-06,
+      "loss": 0.7813,
+      "step": 3096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5713221267420075e-06,
+      "loss": 0.7647,
+      "step": 3097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56532823306082e-06,
+      "loss": 0.7379,
+      "step": 3098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559335269050166e-06,
+      "loss": 0.7998,
+      "step": 3099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553343236998442e-06,
+      "loss": 0.8024,
+      "step": 3100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547352139193682e-06,
+      "loss": 0.6432,
+      "step": 3101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541361977923564e-06,
+      "loss": 0.699,
+      "step": 3102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535372755475411e-06,
+      "loss": 0.8608,
+      "step": 3103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529384474136182e-06,
+      "loss": 0.7128,
+      "step": 3104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523397136192483e-06,
+      "loss": 0.6515,
+      "step": 3105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51741074393056e-06,
+      "loss": 0.7284,
+      "step": 3106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511425299636289e-06,
+      "loss": 0.7688,
+      "step": 3107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505440805595196e-06,
+      "loss": 0.7574,
+      "step": 3108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499457264092433e-06,
+      "loss": 0.8038,
+      "step": 3109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 0.7039,
+      "step": 3110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487493047840717e-06,
+      "loss": 0.797,
+      "step": 3111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481512377660251e-06,
+      "loss": 0.7074,
+      "step": 3112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475532669155106e-06,
+      "loss": 0.8716,
+      "step": 3113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469553924608601e-06,
+      "loss": 0.711,
+      "step": 3114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463576146303705e-06,
+      "loss": 0.7647,
+      "step": 3115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457599336523011e-06,
+      "loss": 0.791,
+      "step": 3116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.451623497548737e-06,
+      "loss": 0.8853,
+      "step": 3117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44564863166274e-06,
+      "loss": 0.7988,
+      "step": 3118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4396747411465e-06,
+      "loss": 0.8191,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4337018282811225e-06,
+      "loss": 0.6927,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427729895347347e-06,
+      "loss": 0.7436,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421758944625528e-06,
+      "loss": 0.7979,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415788978395654e-06,
+      "loss": 0.7921,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4098199989373424e-06,
+      "loss": 0.7852,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403852008529811e-06,
+      "loss": 0.764,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397885009451933e-06,
+      "loss": 0.7988,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391919003982168e-06,
+      "loss": 0.7084,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385953994398624e-06,
+      "loss": 0.8077,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379989982979017e-06,
+      "loss": 0.7303,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374026972000681e-06,
+      "loss": 0.6945,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3680649637405735e-06,
+      "loss": 0.7602,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362103960475258e-06,
+      "loss": 0.6333,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356143964480926e-06,
+      "loss": 0.8458,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350184978033386e-06,
+      "loss": 0.6975,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344227003408045e-06,
+      "loss": 0.7648,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33827004287994e-06,
+      "loss": 0.7376,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332314098723717e-06,
+      "loss": 0.7133,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326359173213625e-06,
+      "loss": 0.6959,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320405268623536e-06,
+      "loss": 0.7961,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314452387226922e-06,
+      "loss": 0.7328,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308500531296871e-06,
+      "loss": 0.7583,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302549703106084e-06,
+      "loss": 0.7587,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296599904926849e-06,
+      "loss": 0.7832,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290651139031091e-06,
+      "loss": 0.8499,
+      "step": 3144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28470340769031e-06,
+      "loss": 0.6629,
+      "step": 3145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278756713175632e-06,
+      "loss": 0.7787,
+      "step": 3146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272811057757785e-06,
+      "loss": 0.7931,
+      "step": 3147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266866443707088e-06,
+      "loss": 0.6482,
+      "step": 3148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260922873293477e-06,
+      "loss": 0.7514,
+      "step": 3149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2549803487864755e-06,
+      "loss": 0.6972,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24903887245522e-06,
+      "loss": 0.86,
+      "step": 3151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243098446568442e-06,
+      "loss": 0.7881,
+      "step": 3152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237159073394467e-06,
+      "loss": 0.7291,
+      "step": 3153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231220755201225e-06,
+      "loss": 0.8422,
+      "step": 3154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225283494256246e-06,
+      "loss": 0.8416,
+      "step": 3155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219347292826644e-06,
+      "loss": 0.6497,
+      "step": 3156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2134121531791405e-06,
+      "loss": 0.769,
+      "step": 3157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207478077580043e-06,
+      "loss": 0.7721,
+      "step": 3158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20154506829526e-06,
+      "loss": 0.7354,
+      "step": 3159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195613127590292e-06,
+      "loss": 0.825,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18968225773022e-06,
+      "loss": 0.7042,
+      "step": 3161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183752460979737e-06,
+      "loss": 0.7155,
+      "step": 3162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177823739603102e-06,
+      "loss": 0.7472,
+      "step": 3163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171896095864183e-06,
+      "loss": 0.7282,
+      "step": 3164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16596953202643e-06,
+      "loss": 0.7677,
+      "step": 3165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160044050352874e-06,
+      "loss": 0.7455,
+      "step": 3166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154119653106147e-06,
+      "loss": 0.734,
+      "step": 3167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148196342548449e-06,
+      "loss": 0.6868,
+      "step": 3168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142274120941579e-06,
+      "loss": 0.7542,
+      "step": 3169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13635299054692e-06,
+      "loss": 0.6818,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130432953625425e-06,
+      "loss": 0.7997,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124514012437645e-06,
+      "loss": 0.6125,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118596169243706e-06,
+      "loss": 0.9324,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112679426303312e-06,
+      "loss": 0.7719,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106763785875752e-06,
+      "loss": 0.748,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100849250219889e-06,
+      "loss": 0.8062,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094935821594166e-06,
+      "loss": 0.8486,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089023502256613e-06,
+      "loss": 0.7967,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083112294464816e-06,
+      "loss": 0.7683,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077202200475959e-06,
+      "loss": 0.7447,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071293222546779e-06,
+      "loss": 0.728,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065385362933603e-06,
+      "loss": 0.8051,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0594786238923306e-06,
+      "loss": 0.7415,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053573007678421e-06,
+      "loss": 0.6709,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047668516546915e-06,
+      "loss": 0.7132,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041765152752419e-06,
+      "loss": 0.7835,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035862918549113e-06,
+      "loss": 0.7821,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029961816190747e-06,
+      "loss": 0.7129,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024061847930628e-06,
+      "loss": 0.7507,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01816301602164e-06,
+      "loss": 0.7701,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012265322716237e-06,
+      "loss": 0.8773,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006368770266421e-06,
+      "loss": 0.6775,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0004733609237784e-06,
+      "loss": 0.7745,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994579096939442e-06,
+      "loss": 0.7737,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988685980564121e-06,
+      "loss": 0.71,
+      "step": 3195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9827940140480776e-06,
+      "loss": 0.7424,
+      "step": 3196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976903199641138e-06,
+      "loss": 0.7253,
+      "step": 3197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97101353959269e-06,
+      "loss": 0.8196,
+      "step": 3198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965125036151675e-06,
+      "loss": 0.7549,
+      "step": 3199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9592376915666005e-06,
+      "loss": 0.799,
+      "step": 3200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953351508085527e-06,
+      "loss": 0.7941,
+      "step": 3201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947466487956067e-06,
+      "loss": 0.7704,
+      "step": 3202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941582633425403e-06,
+      "loss": 0.8222,
+      "step": 3203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935699946740255e-06,
+      "loss": 0.7258,
+      "step": 3204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929818430146909e-06,
+      "loss": 0.7887,
+      "step": 3205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923938085891202e-06,
+      "loss": 0.809,
+      "step": 3206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91805891621852e-06,
+      "loss": 0.716,
+      "step": 3207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9121809233738e-06,
+      "loss": 0.7911,
+      "step": 3208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906304109601542e-06,
+      "loss": 0.8083,
+      "step": 3209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900428477145773e-06,
+      "loss": 0.6777,
+      "step": 3210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894554028250093e-06,
+      "loss": 0.7389,
+      "step": 3211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88868076515763e-06,
+      "loss": 0.793,
+      "step": 3212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882808690111073e-06,
+      "loss": 0.7146,
+      "step": 3213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876937805352653e-06,
+      "loss": 0.7838,
+      "step": 3214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871068113124142e-06,
+      "loss": 0.7163,
+      "step": 3215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8651996156668656e-06,
+      "loss": 0.7452,
+      "step": 3216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859332315221683e-06,
+      "loss": 0.7412,
+      "step": 3217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853466214029004e-06,
+      "loss": 0.7287,
+      "step": 3218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84760131432878e-06,
+      "loss": 0.7676,
+      "step": 3219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841737618360498e-06,
+      "loss": 0.7864,
+      "step": 3220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8358751283631896e-06,
+      "loss": 0.7847,
+      "step": 3221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83001384657543e-06,
+      "loss": 0.769,
+      "step": 3222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8241537752353205e-06,
+      "loss": 0.8497,
+      "step": 3223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818294916580513e-06,
+      "loss": 0.6916,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812437272848189e-06,
+      "loss": 0.8153,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806580846275068e-06,
+      "loss": 0.7325,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800725639097412e-06,
+      "loss": 0.7531,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7948716535509995e-06,
+      "loss": 0.8177,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7890188918711655e-06,
+      "loss": 0.6774,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.783167356292753e-06,
+      "loss": 0.7811,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7773170490501585e-06,
+      "loss": 0.8171,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7714679723772996e-06,
+      "loss": 0.769,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7656201285076195e-06,
+      "loss": 0.8394,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759773519674103e-06,
+      "loss": 0.8098,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753928148109252e-06,
+      "loss": 0.7164,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7480840160451e-06,
+      "loss": 0.6982,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7422411257132115e-06,
+      "loss": 0.7804,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736399479344669e-06,
+      "loss": 0.7614,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730559079170082e-06,
+      "loss": 0.7334,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7247199274195966e-06,
+      "loss": 0.8449,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718882026322857e-06,
+      "loss": 0.6533,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713045378109058e-06,
+      "loss": 0.7109,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707209985006893e-06,
+      "loss": 0.7507,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701375849244585e-06,
+      "loss": 0.7177,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695542973049889e-06,
+      "loss": 0.7087,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689711358650052e-06,
+      "loss": 0.7006,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683881008271869e-06,
+      "loss": 0.8432,
+      "step": 3247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678051924141625e-06,
+      "loss": 0.8347,
+      "step": 3248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672224108485143e-06,
+      "loss": 0.7183,
+      "step": 3249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666397563527753e-06,
+      "loss": 0.6941,
+      "step": 3250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660572291494297e-06,
+      "loss": 0.6115,
+      "step": 3251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654748294609137e-06,
+      "loss": 0.6345,
+      "step": 3252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648925575096139e-06,
+      "loss": 0.7027,
+      "step": 3253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643104135178693e-06,
+      "loss": 0.7805,
+      "step": 3254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637283977079694e-06,
+      "loss": 0.7346,
+      "step": 3255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6314651030215435e-06,
+      "loss": 0.7838,
+      "step": 3256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62564751522616e-06,
+      "loss": 0.8249,
+      "step": 3257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619831215914974e-06,
+      "loss": 0.7588,
+      "step": 3258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614016207308908e-06,
+      "loss": 0.7537,
+      "step": 3259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608202491628412e-06,
+      "loss": 0.6806,
+      "step": 3260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602390071093423e-06,
+      "loss": 0.7373,
+      "step": 3261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596578947923395e-06,
+      "loss": 0.6888,
+      "step": 3262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590769124337291e-06,
+      "loss": 0.7319,
+      "step": 3263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584960602553563e-06,
+      "loss": 0.7178,
+      "step": 3264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5791533847901775e-06,
+      "loss": 0.7456,
+      "step": 3265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573347473264598e-06,
+      "loss": 0.8482,
+      "step": 3266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56754287019379e-06,
+      "loss": 0.8265,
+      "step": 3267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561739577794224e-06,
+      "loss": 0.7812,
+      "step": 3268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555937598281861e-06,
+      "loss": 0.6401,
+      "step": 3269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550136933872169e-06,
+      "loss": 0.6888,
+      "step": 3270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544337586780109e-06,
+      "loss": 0.6271,
+      "step": 3271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538539559220141e-06,
+      "loss": 0.734,
+      "step": 3272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5327428534062205e-06,
+      "loss": 0.7011,
+      "step": 3273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 0.7977,
+      "step": 3274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521153415869819e-06,
+      "loss": 0.7801,
+      "step": 3275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515360688572727e-06,
+      "loss": 0.6823,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5095692918724465e-06,
+      "loss": 0.7273,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503779227980409e-06,
+      "loss": 0.6887,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497990499107518e-06,
+      "loss": 0.682,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49220310746419e-06,
+      "loss": 0.7884,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486417055260314e-06,
+      "loss": 0.672,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480632344705274e-06,
+      "loss": 0.7759,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474848978007942e-06,
+      "loss": 0.8371,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469066957376672e-06,
+      "loss": 0.6848,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463286285019309e-06,
+      "loss": 0.8083,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4575069631431845e-06,
+      "loss": 0.6948,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451728993955107e-06,
+      "loss": 0.786,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445952379661377e-06,
+      "loss": 0.8001,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440177122467769e-06,
+      "loss": 0.7412,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434403224579546e-06,
+      "loss": 0.6905,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428630688201453e-06,
+      "loss": 0.6498,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422859515537709e-06,
+      "loss": 0.5451,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4170897087920125e-06,
+      "loss": 0.7583,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411321270167552e-06,
+      "loss": 0.6741,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4055542018669724e-06,
+      "loss": 0.7302,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.399788506092421e-06,
+      "loss": 0.8079,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394024185045495e-06,
+      "loss": 0.7073,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38826124092729e-06,
+      "loss": 0.7926,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382499675938358e-06,
+      "loss": 0.6384,
+      "step": 3299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376739492278735e-06,
+      "loss": 0.693,
+      "step": 3300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370980692147927e-06,
+      "loss": 0.6802,
+      "step": 3301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365223277744907e-06,
+      "loss": 0.7014,
+      "step": 3302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359467251268125e-06,
+      "loss": 0.7193,
+      "step": 3303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3537126149154995e-06,
+      "loss": 0.8281,
+      "step": 3304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347959370884416e-06,
+      "loss": 0.7212,
+      "step": 3305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34220752137173e-06,
+      "loss": 0.729,
+      "step": 3306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336457068573765e-06,
+      "loss": 0.6996,
+      "step": 3307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330708014686307e-06,
+      "loss": 0.8089,
+      "step": 3308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3249603619046194e-06,
+      "loss": 0.6529,
+      "step": 3309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319214112423412e-06,
+      "loss": 0.7516,
+      "step": 3310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3134692684368756e-06,
+      "loss": 0.664,
+      "step": 3311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3077258321386604e-06,
+      "loss": 0.8128,
+      "step": 3312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30198380572187e-06,
+      "loss": 0.6921,
+      "step": 3313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296243191379081e-06,
+      "loss": 0.6819,
+      "step": 3314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290503991302324e-06,
+      "loss": 0.7849,
+      "step": 3315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284766207683089e-06,
+      "loss": 0.6734,
+      "step": 3316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279029842712334e-06,
+      "loss": 0.6426,
+      "step": 3317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2732948985804645e-06,
+      "loss": 0.6878,
+      "step": 3318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267561377477351e-06,
+      "loss": 0.6882,
+      "step": 3319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261829281592313e-06,
+      "loss": 0.761,
+      "step": 3320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256098613114134e-06,
+      "loss": 0.7703,
+      "step": 3321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25036937423105e-06,
+      "loss": 0.7932,
+      "step": 3322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244641567130747e-06,
+      "loss": 0.6128,
+      "step": 3323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238915194000369e-06,
+      "loss": 0.8425,
+      "step": 3324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233190257026509e-06,
+      "loss": 0.755,
+      "step": 3325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227466758395211e-06,
+      "loss": 0.7487,
+      "step": 3326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2217447002919824e-06,
+      "loss": 0.7809,
+      "step": 3327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216024084901757e-06,
+      "loss": 0.6851,
+      "step": 3328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21030491440894e-06,
+      "loss": 0.7352,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204587190997374e-06,
+      "loss": 0.8003,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198870916850348e-06,
+      "loss": 0.7479,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1931560941506055e-06,
+      "loss": 0.7893,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187442725080323e-06,
+      "loss": 0.7317,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181730811821136e-06,
+      "loss": 0.7234,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1760203565541175e-06,
+      "loss": 0.7631,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1703113614597805e-06,
+      "loss": 0.8313,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1646038287180876e-06,
+      "loss": 0.7115,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1588977605084365e-06,
+      "loss": 0.7647,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15319315900967e-06,
+      "loss": 0.7964,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147490026400069e-06,
+      "loss": 0.7521,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141788364857353e-06,
+      "loss": 0.7551,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136088176558683e-06,
+      "loss": 0.6403,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130389463680653e-06,
+      "loss": 0.7351,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124692228399294e-06,
+      "loss": 0.6982,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118996472890082e-06,
+      "loss": 0.7543,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113302199327911e-06,
+      "loss": 0.7949,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107609409887124e-06,
+      "loss": 0.7033,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101918106741494e-06,
+      "loss": 0.8124,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096228292064218e-06,
+      "loss": 0.7623,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090539968027937e-06,
+      "loss": 0.776,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084853136804711e-06,
+      "loss": 0.7495,
+      "step": 3351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07916780056604e-06,
+      "loss": 0.7856,
+      "step": 3352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07348396148285e-06,
+      "loss": 0.7208,
+      "step": 3353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067801621725488e-06,
+      "loss": 0.692,
+      "step": 3354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062120783463743e-06,
+      "loss": 0.7381,
+      "step": 3355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056441448866817e-06,
+      "loss": 0.726,
+      "step": 3356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050763620103342e-06,
+      "loss": 0.6939,
+      "step": 3357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045087299341381e-06,
+      "loss": 0.7056,
+      "step": 3358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039412488748411e-06,
+      "loss": 0.7483,
+      "step": 3359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033739190491343e-06,
+      "loss": 0.8127,
+      "step": 3360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0280674067365016e-06,
+      "loss": 0.7751,
+      "step": 3361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022397139649636e-06,
+      "loss": 0.692,
+      "step": 3362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016728391395925e-06,
+      "loss": 0.7712,
+      "step": 3363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0110611641399486e-06,
+      "loss": 0.7282,
+      "step": 3364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005395460045722e-06,
+      "loss": 0.7594,
+      "step": 3365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999731281276675e-06,
+      "loss": 0.7605,
+      "step": 3366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994068629995651e-06,
+      "loss": 0.7794,
+      "step": 3367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988407508364915e-06,
+      "loss": 0.6878,
+      "step": 3368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982747918546144e-06,
+      "loss": 0.6607,
+      "step": 3369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977089862700432e-06,
+      "loss": 0.731,
+      "step": 3370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971433342988287e-06,
+      "loss": 0.7859,
+      "step": 3371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96577836156963e-06,
+      "loss": 0.7194,
+      "step": 3372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960124920603798e-06,
+      "loss": 0.7993,
+      "step": 3373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954473022249532e-06,
+      "loss": 0.7407,
+      "step": 3374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948822668664989e-06,
+      "loss": 0.7588,
+      "step": 3375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943173862007743e-06,
+      "loss": 0.7635,
+      "step": 3376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93752660443476e-06,
+      "loss": 0.6547,
+      "step": 3377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931880898102437e-06,
+      "loss": 0.7241,
+      "step": 3378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926236745166553e-06,
+      "loss": 0.6208,
+      "step": 3379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920594147782315e-06,
+      "loss": 0.8093,
+      "step": 3380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914953108104329e-06,
+      "loss": 0.6412,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9093136282866014e-06,
+      "loss": 0.7115,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90367571048255e-06,
+      "loss": 0.8314,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898039356844993e-06,
+      "loss": 0.6749,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8924045695261486e-06,
+      "loss": 0.7776,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886771350677642e-06,
+      "loss": 0.6975,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8811397024504976e-06,
+      "loss": 0.6477,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875509626995139e-06,
+      "loss": 0.7536,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869881126461392e-06,
+      "loss": 0.7184,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8642542029984736e-06,
+      "loss": 0.8491,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858628858755011e-06,
+      "loss": 0.753,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853005095879015e-06,
+      "loss": 0.6909,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847382916517901e-06,
+      "loss": 0.7954,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841762322818483e-06,
+      "loss": 0.6404,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8361433169269555e-06,
+      "loss": 0.7186,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830525900988925e-06,
+      "loss": 0.6571,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824910077149372e-06,
+      "loss": 0.6803,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819295847552682e-06,
+      "loss": 0.733,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813683214342629e-06,
+      "loss": 0.7831,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808072179662375e-06,
+      "loss": 0.6471,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802462745654475e-06,
+      "loss": 0.7032,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796854914460873e-06,
+      "loss": 0.6966,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791248688222889e-06,
+      "loss": 0.8338,
+      "step": 3403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785644069081256e-06,
+      "loss": 0.6296,
+      "step": 3404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780041059176061e-06,
+      "loss": 0.7302,
+      "step": 3405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774439660646802e-06,
+      "loss": 0.8237,
+      "step": 3406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768839875632351e-06,
+      "loss": 0.6922,
+      "step": 3407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7632417062709654e-06,
+      "loss": 0.6985,
+      "step": 3408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757645154700291e-06,
+      "loss": 0.7694,
+      "step": 3409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75205022305734e-06,
+      "loss": 0.8897,
+      "step": 3410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746456913478522e-06,
+      "loss": 0.6828,
+      "step": 3411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740865228099621e-06,
+      "loss": 0.6616,
+      "step": 3412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7352751690558025e-06,
+      "loss": 0.7075,
+      "step": 3413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729686738481616e-06,
+      "loss": 0.7503,
+      "step": 3414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724099938510972e-06,
+      "loss": 0.7321,
+      "step": 3415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718514771277175e-06,
+      "loss": 0.7379,
+      "step": 3416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7129312389129025e-06,
+      "loss": 0.7707,
+      "step": 3417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707349343550203e-06,
+      "loss": 0.6958,
+      "step": 3418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701769087320506e-06,
+      "loss": 0.7642,
+      "step": 3419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.696190472354613e-06,
+      "loss": 0.692,
+      "step": 3420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690613500782687e-06,
+      "loss": 0.7343,
+      "step": 3421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68503817473429e-06,
+      "loss": 0.7437,
+      "step": 3422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679464496338328e-06,
+      "loss": 0.8853,
+      "step": 3423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673892467723094e-06,
+      "loss": 0.8028,
+      "step": 3424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668322091016249e-06,
+      "loss": 0.7425,
+      "step": 3425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662753368344817e-06,
+      "loss": 0.6401,
+      "step": 3426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6571863018352e-06,
+      "loss": 0.6929,
+      "step": 3427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651620893613156e-06,
+      "loss": 0.7542,
+      "step": 3428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646057145803817e-06,
+      "loss": 0.847,
+      "step": 3429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640495060531681e-06,
+      "loss": 0.7149,
+      "step": 3430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6349346399206105e-06,
+      "loss": 0.759,
+      "step": 3431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629375886093835e-06,
+      "loss": 0.6696,
+      "step": 3432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623818801173937e-06,
+      "loss": 0.7515,
+      "step": 3433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6182633872828695e-06,
+      "loss": 0.6281,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612709646541957e-06,
+      "loss": 0.7336,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6071575810718644e-06,
+      "loss": 0.7639,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601607192992633e-06,
+      "loss": 0.7184,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 0.7361,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590511457483689e-06,
+      "loss": 0.6802,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584966114290849e-06,
+      "loss": 0.746,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579422456962593e-06,
+      "loss": 0.7028,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573880487615755e-06,
+      "loss": 0.8142,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.568340208366513e-06,
+      "loss": 0.7606,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562801621330402e-06,
+      "loss": 0.6981,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557264728622318e-06,
+      "loss": 0.7292,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551729532356494e-06,
+      "loss": 0.7323,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546196034646527e-06,
+      "loss": 0.762,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540664237605365e-06,
+      "loss": 0.7116,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5351341433453044e-06,
+      "loss": 0.7085,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529605753977993e-06,
+      "loss": 0.8586,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524079071614421e-06,
+      "loss": 0.7164,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518554098364932e-06,
+      "loss": 0.8297,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513030836339226e-06,
+      "loss": 0.7637,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507509287646331e-06,
+      "loss": 0.7983,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501989454394634e-06,
+      "loss": 0.7547,
+      "step": 3455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496471338691862e-06,
+      "loss": 0.6313,
+      "step": 3456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490954942645088e-06,
+      "loss": 0.7449,
+      "step": 3457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485440268360731e-06,
+      "loss": 0.606,
+      "step": 3458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479927317944542e-06,
+      "loss": 0.7645,
+      "step": 3459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474416093501627e-06,
+      "loss": 0.6247,
+      "step": 3460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468906597136423e-06,
+      "loss": 0.6658,
+      "step": 3461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463398830952714e-06,
+      "loss": 0.6752,
+      "step": 3462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457892797053623e-06,
+      "loss": 0.6882,
+      "step": 3463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452388497541603e-06,
+      "loss": 0.6399,
+      "step": 3464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446885934518451e-06,
+      "loss": 0.5934,
+      "step": 3465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441385110085304e-06,
+      "loss": 0.6594,
+      "step": 3466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4358860263426314e-06,
+      "loss": 0.7642,
+      "step": 3467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430388685390236e-06,
+      "loss": 0.8333,
+      "step": 3468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424893089327263e-06,
+      "loss": 0.7954,
+      "step": 3469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419399240252174e-06,
+      "loss": 0.7762,
+      "step": 3470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413907140262791e-06,
+      "loss": 0.7498,
+      "step": 3471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408416791456239e-06,
+      "loss": 0.6964,
+      "step": 3472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402928195928994e-06,
+      "loss": 0.8244,
+      "step": 3473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397441355776855e-06,
+      "loss": 0.7475,
+      "step": 3474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391956273094952e-06,
+      "loss": 0.7205,
+      "step": 3475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386472949977745e-06,
+      "loss": 0.7215,
+      "step": 3476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380991388519019e-06,
+      "loss": 0.7593,
+      "step": 3477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375511590811888e-06,
+      "loss": 0.6234,
+      "step": 3478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370033558948793e-06,
+      "loss": 0.82,
+      "step": 3479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364557295021502e-06,
+      "loss": 0.6733,
+      "step": 3480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3590828011211095e-06,
+      "loss": 0.6818,
+      "step": 3481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3536100793380234e-06,
+      "loss": 0.7015,
+      "step": 3482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34813913176198e-06,
+      "loss": 0.671,
+      "step": 3483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3426699604820574e-06,
+      "loss": 0.728,
+      "step": 3484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337202567586624e-06,
+      "loss": 0.7956,
+      "step": 3485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331736955163387e-06,
+      "loss": 0.6705,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326273125299377e-06,
+      "loss": 0.793,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320811080080923e-06,
+      "loss": 0.858,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315350821593706e-06,
+      "loss": 0.7999,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3098923519226904e-06,
+      "loss": 0.6857,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304435673152179e-06,
+      "loss": 0.7687,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298980787365785e-06,
+      "loss": 0.646,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293527696646435e-06,
+      "loss": 0.8314,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288076403076377e-06,
+      "loss": 0.6495,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282626908737161e-06,
+      "loss": 0.8029,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277179215709657e-06,
+      "loss": 0.6119,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271733326074049e-06,
+      "loss": 0.7738,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2662892419098296e-06,
+      "loss": 0.7325,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2608469652958065e-06,
+      "loss": 0.7629,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255406498310085e-06,
+      "loss": 0.7497,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249967843030088e-06,
+      "loss": 0.7758,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244531001532558e-06,
+      "loss": 0.791,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239095975893522e-06,
+      "loss": 0.8307,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233662768188327e-06,
+      "loss": 0.8198,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228231380491624e-06,
+      "loss": 0.8128,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22280181487737e-06,
+      "loss": 0.7193,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217374073418827e-06,
+      "loss": 0.6749,
+      "step": 3507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211948158188551e-06,
+      "loss": 0.786,
+      "step": 3508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2065240712584135e-06,
+      "loss": 0.7536,
+      "step": 3509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201101814699578e-06,
+      "loss": 0.6737,
+      "step": 3510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195681390582519e-06,
+      "loss": 0.8175,
+      "step": 3511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190262800977007e-06,
+      "loss": 0.7169,
+      "step": 3512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184846047952101e-06,
+      "loss": 0.6937,
+      "step": 3513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179431133576174e-06,
+      "loss": 0.765,
+      "step": 3514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174018059916892e-06,
+      "loss": 0.7176,
+      "step": 3515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1686068290412175e-06,
+      "loss": 0.7553,
+      "step": 3516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163197443015411e-06,
+      "loss": 0.6903,
+      "step": 3517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157789903905022e-06,
+      "loss": 0.7873,
+      "step": 3518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152384213774897e-06,
+      "loss": 0.7429,
+      "step": 3519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146980374689192e-06,
+      "loss": 0.6445,
+      "step": 3520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14157838871133e-06,
+      "loss": 0.7947,
+      "step": 3521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136178257904048e-06,
+      "loss": 0.7915,
+      "step": 3522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130779984329361e-06,
+      "loss": 0.8696,
+      "step": 3523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125383570048583e-06,
+      "loss": 0.7093,
+      "step": 3524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119989017122318e-06,
+      "loss": 0.7595,
+      "step": 3525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114596327610448e-06,
+      "loss": 0.6713,
+      "step": 3526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1092055035721575e-06,
+      "loss": 0.7281,
+      "step": 3527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103816547065912e-06,
+      "loss": 0.7242,
+      "step": 3528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098429460149466e-06,
+      "loss": 0.7503,
+      "step": 3529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09304424487986e-06,
+      "loss": 0.7844,
+      "step": 3530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087660903313414e-06,
+      "loss": 0.6642,
+      "step": 3531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082279437505739e-06,
+      "loss": 0.7458,
+      "step": 3532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076899849511731e-06,
+      "loss": 0.7558,
+      "step": 3533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071522141385563e-06,
+      "loss": 0.7066,
+      "step": 3534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066146315180699e-06,
+      "loss": 0.7341,
+      "step": 3535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0607723729498696e-06,
+      "loss": 0.7785,
+      "step": 3536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055400316745096e-06,
+      "loss": 0.6891,
+      "step": 3537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050030148617691e-06,
+      "loss": 0.7885,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044661870618218e-06,
+      "loss": 0.7359,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0392954847965425e-06,
+      "loss": 0.7824,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033930993201799e-06,
+      "loss": 0.7964,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028568397882397e-06,
+      "loss": 0.7378,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023207700886027e-06,
+      "loss": 0.729,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017848904259648e-06,
+      "loss": 0.6997,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012492010049497e-06,
+      "loss": 0.8554,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0071370203010846e-06,
+      "loss": 0.75,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001783937059198e-06,
+      "loss": 0.6118,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996432762367893e-06,
+      "loss": 0.6322,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991083498270489e-06,
+      "loss": 0.7898,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985736146809583e-06,
+      "loss": 0.7347,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980390710027054e-06,
+      "loss": 0.7146,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975047189964027e-06,
+      "loss": 0.8201,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969705588660909e-06,
+      "loss": 0.7348,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964365908157363e-06,
+      "loss": 0.6991,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959028150492338e-06,
+      "loss": 0.6942,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9536923177040365e-06,
+      "loss": 0.7104,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948358411829919e-06,
+      "loss": 0.7338,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9430264349067216e-06,
+      "loss": 0.7526,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937696388970441e-06,
+      "loss": 0.7486,
+      "step": 3559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932368276056335e-06,
+      "loss": 0.7351,
+      "step": 3560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9270420981989295e-06,
+      "loss": 0.7968,
+      "step": 3561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921717857431997e-06,
+      "loss": 0.771,
+      "step": 3562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9163955557885836e-06,
+      "loss": 0.7654,
+      "step": 3563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911075195300991e-06,
+      "loss": 0.7855,
+      "step": 3564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905756778000778e-06,
+      "loss": 0.6795,
+      "step": 3565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900440305918769e-06,
+      "loss": 0.7857,
+      "step": 3566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8951257810850304e-06,
+      "loss": 0.8129,
+      "step": 3567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889813205528895e-06,
+      "loss": 0.7029,
+      "step": 3568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8845025812789605e-06,
+      "loss": 0.7147,
+      "step": 3569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8791939103630594e-06,
+      "loss": 0.8075,
+      "step": 3570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8738871948082956e-06,
+      "loss": 0.8009,
+      "step": 3571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868582436641006e-06,
+      "loss": 0.6866,
+      "step": 3572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863279637886807e-06,
+      "loss": 0.7602,
+      "step": 3573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857978800570553e-06,
+      "loss": 0.7598,
+      "step": 3574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8526799267163395e-06,
+      "loss": 0.7313,
+      "step": 3575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847383018347528e-06,
+      "loss": 0.787,
+      "step": 3576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842088077486722e-06,
+      "loss": 0.6968,
+      "step": 3577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836795106155776e-06,
+      "loss": 0.8207,
+      "step": 3578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831504106375797e-06,
+      "loss": 0.8464,
+      "step": 3579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826215080167125e-06,
+      "loss": 0.7386,
+      "step": 3580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820928029549359e-06,
+      "loss": 0.7553,
+      "step": 3581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81564295654134e-06,
+      "loss": 0.8343,
+      "step": 3582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810359863161155e-06,
+      "loss": 0.7155,
+      "step": 3583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805078751426135e-06,
+      "loss": 0.7828,
+      "step": 3584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7997996233528496e-06,
+      "loss": 0.7793,
+      "step": 3585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794522480957112e-06,
+      "loss": 0.6679,
+      "step": 3586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789247326253993e-06,
+      "loss": 0.6377,
+      "step": 3587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7839741612577776e-06,
+      "loss": 0.6758,
+      "step": 3588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778702987982013e-06,
+      "loss": 0.7315,
+      "step": 3589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7734338084394675e-06,
+      "loss": 0.8398,
+      "step": 3590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768166624642167e-06,
+      "loss": 0.8241,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762901438601368e-06,
+      "loss": 0.6653,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7576382523275555e-06,
+      "loss": 0.6546,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75237706783046e-06,
+      "loss": 0.7703,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747117887119047e-06,
+      "loss": 0.7415,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741860712201516e-06,
+      "loss": 0.802,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736605545085302e-06,
+      "loss": 0.6817,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731352387777065e-06,
+      "loss": 0.7692,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726101242282708e-06,
+      "loss": 0.775,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720852110607361e-06,
+      "loss": 0.6941,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715604994755386e-06,
+      "loss": 0.7591,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 0.7559,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705116818535154e-06,
+      "loss": 0.7708,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699875762171762e-06,
+      "loss": 0.7282,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694636729641491e-06,
+      "loss": 0.6577,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689399722944835e-06,
+      "loss": 0.7612,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684164744081536e-06,
+      "loss": 0.7037,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678931795050539e-06,
+      "loss": 0.6779,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673700877850036e-06,
+      "loss": 0.7033,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668471994477436e-06,
+      "loss": 0.7056,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66324514692936e-06,
+      "loss": 0.6866,
+      "step": 3611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658020337201666e-06,
+      "loss": 0.6542,
+      "step": 3612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652797567289428e-06,
+      "loss": 0.8471,
+      "step": 3613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64757683918694e-06,
+      "loss": 0.7185,
+      "step": 3614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642358154887725e-06,
+      "loss": 0.6714,
+      "step": 3615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637141516384508e-06,
+      "loss": 0.6757,
+      "step": 3616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631926925669245e-06,
+      "loss": 0.7375,
+      "step": 3617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626714384733118e-06,
+      "loss": 0.745,
+      "step": 3618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621503895566507e-06,
+      "loss": 0.874,
+      "step": 3619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6162954601590256e-06,
+      "loss": 0.6771,
+      "step": 3620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611089080499486e-06,
+      "loss": 0.8092,
+      "step": 3621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6058847585759335e-06,
+      "loss": 0.7893,
+      "step": 3622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6006824963756226e-06,
+      "loss": 0.7836,
+      "step": 3623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595482295885007e-06,
+      "loss": 0.661,
+      "step": 3624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590284159089772e-06,
+      "loss": 0.7464,
+      "step": 3625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585088087974804e-06,
+      "loss": 0.7286,
+      "step": 3626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579894084524206e-06,
+      "loss": 0.7021,
+      "step": 3627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574702150721292e-06,
+      "loss": 0.6897,
+      "step": 3628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5695122885485755e-06,
+      "loss": 0.7783,
+      "step": 3629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56432449998779e-06,
+      "loss": 0.7823,
+      "step": 3630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559138787019874e-06,
+      "loss": 0.7479,
+      "step": 3631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5539551516249735e-06,
+      "loss": 0.6726,
+      "step": 3632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5487735957824446e-06,
+      "loss": 0.6861,
+      "step": 3633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543594121470835e-06,
+      "loss": 0.6566,
+      "step": 3634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538416730667914e-06,
+      "loss": 0.7271,
+      "step": 3635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5332414253506554e-06,
+      "loss": 0.6634,
+      "step": 3636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528068207495221e-06,
+      "loss": 0.7371,
+      "step": 3637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522897079076993e-06,
+      "loss": 0.803,
+      "step": 3638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517728042070537e-06,
+      "loss": 0.7339,
+      "step": 3639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512561098449642e-06,
+      "loss": 0.8633,
+      "step": 3640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507396250187287e-06,
+      "loss": 0.7514,
+      "step": 3641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502233499255641e-06,
+      "loss": 0.7313,
+      "step": 3642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497072847626087e-06,
+      "loss": 0.7364,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491914297269201e-06,
+      "loss": 0.7445,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486757850154756e-06,
+      "loss": 0.88,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481603508251727e-06,
+      "loss": 0.7611,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4764512735282724e-06,
+      "loss": 0.6535,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471301147951759e-06,
+      "loss": 0.6959,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466153133488743e-06,
+      "loss": 0.7279,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461007232104974e-06,
+      "loss": 0.7183,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4558634457654e-06,
+      "loss": 0.7675,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450721776434152e-06,
+      "loss": 0.7727,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445582226074553e-06,
+      "loss": 0.7131,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440444796649136e-06,
+      "loss": 0.6578,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435309490119598e-06,
+      "loss": 0.6934,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430176308446844e-06,
+      "loss": 0.6457,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425045253590951e-06,
+      "loss": 0.6505,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4199163275112024e-06,
+      "loss": 0.7238,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414789532166062e-06,
+      "loss": 0.6605,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409664869513168e-06,
+      "loss": 0.7029,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40454234150936e-06,
+      "loss": 0.7079,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399421950110657e-06,
+      "loss": 0.6781,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394303697272261e-06,
+      "loss": 0.741,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38918758494856e-06,
+      "loss": 0.6861,
+      "step": 3664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384073615093118e-06,
+      "loss": 0.6685,
+      "step": 3665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378961789658686e-06,
+      "loss": 0.7582,
+      "step": 3666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3738521105972055e-06,
+      "loss": 0.752,
+      "step": 3667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368744579859777e-06,
+      "loss": 0.8033,
+      "step": 3668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363639199396702e-06,
+      "loss": 0.6882,
+      "step": 3669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35853597115744e-06,
+      "loss": 0.6954,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353434897090649e-06,
+      "loss": 0.7177,
+      "step": 3671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348335979144158e-06,
+      "loss": 0.7129,
+      "step": 3672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343239219264962e-06,
+      "loss": 0.7967,
+      "step": 3673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3381446193992475e-06,
+      "loss": 0.7779,
+      "step": 3674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333052181492357e-06,
+      "loss": 0.7766,
+      "step": 3675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327961907488831e-06,
+      "loss": 0.7488,
+      "step": 3676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322873799332373e-06,
+      "loss": 0.7564,
+      "step": 3677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317787858965851e-06,
+      "loss": 0.8763,
+      "step": 3678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312704088331314e-06,
+      "loss": 0.6502,
+      "step": 3679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307622489369985e-06,
+      "loss": 0.7002,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302543064022252e-06,
+      "loss": 0.6739,
+      "step": 3681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297465814227678e-06,
+      "loss": 0.6974,
+      "step": 3682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292390741924987e-06,
+      "loss": 0.7076,
+      "step": 3683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287317849052075e-06,
+      "loss": 0.6717,
+      "step": 3684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282247137546019e-06,
+      "loss": 0.7879,
+      "step": 3685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2771786093430425e-06,
+      "loss": 0.7601,
+      "step": 3686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2721122663785496e-06,
+      "loss": 0.6802,
+      "step": 3687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267048110587096e-06,
+      "loss": 0.8424,
+      "step": 3688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261986143902419e-06,
+      "loss": 0.6159,
+      "step": 3689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256926368257415e-06,
+      "loss": 0.6945,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251868785584132e-06,
+      "loss": 0.6927,
+      "step": 3691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2468133978137945e-06,
+      "loss": 0.7756,
+      "step": 3692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241760206876783e-06,
+      "loss": 0.7268,
+      "step": 3693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23670921470264e-06,
+      "loss": 0.7029,
+      "step": 3694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231660423220072e-06,
+      "loss": 0.6838,
+      "step": 3695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226613834356934e-06,
+      "loss": 0.788,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221569450040253e-06,
+      "loss": 0.6738,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216527272196207e-06,
+      "loss": 0.7694,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211487302750133e-06,
+      "loss": 0.6992,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20644954362653e-06,
+      "loss": 0.7917,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201413996749039e-06,
+      "loss": 0.8101,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196380664040468e-06,
+      "loss": 0.6878,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191349547422785e-06,
+      "loss": 0.6979,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186320648817094e-06,
+      "loss": 0.7009,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181293970143667e-06,
+      "loss": 0.645,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176269513321915e-06,
+      "loss": 0.7956,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171247280270417e-06,
+      "loss": 0.6969,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166227272906895e-06,
+      "loss": 0.7271,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1612094931482135e-06,
+      "loss": 0.7094,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156193942910397e-06,
+      "loss": 0.6971,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151180624108616e-06,
+      "loss": 0.6507,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146169538657185e-06,
+      "loss": 0.805,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141160688469576e-06,
+      "loss": 0.6758,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136154075458392e-06,
+      "loss": 0.7718,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131149701535391e-06,
+      "loss": 0.7245,
+      "step": 3715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126147568611477e-06,
+      "loss": 0.6721,
+      "step": 3716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121147678596697e-06,
+      "loss": 0.6785,
+      "step": 3717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1161500334002425e-06,
+      "loss": 0.7211,
+      "step": 3718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111154634930441e-06,
+      "loss": 0.8099,
+      "step": 3719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106161485094765e-06,
+      "loss": 0.6795,
+      "step": 3720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1011705857998415e-06,
+      "loss": 0.7192,
+      "step": 3721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096181938951419e-06,
+      "loss": 0.7014,
+      "step": 3722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091195546454398e-06,
+      "loss": 0.7217,
+      "step": 3723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086211410212806e-06,
+      "loss": 0.7397,
+      "step": 3724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081229532129826e-06,
+      "loss": 0.7522,
+      "step": 3725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076249914107772e-06,
+      "loss": 0.7627,
+      "step": 3726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071272558048085e-06,
+      "loss": 0.8229,
+      "step": 3727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066297465851354e-06,
+      "loss": 0.7309,
+      "step": 3728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061324639417298e-06,
+      "loss": 0.6884,
+      "step": 3729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056354080644774e-06,
+      "loss": 0.7379,
+      "step": 3730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051385791431776e-06,
+      "loss": 0.6966,
+      "step": 3731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046419773675421e-06,
+      "loss": 0.7919,
+      "step": 3732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041456029271962e-06,
+      "loss": 0.7105,
+      "step": 3733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036494560116798e-06,
+      "loss": 0.8459,
+      "step": 3734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031535368104439e-06,
+      "loss": 0.7196,
+      "step": 3735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026578455128539e-06,
+      "loss": 0.7599,
+      "step": 3736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021623823081869e-06,
+      "loss": 0.7114,
+      "step": 3737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016671473856346e-06,
+      "loss": 0.7734,
+      "step": 3738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011721409343008e-06,
+      "loss": 0.8163,
+      "step": 3739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006773631432011e-06,
+      "loss": 0.7591,
+      "step": 3740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001828142012648e-06,
+      "loss": 0.6464,
+      "step": 3741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9968849429733396e-06,
+      "loss": 0.7936,
+      "step": 3742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991944036201625e-06,
+      "loss": 0.7708,
+      "step": 3743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9870054235841774e-06,
+      "loss": 0.6172,
+      "step": 3744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98206910700678e-06,
+      "loss": 0.7503,
+      "step": 3745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97713508835435e-06,
+      "loss": 0.6709,
+      "step": 3746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9722033695109266e-06,
+      "loss": 0.7706,
+      "step": 3747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967273952359666e-06,
+      "loss": 0.7955,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962346838782853e-06,
+      "loss": 0.6768,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957422030661882e-06,
+      "loss": 0.6868,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9524995298772705e-06,
+      "loss": 0.8096,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94757933830867e-06,
+      "loss": 0.7176,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942661457834828e-06,
+      "loss": 0.7926,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937745890333623e-06,
+      "loss": 0.8211,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932832637682039e-06,
+      "loss": 0.8091,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927921701756192e-06,
+      "loss": 0.7499,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923013084431307e-06,
+      "loss": 0.7701,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9181067875817126e-06,
+      "loss": 0.7276,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913202813080865e-06,
+      "loss": 0.7047,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90830116280133e-06,
+      "loss": 0.6705,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903401838614782e-06,
+      "loss": 0.7457,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898504842392017e-06,
+      "loss": 0.6918,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893610176002926e-06,
+      "loss": 0.7359,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888717841316523e-06,
+      "loss": 0.6834,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883827840200927e-06,
+      "loss": 0.7917,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 0.7385,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874054846150194e-06,
+      "loss": 0.7412,
+      "step": 3767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869171856946832e-06,
+      "loss": 0.6881,
+      "step": 3768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864291208777839e-06,
+      "loss": 0.728,
+      "step": 3769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85941290350688e-06,
+      "loss": 0.7664,
+      "step": 3770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854536942996712e-06,
+      "loss": 0.8024,
+      "step": 3771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849663329109206e-06,
+      "loss": 0.6466,
+      "step": 3772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844792063705326e-06,
+      "loss": 0.7659,
+      "step": 3773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839923148645156e-06,
+      "loss": 0.7796,
+      "step": 3774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835056585787873e-06,
+      "loss": 0.7667,
+      "step": 3775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830192376991749e-06,
+      "loss": 0.742,
+      "step": 3776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8253305241141705e-06,
+      "loss": 0.7657,
+      "step": 3777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820471029011615e-06,
+      "loss": 0.7488,
+      "step": 3778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815613893539666e-06,
+      "loss": 0.7368,
+      "step": 3779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8107591195530046e-06,
+      "loss": 0.8127,
+      "step": 3780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8059067089054025e-06,
+      "loss": 0.8186,
+      "step": 3781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801056663449737e-06,
+      "loss": 0.7,
+      "step": 3782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7962089850379814e-06,
+      "loss": 0.7398,
+      "step": 3783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7913636755212034e-06,
+      "loss": 0.728,
+      "step": 3784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7865207367495716e-06,
+      "loss": 0.7452,
+      "step": 3785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.781680170572337e-06,
+      "loss": 0.7408,
+      "step": 3786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7768419788378507e-06,
+      "loss": 0.7626,
+      "step": 3787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.772006163393571e-06,
+      "loss": 0.6439,
+      "step": 3788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7671727260860247e-06,
+      "loss": 0.7756,
+      "step": 3789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7623416687608494e-06,
+      "loss": 0.7321,
+      "step": 3790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7575129932627564e-06,
+      "loss": 0.6655,
+      "step": 3791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7526867014355685e-06,
+      "loss": 0.662,
+      "step": 3792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747862795122188e-06,
+      "loss": 0.6457,
+      "step": 3793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7430412761645974e-06,
+      "loss": 0.6772,
+      "step": 3794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7382221464038815e-06,
+      "loss": 0.6923,
+      "step": 3795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7334054076802064e-06,
+      "loss": 0.6369,
+      "step": 3796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728591061832826e-06,
+      "loss": 0.6702,
+      "step": 3797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7237791107000855e-06,
+      "loss": 0.7763,
+      "step": 3798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7189695561194016e-06,
+      "loss": 0.7072,
+      "step": 3799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7141623999272915e-06,
+      "loss": 0.7415,
+      "step": 3800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.709357643959347e-06,
+      "loss": 0.6946,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70455529005025e-06,
+      "loss": 0.7327,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.699755340033764e-06,
+      "loss": 0.8143,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694957795742721e-06,
+      "loss": 0.7672,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6901626590090565e-06,
+      "loss": 0.7664,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685369931663777e-06,
+      "loss": 0.7295,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680579615536961e-06,
+      "loss": 0.7889,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6757917124577756e-06,
+      "loss": 0.8368,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6710062242544654e-06,
+      "loss": 0.7879,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.66622315275435e-06,
+      "loss": 0.7658,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6614424997838328e-06,
+      "loss": 0.8352,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6566642671683806e-06,
+      "loss": 0.6723,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6518884567325486e-06,
+      "loss": 0.7658,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6471150702999623e-06,
+      "loss": 0.7804,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642344109693322e-06,
+      "loss": 0.7018,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6375755767344047e-06,
+      "loss": 0.7808,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63280947324405e-06,
+      "loss": 0.7612,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6280458010421817e-06,
+      "loss": 0.751,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6232845619477895e-06,
+      "loss": 0.7156,
+      "step": 3819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618525757778937e-06,
+      "loss": 0.741,
+      "step": 3820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613769390352757e-06,
+      "loss": 0.7836,
+      "step": 3821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6090154614854432e-06,
+      "loss": 0.8154,
+      "step": 3822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6042639729922747e-06,
+      "loss": 0.6958,
+      "step": 3823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5995149266875906e-06,
+      "loss": 0.7302,
+      "step": 3824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5947683243847907e-06,
+      "loss": 0.7428,
+      "step": 3825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5900241678963498e-06,
+      "loss": 0.7357,
+      "step": 3826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5852824590338087e-06,
+      "loss": 0.7595,
+      "step": 3827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58054319960777e-06,
+      "loss": 0.7266,
+      "step": 3828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5758063914279064e-06,
+      "loss": 0.6625,
+      "step": 3829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5710720363029437e-06,
+      "loss": 0.7813,
+      "step": 3830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5663401360406834e-06,
+      "loss": 0.794,
+      "step": 3831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.561610692447982e-06,
+      "loss": 0.6848,
+      "step": 3832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5568837073307618e-06,
+      "loss": 0.761,
+      "step": 3833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.552159182494007e-06,
+      "loss": 0.7205,
+      "step": 3834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5474371197417546e-06,
+      "loss": 0.7216,
+      "step": 3835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5427175208771102e-06,
+      "loss": 0.7939,
+      "step": 3836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5380003877022353e-06,
+      "loss": 0.7391,
+      "step": 3837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.533285722018349e-06,
+      "loss": 0.8659,
+      "step": 3838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5285735256257337e-06,
+      "loss": 0.694,
+      "step": 3839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5238638003237134e-06,
+      "loss": 0.7286,
+      "step": 3840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5191565479106914e-06,
+      "loss": 0.7202,
+      "step": 3841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514451770184113e-06,
+      "loss": 0.6958,
+      "step": 3842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5097494689404754e-06,
+      "loss": 0.7927,
+      "step": 3843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5050496459753367e-06,
+      "loss": 0.7896,
+      "step": 3844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5003523030833077e-06,
+      "loss": 0.7579,
+      "step": 3845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4956574420580525e-06,
+      "loss": 0.7967,
+      "step": 3846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49096506469229e-06,
+      "loss": 0.7124,
+      "step": 3847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48627517277778e-06,
+      "loss": 0.7305,
+      "step": 3848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4815877681053433e-06,
+      "loss": 0.7431,
+      "step": 3849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4769028524648494e-06,
+      "loss": 0.7426,
+      "step": 3850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4722204276452164e-06,
+      "loss": 0.687,
+      "step": 3851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467540495434415e-06,
+      "loss": 0.7576,
+      "step": 3852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462863057619448e-06,
+      "loss": 0.739,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4581881159863927e-06,
+      "loss": 0.7832,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453515672320349e-06,
+      "loss": 0.7178,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4488457284054753e-06,
+      "loss": 0.7835,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4441782860249716e-06,
+      "loss": 0.7618,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4395133469610865e-06,
+      "loss": 0.7356,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4348509129951067e-06,
+      "loss": 0.6545,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4301909859073714e-06,
+      "loss": 0.7805,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4255335674772484e-06,
+      "loss": 0.714,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.420878659483161e-06,
+      "loss": 0.7334,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416226263702567e-06,
+      "loss": 0.7073,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4115763819119685e-06,
+      "loss": 0.5883,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4069290158869073e-06,
+      "loss": 0.8502,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402284167401959e-06,
+      "loss": 0.7218,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397641838230743e-06,
+      "loss": 0.7443,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393002030145918e-06,
+      "loss": 0.7913,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3883647449191748e-06,
+      "loss": 0.5951,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38372998432125e-06,
+      "loss": 0.7975,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.379097750121899e-06,
+      "loss": 0.727,
+      "step": 3871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374468044089937e-06,
+      "loss": 0.6671,
+      "step": 3872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3698408679931894e-06,
+      "loss": 0.8686,
+      "step": 3873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3652162235985318e-06,
+      "loss": 0.8085,
+      "step": 3874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3605941126718656e-06,
+      "loss": 0.7397,
+      "step": 3875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355974536978129e-06,
+      "loss": 0.8114,
+      "step": 3876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351357498281289e-06,
+      "loss": 0.7986,
+      "step": 3877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3467429983443477e-06,
+      "loss": 0.6842,
+      "step": 3878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3421310389293305e-06,
+      "loss": 0.7722,
+      "step": 3879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3375216217972996e-06,
+      "loss": 0.8321,
+      "step": 3880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3329147487083423e-06,
+      "loss": 0.7627,
+      "step": 3881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328310421421579e-06,
+      "loss": 0.7551,
+      "step": 3882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323708641695157e-06,
+      "loss": 0.703,
+      "step": 3883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3191094112862434e-06,
+      "loss": 0.8623,
+      "step": 3884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3145127319510404e-06,
+      "loss": 0.7982,
+      "step": 3885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.309918605444773e-06,
+      "loss": 0.7892,
+      "step": 3886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3053270335216923e-06,
+      "loss": 0.775,
+      "step": 3887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3007380179350767e-06,
+      "loss": 0.6464,
+      "step": 3888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296151560437214e-06,
+      "loss": 0.7904,
+      "step": 3889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2915676627794403e-06,
+      "loss": 0.7603,
+      "step": 3890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2869863267120906e-06,
+      "loss": 0.7677,
+      "step": 3891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2824075539845334e-06,
+      "loss": 0.6679,
+      "step": 3892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2778313463451585e-06,
+      "loss": 0.7019,
+      "step": 3893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2732577055413716e-06,
+      "loss": 0.752,
+      "step": 3894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2686866333196023e-06,
+      "loss": 0.7758,
+      "step": 3895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264118131425301e-06,
+      "loss": 0.75,
+      "step": 3896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2595522016029268e-06,
+      "loss": 0.782,
+      "step": 3897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2549888455959667e-06,
+      "loss": 0.7905,
+      "step": 3898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250428065146921e-06,
+      "loss": 0.7796,
+      "step": 3899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2458698619973084e-06,
+      "loss": 0.7768,
+      "step": 3900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241314237887664e-06,
+      "loss": 0.6909,
+      "step": 3901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2367611945575308e-06,
+      "loss": 0.6284,
+      "step": 3902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2322107337454733e-06,
+      "loss": 0.6636,
+      "step": 3903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227662857189071e-06,
+      "loss": 0.7514,
+      "step": 3904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2231175666249114e-06,
+      "loss": 0.7619,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218574863788603e-06,
+      "loss": 0.6973,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2140347504147473e-06,
+      "loss": 0.6498,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2094972282369863e-06,
+      "loss": 0.8224,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2049622989879446e-06,
+      "loss": 0.7006,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200429964399272e-06,
+      "loss": 0.7203,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1959002262016247e-06,
+      "loss": 0.76,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191373086124666e-06,
+      "loss": 0.7149,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1868485458970675e-06,
+      "loss": 0.8389,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182326607246512e-06,
+      "loss": 0.7812,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.17780727189968e-06,
+      "loss": 0.7254,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1732905415822645e-06,
+      "loss": 0.6721,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168776418018963e-06,
+      "loss": 0.7878,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.164264902933478e-06,
+      "loss": 0.723,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159755998048518e-06,
+      "loss": 0.6397,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1552497050857812e-06,
+      "loss": 0.7299,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150746025765994e-06,
+      "loss": 0.6934,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1462449618088576e-06,
+      "loss": 0.6863,
+      "step": 3922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1417465149330926e-06,
+      "loss": 0.7274,
+      "step": 3923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1372506868564135e-06,
+      "loss": 0.7729,
+      "step": 3924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132757479295535e-06,
+      "loss": 0.7154,
+      "step": 3925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1282668939661777e-06,
+      "loss": 0.6771,
+      "step": 3926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1237789325830457e-06,
+      "loss": 0.7806,
+      "step": 3927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1192935968598558e-06,
+      "loss": 0.7376,
+      "step": 3928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1148108885093166e-06,
+      "loss": 0.7035,
+      "step": 3929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 0.707,
+      "step": 3930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105853360772011e-06,
+      "loss": 0.7182,
+      "step": 3931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1013785448056454e-06,
+      "loss": 0.6603,
+      "step": 3932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0969063630527264e-06,
+      "loss": 0.7395,
+      "step": 3933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0924368172209408e-06,
+      "loss": 0.776,
+      "step": 3934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0879699090169702e-06,
+      "loss": 0.768,
+      "step": 3935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083505640146486e-06,
+      "loss": 0.6945,
+      "step": 3936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079044012314156e-06,
+      "loss": 0.7372,
+      "step": 3937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074585027223628e-06,
+      "loss": 0.6121,
+      "step": 3938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0701286865775614e-06,
+      "loss": 0.6278,
+      "step": 3939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065674992077584e-06,
+      "loss": 0.784,
+      "step": 3940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0612239454243266e-06,
+      "loss": 0.7552,
+      "step": 3941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0567755483174043e-06,
+      "loss": 0.7107,
+      "step": 3942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0523298024554215e-06,
+      "loss": 0.6736,
+      "step": 3943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0478867095359733e-06,
+      "loss": 0.6742,
+      "step": 3944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.043446271255632e-06,
+      "loss": 0.7801,
+      "step": 3945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039008489309967e-06,
+      "loss": 0.6564,
+      "step": 3946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0345733653935285e-06,
+      "loss": 0.7221,
+      "step": 3947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0301409011998515e-06,
+      "loss": 0.6074,
+      "step": 3948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0257110984214577e-06,
+      "loss": 0.6859,
+      "step": 3949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0212839587498544e-06,
+      "loss": 0.6504,
+      "step": 3950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0168594838755203e-06,
+      "loss": 0.6219,
+      "step": 3951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0124376754879305e-06,
+      "loss": 0.6994,
+      "step": 3952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008018535275534e-06,
+      "loss": 0.7777,
+      "step": 3953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0036020649257657e-06,
+      "loss": 0.6595,
+      "step": 3954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9991882661250416e-06,
+      "loss": 0.834,
+      "step": 3955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9947771405587434e-06,
+      "loss": 0.7307,
+      "step": 3956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9903686899112584e-06,
+      "loss": 0.742,
+      "step": 3957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9859629158659253e-06,
+      "loss": 0.7433,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981559820105079e-06,
+      "loss": 0.8194,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9771594043100227e-06,
+      "loss": 0.6899,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9727616701610407e-06,
+      "loss": 0.7125,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968366619337394e-06,
+      "loss": 0.7472,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9639742535173123e-06,
+      "loss": 0.7383,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959584574378005e-06,
+      "loss": 0.7031,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955197583595657e-06,
+      "loss": 0.7122,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.950813282845424e-06,
+      "loss": 0.807,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9464316738014353e-06,
+      "loss": 0.7155,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942052758136795e-06,
+      "loss": 0.6985,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937676537523568e-06,
+      "loss": 0.707,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.93330301363281e-06,
+      "loss": 0.7123,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.7329,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9245640626977012e-06,
+      "loss": 0.6319,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.920198638990295e-06,
+      "loss": 0.7846,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.915835918679215e-06,
+      "loss": 0.746,
+      "step": 3974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911475903430365e-06,
+      "loss": 0.7965,
+      "step": 3975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9071185949085923e-06,
+      "loss": 0.7429,
+      "step": 3976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9027639947777207e-06,
+      "loss": 0.7757,
+      "step": 3977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8984121047005385e-06,
+      "loss": 0.7697,
+      "step": 3978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8940629263388e-06,
+      "loss": 0.7676,
+      "step": 3979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8897164613532224e-06,
+      "loss": 0.7346,
+      "step": 3980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8853727114034902e-06,
+      "loss": 0.7376,
+      "step": 3981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881031678148244e-06,
+      "loss": 0.6977,
+      "step": 3982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876693363245092e-06,
+      "loss": 0.7057,
+      "step": 3983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8723577683506063e-06,
+      "loss": 0.7059,
+      "step": 3984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8680248951203173e-06,
+      "loss": 0.6281,
+      "step": 3985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.863694745208719e-06,
+      "loss": 0.7664,
+      "step": 3986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8593673202692562e-06,
+      "loss": 0.6265,
+      "step": 3987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8550426219543504e-06,
+      "loss": 0.7372,
+      "step": 3988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850720651915363e-06,
+      "loss": 0.7097,
+      "step": 3989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8464014118026274e-06,
+      "loss": 0.8104,
+      "step": 3990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8420849032654273e-06,
+      "loss": 0.6763,
+      "step": 3991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837771127952007e-06,
+      "loss": 0.6715,
+      "step": 3992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8334600875095697e-06,
+      "loss": 0.7059,
+      "step": 3993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829151783584262e-06,
+      "loss": 0.6875,
+      "step": 3994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8248462178212e-06,
+      "loss": 0.6848,
+      "step": 3995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8205433918644466e-06,
+      "loss": 0.7012,
+      "step": 3996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8162433073570196e-06,
+      "loss": 0.7898,
+      "step": 3997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811945965940892e-06,
+      "loss": 0.6698,
+      "step": 3998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8076513692569897e-06,
+      "loss": 0.7555,
+      "step": 3999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8033595189451846e-06,
+      "loss": 0.6683,
+      "step": 4000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799070416644305e-06,
+      "loss": 0.7199,
+      "step": 4001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7947840639921308e-06,
+      "loss": 0.7194,
+      "step": 4002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7905004626253886e-06,
+      "loss": 0.7069,
+      "step": 4003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7862196141797593e-06,
+      "loss": 0.6737,
+      "step": 4004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7819415202898613e-06,
+      "loss": 0.6751,
+      "step": 4005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7776661825892804e-06,
+      "loss": 0.6417,
+      "step": 4006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77339360271053e-06,
+      "loss": 0.7169,
+      "step": 4007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769123782285081e-06,
+      "loss": 0.7667,
+      "step": 4008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76485672294335e-06,
+      "loss": 0.789,
+      "step": 4009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760592426314698e-06,
+      "loss": 0.6974,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7563308940274335e-06,
+      "loss": 0.8295,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7520721277088023e-06,
+      "loss": 0.6849,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7478161289850005e-06,
+      "loss": 0.7515,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7435628994811648e-06,
+      "loss": 0.7016,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739312440821378e-06,
+      "loss": 0.6651,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7350647546286604e-06,
+      "loss": 0.6263,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73081984252498e-06,
+      "loss": 0.7667,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7265777061312326e-06,
+      "loss": 0.7281,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722338347067267e-06,
+      "loss": 0.7178,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7181017669518682e-06,
+      "loss": 0.7891,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713867967402757e-06,
+      "loss": 0.7149,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709636950036597e-06,
+      "loss": 0.7541,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705408716468979e-06,
+      "loss": 0.7123,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701183268314451e-06,
+      "loss": 0.711,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696960607186474e-06,
+      "loss": 0.7216,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692740734697461e-06,
+      "loss": 0.6417,
+      "step": 4026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688523652458753e-06,
+      "loss": 0.6553,
+      "step": 4027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6843093620806292e-06,
+      "loss": 0.7146,
+      "step": 4028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680097865172303e-06,
+      "loss": 0.6709,
+      "step": 4029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6758891633419136e-06,
+      "loss": 0.7378,
+      "step": 4030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6716832581965402e-06,
+      "loss": 0.6458,
+      "step": 4031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6674801513421945e-06,
+      "loss": 0.6525,
+      "step": 4032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663279844383815e-06,
+      "loss": 0.8068,
+      "step": 4033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6590823389252753e-06,
+      "loss": 0.6815,
+      "step": 4034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654887636569381e-06,
+      "loss": 0.7534,
+      "step": 4035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6506957389178532e-06,
+      "loss": 0.7681,
+      "step": 4036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6465066475713653e-06,
+      "loss": 0.7327,
+      "step": 4037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642320364129498e-06,
+      "loss": 0.7267,
+      "step": 4038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.638136890190772e-06,
+      "loss": 0.6578,
+      "step": 4039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633956227352631e-06,
+      "loss": 0.7274,
+      "step": 4040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629778377211445e-06,
+      "loss": 0.8206,
+      "step": 4041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6256033413625136e-06,
+      "loss": 0.7585,
+      "step": 4042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6214311214000555e-06,
+      "loss": 0.8076,
+      "step": 4043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617261718917219e-06,
+      "loss": 0.7807,
+      "step": 4044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613095135506074e-06,
+      "loss": 0.7228,
+      "step": 4045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608931372757617e-06,
+      "loss": 0.7374,
+      "step": 4046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6047704322617683e-06,
+      "loss": 0.6523,
+      "step": 4047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6006123156073625e-06,
+      "loss": 0.7255,
+      "step": 4048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596457024382163e-06,
+      "loss": 0.7138,
+      "step": 4049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5923045601728525e-06,
+      "loss": 0.7524,
+      "step": 4050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588154924565035e-06,
+      "loss": 0.7159,
+      "step": 4051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584008119143234e-06,
+      "loss": 0.7862,
+      "step": 4052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5798641454908945e-06,
+      "loss": 0.7416,
+      "step": 4053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575723005190369e-06,
+      "loss": 0.7496,
+      "step": 4054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5715846998229486e-06,
+      "loss": 0.7384,
+      "step": 4055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5674492309688213e-06,
+      "loss": 0.7409,
+      "step": 4056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563316600207103e-06,
+      "loss": 0.7517,
+      "step": 4057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5591868091158257e-06,
+      "loss": 0.6833,
+      "step": 4058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5550598592719324e-06,
+      "loss": 0.7163,
+      "step": 4059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.550935752251289e-06,
+      "loss": 0.7387,
+      "step": 4060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546814489628664e-06,
+      "loss": 0.7423,
+      "step": 4061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5426960729777496e-06,
+      "loss": 0.8097,
+      "step": 4062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5385805038711474e-06,
+      "loss": 0.699,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534467783880373e-06,
+      "loss": 0.765,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5303579145758574e-06,
+      "loss": 0.7885,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526250897526932e-06,
+      "loss": 0.6805,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5221467343018502e-06,
+      "loss": 0.6575,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518045426467771e-06,
+      "loss": 0.7409,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513946975590764e-06,
+      "loss": 0.6636,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5098513832358074e-06,
+      "loss": 0.711,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505758650966792e-06,
+      "loss": 0.7507,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5016687803465033e-06,
+      "loss": 0.6864,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4975817729366558e-06,
+      "loss": 0.7164,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49349763029785e-06,
+      "loss": 0.6868,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489416353989601e-06,
+      "loss": 0.8176,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485337945570333e-06,
+      "loss": 0.7218,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4812624065973688e-06,
+      "loss": 0.6611,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4771897386269416e-06,
+      "loss": 0.6787,
+      "step": 4078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4731199432141796e-06,
+      "loss": 0.7966,
+      "step": 4079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4690530219131214e-06,
+      "loss": 0.7283,
+      "step": 4080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464988976276708e-06,
+      "loss": 0.6702,
+      "step": 4081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.460927807856778e-06,
+      "loss": 0.7472,
+      "step": 4082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.456869518204077e-06,
+      "loss": 0.8223,
+      "step": 4083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4528141088682435e-06,
+      "loss": 0.7395,
+      "step": 4084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.448761581397824e-06,
+      "loss": 0.7148,
+      "step": 4085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4447119373402596e-06,
+      "loss": 0.7727,
+      "step": 4086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440665178241892e-06,
+      "loss": 0.6939,
+      "step": 4087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436621305647964e-06,
+      "loss": 0.7304,
+      "step": 4088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4325803211026133e-06,
+      "loss": 0.7,
+      "step": 4089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4285422261488688e-06,
+      "loss": 0.7037,
+      "step": 4090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424507022328674e-06,
+      "loss": 0.767,
+      "step": 4091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4204747111828463e-06,
+      "loss": 0.7386,
+      "step": 4092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416445294251112e-06,
+      "loss": 0.7085,
+      "step": 4093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 0.6678,
+      "step": 4094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4083951491832947e-06,
+      "loss": 0.7683,
+      "step": 4095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404374424121132e-06,
+      "loss": 0.8365,
+      "step": 4096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4003565994208967e-06,
+      "loss": 0.6916,
+      "step": 4097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3963416766167834e-06,
+      "loss": 0.7505,
+      "step": 4098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392329657241875e-06,
+      "loss": 0.6957,
+      "step": 4099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3883205428281473e-06,
+      "loss": 0.707,
+      "step": 4100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384314334906468e-06,
+      "loss": 0.7169,
+      "step": 4101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3803110350065884e-06,
+      "loss": 0.6473,
+      "step": 4102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3763106446571537e-06,
+      "loss": 0.717,
+      "step": 4103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372313165385707e-06,
+      "loss": 0.7445,
+      "step": 4104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3683185987186606e-06,
+      "loss": 0.776,
+      "step": 4105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364326946181331e-06,
+      "loss": 0.7237,
+      "step": 4106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3603382092979143e-06,
+      "loss": 0.7889,
+      "step": 4107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3563523895914962e-06,
+      "loss": 0.7914,
+      "step": 4108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35236948858405e-06,
+      "loss": 0.6723,
+      "step": 4109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3483895077964257e-06,
+      "loss": 0.7526,
+      "step": 4110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344412448748368e-06,
+      "loss": 0.6339,
+      "step": 4111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3404383129585018e-06,
+      "loss": 0.7639,
+      "step": 4112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3364671019443363e-06,
+      "loss": 0.7174,
+      "step": 4113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3324988172222673e-06,
+      "loss": 0.7728,
+      "step": 4114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3285334603075637e-06,
+      "loss": 0.7134,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324571032714387e-06,
+      "loss": 0.8334,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3206115359557724e-06,
+      "loss": 0.6864,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3166549715436427e-06,
+      "loss": 0.7537,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312701340988799e-06,
+      "loss": 0.6887,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3087506458009144e-06,
+      "loss": 0.771,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3048028874885486e-06,
+      "loss": 0.6117,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3008580675591462e-06,
+      "loss": 0.7468,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2969161875190147e-06,
+      "loss": 0.7067,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2929772488733494e-06,
+      "loss": 0.6719,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2890412531262194e-06,
+      "loss": 0.7084,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2851082017805704e-06,
+      "loss": 0.6673,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2811780963382258e-06,
+      "loss": 0.8396,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2772509382998775e-06,
+      "loss": 0.7181,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2733267291650983e-06,
+      "loss": 0.7098,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2694054704323345e-06,
+      "loss": 0.7183,
+      "step": 4130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265487163598903e-06,
+      "loss": 0.7928,
+      "step": 4131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2615718101609986e-06,
+      "loss": 0.6898,
+      "step": 4132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2576594116136786e-06,
+      "loss": 0.7838,
+      "step": 4133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.253749969450881e-06,
+      "loss": 0.7452,
+      "step": 4134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2498434851654125e-06,
+      "loss": 0.762,
+      "step": 4135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2459399602489507e-06,
+      "loss": 0.6326,
+      "step": 4136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2420393961920428e-06,
+      "loss": 0.6623,
+      "step": 4137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238141794484101e-06,
+      "loss": 0.7276,
+      "step": 4138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2342471566134103e-06,
+      "loss": 0.8207,
+      "step": 4139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2303554840671326e-06,
+      "loss": 0.7989,
+      "step": 4140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2264667783312786e-06,
+      "loss": 0.7339,
+      "step": 4141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222581040890741e-06,
+      "loss": 0.7048,
+      "step": 4142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218698273229275e-06,
+      "loss": 0.7409,
+      "step": 4143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214818476829499e-06,
+      "loss": 0.7709,
+      "step": 4144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2109416531729032e-06,
+      "loss": 0.6851,
+      "step": 4145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207067803739834e-06,
+      "loss": 0.7193,
+      "step": 4146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203196930009508e-06,
+      "loss": 0.7908,
+      "step": 4147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199329033460005e-06,
+      "loss": 0.6859,
+      "step": 4148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1954641155682654e-06,
+      "loss": 0.7472,
+      "step": 4149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1916021778100994e-06,
+      "loss": 0.6826,
+      "step": 4150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1877432216601667e-06,
+      "loss": 0.732,
+      "step": 4151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183887248591996e-06,
+      "loss": 0.8643,
+      "step": 4152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1800342600779855e-06,
+      "loss": 0.8032,
+      "step": 4153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1761842575893757e-06,
+      "loss": 0.7185,
+      "step": 4154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1723372425962817e-06,
+      "loss": 0.7124,
+      "step": 4155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168493216567664e-06,
+      "loss": 0.8337,
+      "step": 4156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1646521809713583e-06,
+      "loss": 0.8438,
+      "step": 4157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16081413727405e-06,
+      "loss": 0.6959,
+      "step": 4158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1569790869412755e-06,
+      "loss": 0.6912,
+      "step": 4159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1531470314374394e-06,
+      "loss": 0.7091,
+      "step": 4160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1493179722257973e-06,
+      "loss": 0.6817,
+      "step": 4161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1454919107684615e-06,
+      "loss": 0.7611,
+      "step": 4162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1416688485264016e-06,
+      "loss": 0.8793,
+      "step": 4163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1378487869594357e-06,
+      "loss": 0.7041,
+      "step": 4164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1340317275262423e-06,
+      "loss": 0.8048,
+      "step": 4165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1302176716843524e-06,
+      "loss": 0.7688,
+      "step": 4166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1264066208901477e-06,
+      "loss": 0.7824,
+      "step": 4167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.122598576598869e-06,
+      "loss": 0.7744,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1187935402645976e-06,
+      "loss": 0.7251,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114991513340272e-06,
+      "loss": 0.7597,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.111192497277692e-06,
+      "loss": 0.62,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107396493527489e-06,
+      "loss": 0.7946,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1036035035391587e-06,
+      "loss": 0.6557,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099813528761032e-06,
+      "loss": 0.7836,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096026570640306e-06,
+      "loss": 0.7298,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092242630623016e-06,
+      "loss": 0.7347,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0884617101540404e-06,
+      "loss": 0.704,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0846838106771127e-06,
+      "loss": 0.7325,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080908933634811e-06,
+      "loss": 0.6847,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0771370804685564e-06,
+      "loss": 0.7263,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0733682526186216e-06,
+      "loss": 0.7715,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069602451524114e-06,
+      "loss": 0.7554,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0658396786229927e-06,
+      "loss": 0.7014,
+      "step": 4183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0620799353520602e-06,
+      "loss": 0.7134,
+      "step": 4184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0583232231469605e-06,
+      "loss": 0.7275,
+      "step": 4185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054569543442183e-06,
+      "loss": 0.8061,
+      "step": 4186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.050818897671051e-06,
+      "loss": 0.7349,
+      "step": 4187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047071287265735e-06,
+      "loss": 0.6898,
+      "step": 4188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0433267136572533e-06,
+      "loss": 0.8277,
+      "step": 4189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0395851782754515e-06,
+      "loss": 0.7581,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035846682549026e-06,
+      "loss": 0.7542,
+      "step": 4191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0321112279055e-06,
+      "loss": 0.7598,
+      "step": 4192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0283788157712504e-06,
+      "loss": 0.7705,
+      "step": 4193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024649447571485e-06,
+      "loss": 0.7453,
+      "step": 4194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020923124730246e-06,
+      "loss": 0.6504,
+      "step": 4195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017199848670417e-06,
+      "loss": 0.7255,
+      "step": 4196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0134796208137175e-06,
+      "loss": 0.6487,
+      "step": 4197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009762442580704e-06,
+      "loss": 0.6228,
+      "step": 4198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0060483153907693e-06,
+      "loss": 0.6776,
+      "step": 4199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0023372406621344e-06,
+      "loss": 0.6674,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9986292198118616e-06,
+      "loss": 0.7008,
+      "step": 4201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9949242542558466e-06,
+      "loss": 0.7161,
+      "step": 4202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9912223454088155e-06,
+      "loss": 0.6787,
+      "step": 4203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9875234946843314e-06,
+      "loss": 0.7105,
+      "step": 4204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9838277034947826e-06,
+      "loss": 0.7603,
+      "step": 4205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980134973251393e-06,
+      "loss": 0.6773,
+      "step": 4206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9764453053642265e-06,
+      "loss": 0.7517,
+      "step": 4207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972758701242161e-06,
+      "loss": 0.7866,
+      "step": 4208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969075162292918e-06,
+      "loss": 0.6679,
+      "step": 4209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965394689923036e-06,
+      "loss": 0.6738,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9617172855378974e-06,
+      "loss": 0.6777,
+      "step": 4211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9580429505417054e-06,
+      "loss": 0.6948,
+      "step": 4212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9543716863374874e-06,
+      "loss": 0.8134,
+      "step": 4213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950703494327104e-06,
+      "loss": 0.7645,
+      "step": 4214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94703837591124e-06,
+      "loss": 0.7345,
+      "step": 4215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9433763324894093e-06,
+      "loss": 0.8099,
+      "step": 4216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939717365459952e-06,
+      "loss": 0.7851,
+      "step": 4217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936061476220025e-06,
+      "loss": 0.7879,
+      "step": 4218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932408666165617e-06,
+      "loss": 0.6839,
+      "step": 4219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9287589366915484e-06,
+      "loss": 0.7586,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9251122891914464e-06,
+      "loss": 0.6631,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9214687250577766e-06,
+      "loss": 0.8066,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917828245681812e-06,
+      "loss": 0.6592,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9141908524536633e-06,
+      "loss": 0.7847,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105565467622578e-06,
+      "loss": 0.6761,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906925329995335e-06,
+      "loss": 0.7349,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032972035394649e-06,
+      "loss": 0.6605,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8996721687800346e-06,
+      "loss": 0.745,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8960502271012504e-06,
+      "loss": 0.7405,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8924313798861405e-06,
+      "loss": 0.75,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8888156285165427e-06,
+      "loss": 0.6478,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8852029743731203e-06,
+      "loss": 0.6834,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8815934188353536e-06,
+      "loss": 0.7853,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8779869632815384e-06,
+      "loss": 0.6324,
+      "step": 4234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8743836090887868e-06,
+      "loss": 0.7003,
+      "step": 4235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8707833576330237e-06,
+      "loss": 0.7205,
+      "step": 4236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8671862102889916e-06,
+      "loss": 0.6716,
+      "step": 4237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8635921684302548e-06,
+      "loss": 0.6486,
+      "step": 4238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8600012334291784e-06,
+      "loss": 0.7314,
+      "step": 4239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8564134066569529e-06,
+      "loss": 0.7135,
+      "step": 4240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8528286894835678e-06,
+      "loss": 0.6904,
+      "step": 4241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8492470832778442e-06,
+      "loss": 0.6327,
+      "step": 4242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8456685894074034e-06,
+      "loss": 0.7756,
+      "step": 4243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8420932092386756e-06,
+      "loss": 0.6593,
+      "step": 4244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.83852094413691e-06,
+      "loss": 0.738,
+      "step": 4245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8349517954661633e-06,
+      "loss": 0.6912,
+      "step": 4246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8313857645892996e-06,
+      "loss": 0.7835,
+      "step": 4247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8278228528679998e-06,
+      "loss": 0.6835,
+      "step": 4248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8242630616627433e-06,
+      "loss": 0.6957,
+      "step": 4249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820706392332824e-06,
+      "loss": 0.7068,
+      "step": 4250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8171528462363453e-06,
+      "loss": 0.7687,
+      "step": 4251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8136024247302152e-06,
+      "loss": 0.6678,
+      "step": 4252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8100551291701507e-06,
+      "loss": 0.7633,
+      "step": 4253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80651096091067e-06,
+      "loss": 0.6463,
+      "step": 4254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802969921305101e-06,
+      "loss": 0.7453,
+      "step": 4255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799432011705583e-06,
+      "loss": 0.77,
+      "step": 4256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7958972334630465e-06,
+      "loss": 0.803,
+      "step": 4257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 0.7458,
+      "step": 4258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7888370764467e-06,
+      "loss": 0.743,
+      "step": 4259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7853117003687848e-06,
+      "loss": 0.7126,
+      "step": 4260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7817894610396458e-06,
+      "loss": 0.6809,
+      "step": 4261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7782703598042327e-06,
+      "loss": 0.7307,
+      "step": 4262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7747543980063041e-06,
+      "loss": 0.8458,
+      "step": 4263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7712415769884162e-06,
+      "loss": 0.7135,
+      "step": 4264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7677318980919278e-06,
+      "loss": 0.7464,
+      "step": 4265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7642253626569995e-06,
+      "loss": 0.8192,
+      "step": 4266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7607219720225832e-06,
+      "loss": 0.6683,
+      "step": 4267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7572217275264359e-06,
+      "loss": 0.77,
+      "step": 4268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7537246305051213e-06,
+      "loss": 0.799,
+      "step": 4269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7502306822939862e-06,
+      "loss": 0.7303,
+      "step": 4270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7467398842271854e-06,
+      "loss": 0.6976,
+      "step": 4271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432522376376637e-06,
+      "loss": 0.7751,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7397677438571658e-06,
+      "loss": 0.7017,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7362864042162408e-06,
+      "loss": 0.729,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7328082200442176e-06,
+      "loss": 0.6967,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7293331926692336e-06,
+      "loss": 0.7043,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7258613234182075e-06,
+      "loss": 0.7244,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7223926136168668e-06,
+      "loss": 0.6999,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718927064589726e-06,
+      "loss": 0.6939,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7154646776600881e-06,
+      "loss": 0.7306,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7120054541500552e-06,
+      "loss": 0.6962,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085493953805187e-06,
+      "loss": 0.7167,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705096502671163e-06,
+      "loss": 0.6423,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016467773404644e-06,
+      "loss": 0.6473,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6982002207056825e-06,
+      "loss": 0.7008,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947568340828724e-06,
+      "loss": 0.6777,
+      "step": 4286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6913166187868868e-06,
+      "loss": 0.7036,
+      "step": 4287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6878795761313515e-06,
+      "loss": 0.6149,
+      "step": 4288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6844457074286946e-06,
+      "loss": 0.6888,
+      "step": 4289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6810150139901182e-06,
+      "loss": 0.7396,
+      "step": 4290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6775874971256278e-06,
+      "loss": 0.735,
+      "step": 4291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6741631581440066e-06,
+      "loss": 0.6941,
+      "step": 4292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670741998352824e-06,
+      "loss": 0.6692,
+      "step": 4293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6673240190584371e-06,
+      "loss": 0.6856,
+      "step": 4294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6639092215659901e-06,
+      "loss": 0.7293,
+      "step": 4295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660497607179411e-06,
+      "loss": 0.6532,
+      "step": 4296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.657089177201414e-06,
+      "loss": 0.7559,
+      "step": 4297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653683932933492e-06,
+      "loss": 0.7121,
+      "step": 4298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6502818756759275e-06,
+      "loss": 0.739,
+      "step": 4299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.646883006727782e-06,
+      "loss": 0.733,
+      "step": 4300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6434873273869023e-06,
+      "loss": 0.7412,
+      "step": 4301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6400948389499194e-06,
+      "loss": 0.6193,
+      "step": 4302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6367055427122358e-06,
+      "loss": 0.6952,
+      "step": 4303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.633319439968044e-06,
+      "loss": 0.7766,
+      "step": 4304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6299365320103211e-06,
+      "loss": 0.6905,
+      "step": 4305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6265568201308112e-06,
+      "loss": 0.7459,
+      "step": 4306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623180305620049e-06,
+      "loss": 0.6112,
+      "step": 4307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6198069897673386e-06,
+      "loss": 0.7269,
+      "step": 4308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6164368738607739e-06,
+      "loss": 0.7022,
+      "step": 4309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6130699591872233e-06,
+      "loss": 0.6523,
+      "step": 4310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6097062470323244e-06,
+      "loss": 0.7693,
+      "step": 4311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6063457386805004e-06,
+      "loss": 0.7428,
+      "step": 4312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6029884354149517e-06,
+      "loss": 0.6997,
+      "step": 4313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5996343385176506e-06,
+      "loss": 0.8239,
+      "step": 4314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596283449269348e-06,
+      "loss": 0.751,
+      "step": 4315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592935768949566e-06,
+      "loss": 0.7152,
+      "step": 4316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5895912988366068e-06,
+      "loss": 0.7113,
+      "step": 4317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5862500402075421e-06,
+      "loss": 0.6567,
+      "step": 4318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.582911994338221e-06,
+      "loss": 0.7995,
+      "step": 4319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5795771625032653e-06,
+      "loss": 0.7417,
+      "step": 4320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576245545976064e-06,
+      "loss": 0.6854,
+      "step": 4321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.572917146028783e-06,
+      "loss": 0.707,
+      "step": 4322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5695919639323676e-06,
+      "loss": 0.7067,
+      "step": 4323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5662700009565178e-06,
+      "loss": 0.7436,
+      "step": 4324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5629512583697181e-06,
+      "loss": 0.7919,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5596357374392124e-06,
+      "loss": 0.5979,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5563234394310257e-06,
+      "loss": 0.7314,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5530143656099473e-06,
+      "loss": 0.8338,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.549708517239532e-06,
+      "loss": 0.8657,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.546405895582106e-06,
+      "loss": 0.7778,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5431065018987645e-06,
+      "loss": 0.6527,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539810337449369e-06,
+      "loss": 0.7422,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.536517403492549e-06,
+      "loss": 0.741,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5332277012856966e-06,
+      "loss": 0.7834,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5299412320849716e-06,
+      "loss": 0.6939,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5266579971453065e-06,
+      "loss": 0.7187,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5233779977203867e-06,
+      "loss": 0.7626,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5201012350626732e-06,
+      "loss": 0.6958,
+      "step": 4338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5168277104233765e-06,
+      "loss": 0.7308,
+      "step": 4339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135574250524898e-06,
+      "loss": 0.7441,
+      "step": 4340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5102903801987578e-06,
+      "loss": 0.7416,
+      "step": 4341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507026577109686e-06,
+      "loss": 0.8126,
+      "step": 4342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503766017031547e-06,
+      "loss": 0.6556,
+      "step": 4343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5005087012093755e-06,
+      "loss": 0.6772,
+      "step": 4344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497254630886964e-06,
+      "loss": 0.7732,
+      "step": 4345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940038073068696e-06,
+      "loss": 0.6624,
+      "step": 4346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4907562317104041e-06,
+      "loss": 0.8071,
+      "step": 4347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487511905337644e-06,
+      "loss": 0.7894,
+      "step": 4348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4842708294274243e-06,
+      "loss": 0.6881,
+      "step": 4349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.481033005217337e-06,
+      "loss": 0.7416,
+      "step": 4350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4777984339437357e-06,
+      "loss": 0.6975,
+      "step": 4351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4745671168417265e-06,
+      "loss": 0.6866,
+      "step": 4352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4713390551451733e-06,
+      "loss": 0.7459,
+      "step": 4353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4681142500867097e-06,
+      "loss": 0.7949,
+      "step": 4354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4648927028977066e-06,
+      "loss": 0.669,
+      "step": 4355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4616744148083062e-06,
+      "loss": 0.7622,
+      "step": 4356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4584593870473918e-06,
+      "loss": 0.6635,
+      "step": 4357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455247620842618e-06,
+      "loss": 0.8135,
+      "step": 4358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4520391174203851e-06,
+      "loss": 0.7031,
+      "step": 4359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4488338780058452e-06,
+      "loss": 0.698,
+      "step": 4360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4456319038229084e-06,
+      "loss": 0.7966,
+      "step": 4361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.442433196094236e-06,
+      "loss": 0.6966,
+      "step": 4362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.439237756041245e-06,
+      "loss": 0.6324,
+      "step": 4363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436045584884105e-06,
+      "loss": 0.6699,
+      "step": 4364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.432856683841728e-06,
+      "loss": 0.8021,
+      "step": 4365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4296710541317882e-06,
+      "loss": 0.7444,
+      "step": 4366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.426488696970706e-06,
+      "loss": 0.6836,
+      "step": 4367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4233096135736525e-06,
+      "loss": 0.6015,
+      "step": 4368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4201338051545533e-06,
+      "loss": 0.7534,
+      "step": 4369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4169612729260718e-06,
+      "loss": 0.709,
+      "step": 4370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4137920180996312e-06,
+      "loss": 0.7817,
+      "step": 4371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4106260418854033e-06,
+      "loss": 0.7357,
+      "step": 4372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4074633454923015e-06,
+      "loss": 0.698,
+      "step": 4373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4043039301279904e-06,
+      "loss": 0.6707,
+      "step": 4374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4011477969988773e-06,
+      "loss": 0.7226,
+      "step": 4375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3979949473101263e-06,
+      "loss": 0.7171,
+      "step": 4376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3948453822656404e-06,
+      "loss": 0.6847,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3916991030680671e-06,
+      "loss": 0.7272,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3885561109188018e-06,
+      "loss": 0.7403,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3854164070179855e-06,
+      "loss": 0.6362,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3822799925645036e-06,
+      "loss": 0.7911,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.379146868755985e-06,
+      "loss": 0.7485,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3760170367887982e-06,
+      "loss": 0.7485,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3728904978580604e-06,
+      "loss": 0.5841,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3697672531576279e-06,
+      "loss": 0.6546,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366647303880102e-06,
+      "loss": 0.6413,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3635306512168257e-06,
+      "loss": 0.7132,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3604172963578754e-06,
+      "loss": 0.6899,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3573072404920763e-06,
+      "loss": 0.713,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3542004848069979e-06,
+      "loss": 0.7805,
+      "step": 4390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.351097030488937e-06,
+      "loss": 0.7478,
+      "step": 4391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3479968787229402e-06,
+      "loss": 0.6943,
+      "step": 4392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344900030692784e-06,
+      "loss": 0.681,
+      "step": 4393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3418064875809934e-06,
+      "loss": 0.8276,
+      "step": 4394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3387162505688267e-06,
+      "loss": 0.8219,
+      "step": 4395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3356293208362759e-06,
+      "loss": 0.8616,
+      "step": 4396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3325456995620756e-06,
+      "loss": 0.6408,
+      "step": 4397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3294653879236963e-06,
+      "loss": 0.6907,
+      "step": 4398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3263883870973425e-06,
+      "loss": 0.6033,
+      "step": 4399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3233146982579593e-06,
+      "loss": 0.7083,
+      "step": 4400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3202443225792182e-06,
+      "loss": 0.7831,
+      "step": 4401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3171772612335332e-06,
+      "loss": 0.6521,
+      "step": 4402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3141135153920503e-06,
+      "loss": 0.7044,
+      "step": 4403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3110530862246506e-06,
+      "loss": 0.7866,
+      "step": 4404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3079959748999494e-06,
+      "loss": 0.7686,
+      "step": 4405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304942182585287e-06,
+      "loss": 0.7145,
+      "step": 4406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3018917104467489e-06,
+      "loss": 0.6431,
+      "step": 4407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2988445596491484e-06,
+      "loss": 0.8245,
+      "step": 4408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958007313560217e-06,
+      "loss": 0.8182,
+      "step": 4409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292760226729649e-06,
+      "loss": 0.6986,
+      "step": 4410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.289723046931034e-06,
+      "loss": 0.6311,
+      "step": 4411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2866891931199132e-06,
+      "loss": 0.7854,
+      "step": 4412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2836586664547545e-06,
+      "loss": 0.731,
+      "step": 4413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.280631468092749e-06,
+      "loss": 0.7126,
+      "step": 4414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.277607599189824e-06,
+      "loss": 0.6864,
+      "step": 4415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2745870609006327e-06,
+      "loss": 0.7496,
+      "step": 4416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271569854378557e-06,
+      "loss": 0.6966,
+      "step": 4417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2685559807757075e-06,
+      "loss": 0.6772,
+      "step": 4418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.265545441242917e-06,
+      "loss": 0.6554,
+      "step": 4419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2625382369297512e-06,
+      "loss": 0.6903,
+      "step": 4420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2595343689844997e-06,
+      "loss": 0.636,
+      "step": 4421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 0.7737,
+      "step": 4422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2535366467845335e-06,
+      "loss": 0.8434,
+      "step": 4423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250542794820022e-06,
+      "loss": 0.686,
+      "step": 4424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2475522838038433e-06,
+      "loss": 0.7456,
+      "step": 4425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2445651148779126e-06,
+      "loss": 0.6832,
+      "step": 4426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.241581289182866e-06,
+      "loss": 0.7037,
+      "step": 4427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238600807858068e-06,
+      "loss": 0.7207,
+      "step": 4428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2356236720416048e-06,
+      "loss": 0.7284,
+      "step": 4429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2326498828702848e-06,
+      "loss": 0.8098,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.229679441479641e-06,
+      "loss": 0.7441,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267123490039201e-06,
+      "loss": 0.7189,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2237486065760995e-06,
+      "loss": 0.6418,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2207882153278716e-06,
+      "loss": 0.7944,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2178311763896521e-06,
+      "loss": 0.6973,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2148774908905782e-06,
+      "loss": 0.8457,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.211927159958498e-06,
+      "loss": 0.642,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2089801847199878e-06,
+      "loss": 0.7608,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2060365663003393e-06,
+      "loss": 0.7048,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2030963058235634e-06,
+      "loss": 0.7078,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2001594044123876e-06,
+      "loss": 0.7942,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1972258631882527e-06,
+      "loss": 0.7377,
+      "step": 4442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1942956832713259e-06,
+      "loss": 0.6822,
+      "step": 4443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1913688657804856e-06,
+      "loss": 0.7629,
+      "step": 4444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1884454118333244e-06,
+      "loss": 0.67,
+      "step": 4445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1855253225461517e-06,
+      "loss": 0.7003,
+      "step": 4446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1826085990339943e-06,
+      "loss": 0.8097,
+      "step": 4447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1796952424105922e-06,
+      "loss": 0.8004,
+      "step": 4448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767852537884029e-06,
+      "loss": 0.6904,
+      "step": 4449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.17387863427859e-06,
+      "loss": 0.7143,
+      "step": 4450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1709753849910376e-06,
+      "loss": 0.6982,
+      "step": 4451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168075507034341e-06,
+      "loss": 0.7079,
+      "step": 4452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1651790015158082e-06,
+      "loss": 0.755,
+      "step": 4453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1622858695414607e-06,
+      "loss": 0.6711,
+      "step": 4454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1593961122160257e-06,
+      "loss": 0.7565,
+      "step": 4455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1565097306429528e-06,
+      "loss": 0.7479,
+      "step": 4456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1536267259243915e-06,
+      "loss": 0.652,
+      "step": 4457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.150747099161208e-06,
+      "loss": 0.7023,
+      "step": 4458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1478708514529791e-06,
+      "loss": 0.6689,
+      "step": 4459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1449979838979841e-06,
+      "loss": 0.7062,
+      "step": 4460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1421284975932235e-06,
+      "loss": 0.6797,
+      "step": 4461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1392623936343994e-06,
+      "loss": 0.6649,
+      "step": 4462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1363996731159188e-06,
+      "loss": 0.762,
+      "step": 4463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.133540337130903e-06,
+      "loss": 0.7316,
+      "step": 4464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1306843867711802e-06,
+      "loss": 0.6625,
+      "step": 4465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278318231272834e-06,
+      "loss": 0.8102,
+      "step": 4466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1249826472884574e-06,
+      "loss": 0.7065,
+      "step": 4467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.122136860342643e-06,
+      "loss": 0.7452,
+      "step": 4468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192944633764978e-06,
+      "loss": 0.6252,
+      "step": 4469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164554574753794e-06,
+      "loss": 0.7498,
+      "step": 4470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1136198437233535e-06,
+      "loss": 0.7471,
+      "step": 4471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110787623203189e-06,
+      "loss": 0.7382,
+      "step": 4472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1079587969963535e-06,
+      "loss": 0.7038,
+      "step": 4473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.105133366183032e-06,
+      "loss": 0.756,
+      "step": 4474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1023113318420998e-06,
+      "loss": 0.762,
+      "step": 4475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0994926950511408e-06,
+      "loss": 0.7332,
+      "step": 4476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0966774568864424e-06,
+      "loss": 0.698,
+      "step": 4477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0938656184229935e-06,
+      "loss": 0.7364,
+      "step": 4478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0910571807344827e-06,
+      "loss": 0.7151,
+      "step": 4479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0882521448933047e-06,
+      "loss": 0.7439,
+      "step": 4480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0854505119705483e-06,
+      "loss": 0.7701,
+      "step": 4481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0826522830360087e-06,
+      "loss": 0.7415,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07985745915818e-06,
+      "loss": 0.7687,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0770660414042545e-06,
+      "loss": 0.8955,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0742780308401268e-06,
+      "loss": 0.7445,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0714934285303846e-06,
+      "loss": 0.8305,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0687122355383217e-06,
+      "loss": 0.8084,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0659344529259242e-06,
+      "loss": 0.6899,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06316008175388e-06,
+      "loss": 0.6925,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0603891230815744e-06,
+      "loss": 0.7441,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0576215779670807e-06,
+      "loss": 0.7355,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0548574474671835e-06,
+      "loss": 0.8125,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0520967326373544e-06,
+      "loss": 0.7616,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0493394345317598e-06,
+      "loss": 0.6952,
+      "step": 4494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0465855542032644e-06,
+      "loss": 0.6516,
+      "step": 4495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.043835092703429e-06,
+      "loss": 0.8542,
+      "step": 4496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.041088051082505e-06,
+      "loss": 0.6119,
+      "step": 4497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0383444303894453e-06,
+      "loss": 0.7851,
+      "step": 4498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035604231671885e-06,
+      "loss": 0.8004,
+      "step": 4499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.032867455976162e-06,
+      "loss": 0.7366,
+      "step": 4500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0301341043473033e-06,
+      "loss": 0.6811,
+      "step": 4501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027404177829031e-06,
+      "loss": 0.7862,
+      "step": 4502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0246776774637579e-06,
+      "loss": 0.6858,
+      "step": 4503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0219546042925842e-06,
+      "loss": 0.7419,
+      "step": 4504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0192349593553085e-06,
+      "loss": 0.7823,
+      "step": 4505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165187436904179e-06,
+      "loss": 0.6439,
+      "step": 4506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0138059583350868e-06,
+      "loss": 0.794,
+      "step": 4507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0110966043251869e-06,
+      "loss": 0.7466,
+      "step": 4508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083906826952684e-06,
+      "loss": 0.6933,
+      "step": 4509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0056881944785835e-06,
+      "loss": 0.7415,
+      "step": 4510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0029891407070669e-06,
+      "loss": 0.7212,
+      "step": 4511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0002935224113387e-06,
+      "loss": 0.6793,
+      "step": 4512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97601340620713e-07,
+      "loss": 0.8369,
+      "step": 4513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.949125963631912e-07,
+      "loss": 0.6576,
+      "step": 4514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.922272906654574e-07,
+      "loss": 0.7596,
+      "step": 4515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.895454245528902e-07,
+      "loss": 0.7059,
+      "step": 4516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.86866999049546e-07,
+      "loss": 0.7214,
+      "step": 4517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.841920151781736e-07,
+      "loss": 0.6294,
+      "step": 4518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.815204739602047e-07,
+      "loss": 0.7074,
+      "step": 4519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.788523764157599e-07,
+      "loss": 0.6996,
+      "step": 4520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.761877235636419e-07,
+      "loss": 0.7286,
+      "step": 4521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.735265164213349e-07,
+      "loss": 0.6552,
+      "step": 4522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.708687560050179e-07,
+      "loss": 0.6952,
+      "step": 4523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.682144433295414e-07,
+      "loss": 0.7108,
+      "step": 4524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.655635794084472e-07,
+      "loss": 0.708,
+      "step": 4525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.629161652539566e-07,
+      "loss": 0.7218,
+      "step": 4526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.602722018769773e-07,
+      "loss": 0.6873,
+      "step": 4527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.576316902870964e-07,
+      "loss": 0.6355,
+      "step": 4528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.54994631492584e-07,
+      "loss": 0.6908,
+      "step": 4529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.523610265003891e-07,
+      "loss": 0.6529,
+      "step": 4530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.497308763161461e-07,
+      "loss": 0.7369,
+      "step": 4531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471041819441673e-07,
+      "loss": 0.6944,
+      "step": 4532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.44480944387447e-07,
+      "loss": 0.7125,
+      "step": 4533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.418611646476616e-07,
+      "loss": 0.724,
+      "step": 4534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.392448437251589e-07,
+      "loss": 0.6131,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.36631982618974e-07,
+      "loss": 0.6241,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.34022582326819e-07,
+      "loss": 0.6865,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.314166438450833e-07,
+      "loss": 0.7153,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.288141681688367e-07,
+      "loss": 0.8558,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.262151562918209e-07,
+      "loss": 0.6808,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.236196092064642e-07,
+      "loss": 0.757,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.210275279038638e-07,
+      "loss": 0.7737,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.184389133737958e-07,
+      "loss": 0.7225,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.158537666047163e-07,
+      "loss": 0.7036,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.132720885837509e-07,
+      "loss": 0.7242,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.106938802967059e-07,
+      "loss": 0.8089,
+      "step": 4546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081191427280612e-07,
+      "loss": 0.7731,
+      "step": 4547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055478768609694e-07,
+      "loss": 0.7664,
+      "step": 4548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029800836772585e-07,
+      "loss": 0.7057,
+      "step": 4549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.004157641574318e-07,
+      "loss": 0.6837,
+      "step": 4550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978549192806662e-07,
+      "loss": 0.7762,
+      "step": 4551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.952975500248129e-07,
+      "loss": 0.7035,
+      "step": 4552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.927436573663894e-07,
+      "loss": 0.6815,
+      "step": 4553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901932422805914e-07,
+      "loss": 0.7103,
+      "step": 4554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.876463057412876e-07,
+      "loss": 0.7304,
+      "step": 4555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.851028487210145e-07,
+      "loss": 0.7378,
+      "step": 4556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.825628721909851e-07,
+      "loss": 0.742,
+      "step": 4557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.800263771210749e-07,
+      "loss": 0.7192,
+      "step": 4558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.774933644798411e-07,
+      "loss": 0.7716,
+      "step": 4559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.749638352345002e-07,
+      "loss": 0.7559,
+      "step": 4560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.724377903509473e-07,
+      "loss": 0.6781,
+      "step": 4561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69915230793742e-07,
+      "loss": 0.8146,
+      "step": 4562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.673961575261136e-07,
+      "loss": 0.6999,
+      "step": 4563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.648805715099629e-07,
+      "loss": 0.7545,
+      "step": 4564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623684737058579e-07,
+      "loss": 0.7054,
+      "step": 4565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.598598650730306e-07,
+      "loss": 0.7115,
+      "step": 4566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.573547465693866e-07,
+      "loss": 0.7593,
+      "step": 4567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.548531191514953e-07,
+      "loss": 0.7851,
+      "step": 4568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523549837745937e-07,
+      "loss": 0.6744,
+      "step": 4569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.498603413925888e-07,
+      "loss": 0.6546,
+      "step": 4570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.473691929580463e-07,
+      "loss": 0.6325,
+      "step": 4571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.448815394222043e-07,
+      "loss": 0.6706,
+      "step": 4572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.42397381734964e-07,
+      "loss": 0.7332,
+      "step": 4573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.399167208448922e-07,
+      "loss": 0.7339,
+      "step": 4574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.374395576992211e-07,
+      "loss": 0.7087,
+      "step": 4575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34965893243842e-07,
+      "loss": 0.7723,
+      "step": 4576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.324957284233215e-07,
+      "loss": 0.6716,
+      "step": 4577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.300290641808773e-07,
+      "loss": 0.744,
+      "step": 4578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.275659014583992e-07,
+      "loss": 0.7445,
+      "step": 4579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.25106241196435e-07,
+      "loss": 0.7324,
+      "step": 4580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.226500843341989e-07,
+      "loss": 0.7307,
+      "step": 4581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20197431809564e-07,
+      "loss": 0.7366,
+      "step": 4582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.17748284559069e-07,
+      "loss": 0.7489,
+      "step": 4583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.15302643517909e-07,
+      "loss": 0.6944,
+      "step": 4584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128605096199438e-07,
+      "loss": 0.7389,
+      "step": 4585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 0.6225,
+      "step": 4586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.079867669823394e-07,
+      "loss": 0.6334,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.055551601037226e-07,
+      "loss": 0.7002,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.031270640903388e-07,
+      "loss": 0.7679,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.007024798693552e-07,
+      "loss": 0.6617,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.982814083665825e-07,
+      "loss": 0.7235,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.958638505065031e-07,
+      "loss": 0.6692,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.934498072122509e-07,
+      "loss": 0.6187,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.910392794056198e-07,
+      "loss": 0.7165,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.886322680070624e-07,
+      "loss": 0.6934,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.862287739356844e-07,
+      "loss": 0.7272,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.838287981092529e-07,
+      "loss": 0.664,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.814323414441893e-07,
+      "loss": 0.7231,
+      "step": 4598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.790394048555727e-07,
+      "loss": 0.7131,
+      "step": 4599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766499892571366e-07,
+      "loss": 0.7137,
+      "step": 4600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742640955612723e-07,
+      "loss": 0.695,
+      "step": 4601
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.718817246790222e-07,
+      "loss": 0.6092,
+      "step": 4602
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.695028775200863e-07,
+      "loss": 0.6185,
+      "step": 4603
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.671275549928192e-07,
+      "loss": 0.7871,
+      "step": 4604
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.647557580042298e-07,
+      "loss": 0.7589,
+      "step": 4605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.623874874599801e-07,
+      "loss": 0.7399,
+      "step": 4606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.600227442643826e-07,
+      "loss": 0.7265,
+      "step": 4607
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.576615293204126e-07,
+      "loss": 0.6601,
+      "step": 4608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.553038435296845e-07,
+      "loss": 0.6914,
+      "step": 4609
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.529496877924736e-07,
+      "loss": 0.8024,
+      "step": 4610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.505990630077087e-07,
+      "loss": 0.7665,
+      "step": 4611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.48251970072964e-07,
+      "loss": 0.7224,
+      "step": 4612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.459084098844715e-07,
+      "loss": 0.6913,
+      "step": 4613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.435683833371077e-07,
+      "loss": 0.6658,
+      "step": 4614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.412318913244054e-07,
+      "loss": 0.6608,
+      "step": 4615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.388989347385444e-07,
+      "loss": 0.7523,
+      "step": 4616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.365695144703567e-07,
+      "loss": 0.7606,
+      "step": 4617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.342436314093226e-07,
+      "loss": 0.6943,
+      "step": 4618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.319212864435732e-07,
+      "loss": 0.667,
+      "step": 4619
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.296024804598844e-07,
+      "loss": 0.7042,
+      "step": 4620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.27287214343686e-07,
+      "loss": 0.7441,
+      "step": 4621
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.249754889790539e-07,
+      "loss": 0.6988,
+      "step": 4622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22667305248711e-07,
+      "loss": 0.6767,
+      "step": 4623
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.203626640340311e-07,
+      "loss": 0.6719,
+      "step": 4624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.180615662150292e-07,
+      "loss": 0.7029,
+      "step": 4625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.157640126703757e-07,
+      "loss": 0.6896,
+      "step": 4626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.13470004277379e-07,
+      "loss": 0.715,
+      "step": 4627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.111795419119994e-07,
+      "loss": 0.7465,
+      "step": 4628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088926264488427e-07,
+      "loss": 0.7683,
+      "step": 4629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.066092587611573e-07,
+      "loss": 0.7767,
+      "step": 4630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.043294397208411e-07,
+      "loss": 0.7831,
+      "step": 4631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.020531701984334e-07,
+      "loss": 0.6487,
+      "step": 4632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.997804510631179e-07,
+      "loss": 0.6608,
+      "step": 4633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.975112831827258e-07,
+      "loss": 0.7108,
+      "step": 4634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.952456674237307e-07,
+      "loss": 0.804,
+      "step": 4635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.929836046512494e-07,
+      "loss": 0.7554,
+      "step": 4636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.907250957290445e-07,
+      "loss": 0.7637,
+      "step": 4637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.884701415195128e-07,
+      "loss": 0.678,
+      "step": 4638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.862187428837086e-07,
+      "loss": 0.7866,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839709006813155e-07,
+      "loss": 0.688,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817266157706637e-07,
+      "loss": 0.6773,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.794858890087275e-07,
+      "loss": 0.7539,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.77248721251118e-07,
+      "loss": 0.7172,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.750151133520921e-07,
+      "loss": 0.7276,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727850661645418e-07,
+      "loss": 0.6434,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70558580540005e-07,
+      "loss": 0.6462,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.683356573286548e-07,
+      "loss": 0.7252,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.661162973793078e-07,
+      "loss": 0.6848,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.639005015394207e-07,
+      "loss": 0.7522,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.616882706550831e-07,
+      "loss": 0.6797,
+      "step": 4650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.594796055710284e-07,
+      "loss": 0.7589,
+      "step": 4651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.572745071306286e-07,
+      "loss": 0.6447,
+      "step": 4652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.550729761758901e-07,
+      "loss": 0.6666,
+      "step": 4653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.528750135474626e-07,
+      "loss": 0.8174,
+      "step": 4654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.506806200846294e-07,
+      "loss": 0.8576,
+      "step": 4655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.484897966253078e-07,
+      "loss": 0.7379,
+      "step": 4656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.463025440060622e-07,
+      "loss": 0.688,
+      "step": 4657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44118863062082e-07,
+      "loss": 0.709,
+      "step": 4658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.419387546271993e-07,
+      "loss": 0.647,
+      "step": 4659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.397622195338804e-07,
+      "loss": 0.7283,
+      "step": 4660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.375892586132282e-07,
+      "loss": 0.7228,
+      "step": 4661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3541987269498e-07,
+      "loss": 0.633,
+      "step": 4662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.332540626075068e-07,
+      "loss": 0.7693,
+      "step": 4663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.310918291778145e-07,
+      "loss": 0.772,
+      "step": 4664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289331732315463e-07,
+      "loss": 0.7786,
+      "step": 4665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.267780955929758e-07,
+      "loss": 0.7475,
+      "step": 4666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.246265970850129e-07,
+      "loss": 0.6992,
+      "step": 4667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.22478678529197e-07,
+      "loss": 0.6562,
+      "step": 4668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203343407457041e-07,
+      "loss": 0.7606,
+      "step": 4669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.181935845533415e-07,
+      "loss": 0.6975,
+      "step": 4670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.160564107695499e-07,
+      "loss": 0.7502,
+      "step": 4671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.139228202104008e-07,
+      "loss": 0.6878,
+      "step": 4672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.117928136906003e-07,
+      "loss": 0.7001,
+      "step": 4673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09666392023478e-07,
+      "loss": 0.6504,
+      "step": 4674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.075435560210074e-07,
+      "loss": 0.7057,
+      "step": 4675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054243064937803e-07,
+      "loss": 0.6481,
+      "step": 4676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.033086442510261e-07,
+      "loss": 0.8057,
+      "step": 4677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.011965701006029e-07,
+      "loss": 0.6971,
+      "step": 4678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990880848489977e-07,
+      "loss": 0.5794,
+      "step": 4679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.969831893013312e-07,
+      "loss": 0.6573,
+      "step": 4680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948818842613457e-07,
+      "loss": 0.762,
+      "step": 4681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.927841705314175e-07,
+      "loss": 0.7657,
+      "step": 4682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.906900489125533e-07,
+      "loss": 0.7492,
+      "step": 4683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.885995202043848e-07,
+      "loss": 0.7397,
+      "step": 4684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865125852051734e-07,
+      "loss": 0.6399,
+      "step": 4685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844292447118049e-07,
+      "loss": 0.7541,
+      "step": 4686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.823494995197975e-07,
+      "loss": 0.7296,
+      "step": 4687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802733504232938e-07,
+      "loss": 0.7095,
+      "step": 4688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.78200798215065e-07,
+      "loss": 0.7232,
+      "step": 4689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.76131843686506e-07,
+      "loss": 0.7574,
+      "step": 4690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.740664876276403e-07,
+      "loss": 0.6574,
+      "step": 4691
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.720047308271149e-07,
+      "loss": 0.6923,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.699465740722077e-07,
+      "loss": 0.8394,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.678920181488157e-07,
+      "loss": 0.7665,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.658410638414635e-07,
+      "loss": 0.6613,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.637937119333004e-07,
+      "loss": 0.7258,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617499632061019e-07,
+      "loss": 0.6977,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.597098184402671e-07,
+      "loss": 0.7046,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.576732784148131e-07,
+      "loss": 0.7464,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.556403439073887e-07,
+      "loss": 0.7187,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.53611015694262e-07,
+      "loss": 0.7049,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.515852945503241e-07,
+      "loss": 0.7336,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.49563181249091e-07,
+      "loss": 0.7813,
+      "step": 4703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.475446765626968e-07,
+      "loss": 0.764,
+      "step": 4704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.455297812618987e-07,
+      "loss": 0.8371,
+      "step": 4705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.435184961160834e-07,
+      "loss": 0.8045,
+      "step": 4706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.415108218932475e-07,
+      "loss": 0.7655,
+      "step": 4707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.395067593600156e-07,
+      "loss": 0.6376,
+      "step": 4708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375063092816313e-07,
+      "loss": 0.7717,
+      "step": 4709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355094724219601e-07,
+      "loss": 0.7626,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.335162495434876e-07,
+      "loss": 0.6544,
+      "step": 4711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.315266414073161e-07,
+      "loss": 0.7184,
+      "step": 4712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.295406487731714e-07,
+      "loss": 0.6968,
+      "step": 4713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275582723993967e-07,
+      "loss": 0.6789,
+      "step": 4714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255795130429575e-07,
+      "loss": 0.8542,
+      "step": 4715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.236043714594353e-07,
+      "loss": 0.7152,
+      "step": 4716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.216328484030286e-07,
+      "loss": 0.7272,
+      "step": 4717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.196649446265578e-07,
+      "loss": 0.7098,
+      "step": 4718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.177006608814594e-07,
+      "loss": 0.7604,
+      "step": 4719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15739997917788e-07,
+      "loss": 0.6519,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137829564842178e-07,
+      "loss": 0.7291,
+      "step": 4721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.118295373280335e-07,
+      "loss": 0.6585,
+      "step": 4722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098797411951428e-07,
+      "loss": 0.6615,
+      "step": 4723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.079335688300724e-07,
+      "loss": 0.7654,
+      "step": 4724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.059910209759545e-07,
+      "loss": 0.7409,
+      "step": 4725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.040520983745489e-07,
+      "loss": 0.7519,
+      "step": 4726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.021168017662226e-07,
+      "loss": 0.5903,
+      "step": 4727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001851318899642e-07,
+      "loss": 0.7069,
+      "step": 4728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.98257089483375e-07,
+      "loss": 0.6622,
+      "step": 4729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963326752826691e-07,
+      "loss": 0.7713,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.944118900226769e-07,
+      "loss": 0.753,
+      "step": 4731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.924947344368448e-07,
+      "loss": 0.7208,
+      "step": 4732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.905812092572304e-07,
+      "loss": 0.7227,
+      "step": 4733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886713152145083e-07,
+      "loss": 0.6452,
+      "step": 4734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867650530379619e-07,
+      "loss": 0.79,
+      "step": 4735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.848624234554921e-07,
+      "loss": 0.667,
+      "step": 4736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829634271936101e-07,
+      "loss": 0.6409,
+      "step": 4737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.810680649774413e-07,
+      "loss": 0.7507,
+      "step": 4738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.791763375307235e-07,
+      "loss": 0.7762,
+      "step": 4739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772882455758065e-07,
+      "loss": 0.762,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.754037898336472e-07,
+      "loss": 0.7635,
+      "step": 4741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7352297102382317e-07,
+      "loss": 0.7521,
+      "step": 4742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.716457898645155e-07,
+      "loss": 0.6853,
+      "step": 4743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.697722470725197e-07,
+      "loss": 0.6527,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6790234336324016e-07,
+      "loss": 0.7835,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.660360794506946e-07,
+      "loss": 0.6235,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6417345604750865e-07,
+      "loss": 0.7457,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6231447386491767e-07,
+      "loss": 0.6611,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6045913361276726e-07,
+      "loss": 0.7475,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 0.7488,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.567593817322158e-07,
+      "loss": 0.722,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549149715165546e-07,
+      "loss": 0.7187,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5307420605680497e-07,
+      "loss": 0.5662,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5123708605585793e-07,
+      "loss": 0.7382,
+      "step": 4754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4940361221521346e-07,
+      "loss": 0.6716,
+      "step": 4755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.475737852349749e-07,
+      "loss": 0.6961,
+      "step": 4756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.457476058138543e-07,
+      "loss": 0.7582,
+      "step": 4757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.439250746491752e-07,
+      "loss": 0.8191,
+      "step": 4758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.421061924368586e-07,
+      "loss": 0.6169,
+      "step": 4759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4029095987144334e-07,
+      "loss": 0.6825,
+      "step": 4760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384793776460661e-07,
+      "loss": 0.7167,
+      "step": 4761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3667144645247463e-07,
+      "loss": 0.6428,
+      "step": 4762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3486716698102e-07,
+      "loss": 0.711,
+      "step": 4763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.330665399206613e-07,
+      "loss": 0.7469,
+      "step": 4764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3126956595896055e-07,
+      "loss": 0.6609,
+      "step": 4765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2947624578208446e-07,
+      "loss": 0.6987,
+      "step": 4766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.276865800748065e-07,
+      "loss": 0.744,
+      "step": 4767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.259005695205043e-07,
+      "loss": 0.7087,
+      "step": 4768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.241182148011591e-07,
+      "loss": 0.6485,
+      "step": 4769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.223395165973587e-07,
+      "loss": 0.7412,
+      "step": 4770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.205644755882887e-07,
+      "loss": 0.7281,
+      "step": 4771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.187930924517436e-07,
+      "loss": 0.7666,
+      "step": 4772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.170253678641223e-07,
+      "loss": 0.7968,
+      "step": 4773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.152613025004204e-07,
+      "loss": 0.7323,
+      "step": 4774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.135008970342413e-07,
+      "loss": 0.7101,
+      "step": 4775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.117441521377885e-07,
+      "loss": 0.5374,
+      "step": 4776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.099910684818698e-07,
+      "loss": 0.6384,
+      "step": 4777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.082416467358952e-07,
+      "loss": 0.757,
+      "step": 4778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.064958875678715e-07,
+      "loss": 0.7891,
+      "step": 4779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.047537916444122e-07,
+      "loss": 0.8264,
+      "step": 4780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0301535963073066e-07,
+      "loss": 0.7408,
+      "step": 4781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.012805921906393e-07,
+      "loss": 0.6924,
+      "step": 4782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9954948998655595e-07,
+      "loss": 0.7394,
+      "step": 4783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.978220536794919e-07,
+      "loss": 0.6669,
+      "step": 4784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9609828392906504e-07,
+      "loss": 0.6508,
+      "step": 4785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.943781813934888e-07,
+      "loss": 0.7213,
+      "step": 4786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.92661746729579e-07,
+      "loss": 0.7533,
+      "step": 4787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909489805927502e-07,
+      "loss": 0.6875,
+      "step": 4788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.892398836370148e-07,
+      "loss": 0.717,
+      "step": 4789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8753445651498523e-07,
+      "loss": 0.7165,
+      "step": 4790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8583269987787607e-07,
+      "loss": 0.7682,
+      "step": 4791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8413461437549203e-07,
+      "loss": 0.7525,
+      "step": 4792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8244020065624334e-07,
+      "loss": 0.7129,
+      "step": 4793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.807494593671357e-07,
+      "loss": 0.7251,
+      "step": 4794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7906239115377163e-07,
+      "loss": 0.7736,
+      "step": 4795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7737899666035363e-07,
+      "loss": 0.6846,
+      "step": 4796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.756992765296774e-07,
+      "loss": 0.8244,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7402323140314e-07,
+      "loss": 0.6637,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7235086192073166e-07,
+      "loss": 0.8202,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7068216872104157e-07,
+      "loss": 0.6906,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6901715244125445e-07,
+      "loss": 0.6614,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.673558137171496e-07,
+      "loss": 0.7983,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6569815318310276e-07,
+      "loss": 0.6455,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6404417147208657e-07,
+      "loss": 0.7441,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.623938692156681e-07,
+      "loss": 0.7087,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.607472470440099e-07,
+      "loss": 0.6385,
+      "step": 4806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5910430558586673e-07,
+      "loss": 0.762,
+      "step": 4807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574650454685902e-07,
+      "loss": 0.7441,
+      "step": 4808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5582946731812726e-07,
+      "loss": 0.6352,
+      "step": 4809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5419757175901623e-07,
+      "loss": 0.7509,
+      "step": 4810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5256935941439085e-07,
+      "loss": 0.7305,
+      "step": 4811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5094483090597706e-07,
+      "loss": 0.8107,
+      "step": 4812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4932398685409653e-07,
+      "loss": 0.7068,
+      "step": 4813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4770682787766405e-07,
+      "loss": 0.6594,
+      "step": 4814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4609335459418137e-07,
+      "loss": 0.674,
+      "step": 4815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.444835676197511e-07,
+      "loss": 0.7687,
+      "step": 4816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.428774675690638e-07,
+      "loss": 0.7567,
+      "step": 4817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4127505505540226e-07,
+      "loss": 0.6665,
+      "step": 4818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3967633069064256e-07,
+      "loss": 0.7269,
+      "step": 4819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380812950852508e-07,
+      "loss": 0.6775,
+      "step": 4820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.364899488482853e-07,
+      "loss": 0.8,
+      "step": 4821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3490229258739794e-07,
+      "loss": 0.7429,
+      "step": 4822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3331832690882806e-07,
+      "loss": 0.7837,
+      "step": 4823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.317380524174074e-07,
+      "loss": 0.6816,
+      "step": 4824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301614697165567e-07,
+      "loss": 0.7436,
+      "step": 4825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2858857940828993e-07,
+      "loss": 0.6116,
+      "step": 4826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2701938209320995e-07,
+      "loss": 0.6931,
+      "step": 4827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2545387837050744e-07,
+      "loss": 0.7625,
+      "step": 4828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.238920688379654e-07,
+      "loss": 0.6549,
+      "step": 4829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223339540919546e-07,
+      "loss": 0.6453,
+      "step": 4830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.207795347274345e-07,
+      "loss": 0.7208,
+      "step": 4831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1922881133795827e-07,
+      "loss": 0.597,
+      "step": 4832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1768178451565877e-07,
+      "loss": 0.7335,
+      "step": 4833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.161384548512658e-07,
+      "loss": 0.7375,
+      "step": 4834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.145988229340935e-07,
+      "loss": 0.7705,
+      "step": 4835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1306288935204375e-07,
+      "loss": 0.7849,
+      "step": 4836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.115306546916097e-07,
+      "loss": 0.6469,
+      "step": 4837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.100021195378666e-07,
+      "loss": 0.6453,
+      "step": 4838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08477284474481e-07,
+      "loss": 0.5848,
+      "step": 4839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0695615008370814e-07,
+      "loss": 0.6408,
+      "step": 4840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.054387169463835e-07,
+      "loss": 0.7912,
+      "step": 4841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0392498564193685e-07,
+      "loss": 0.6899,
+      "step": 4842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0241495674837695e-07,
+      "loss": 0.6872,
+      "step": 4843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0090863084230705e-07,
+      "loss": 0.7506,
+      "step": 4844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9940600849891144e-07,
+      "loss": 0.7635,
+      "step": 4845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9790709029195895e-07,
+      "loss": 0.6914,
+      "step": 4846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.964118767938062e-07,
+      "loss": 0.724,
+      "step": 4847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9492036857539764e-07,
+      "loss": 0.7044,
+      "step": 4848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.934325662062576e-07,
+      "loss": 0.7106,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9194847025449944e-07,
+      "loss": 0.6147,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.904680812868188e-07,
+      "loss": 0.7346,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.889913998684979e-07,
+      "loss": 0.7203,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8751842656340013e-07,
+      "loss": 0.6741,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.860491619339778e-07,
+      "loss": 0.6884,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.845836065412633e-07,
+      "loss": 0.6781,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8312176094487223e-07,
+      "loss": 0.6677,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8166362570300475e-07,
+      "loss": 0.7528,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802092013724478e-07,
+      "loss": 0.7506,
+      "step": 4858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.787584885085648e-07,
+      "loss": 0.693,
+      "step": 4859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7731148766530714e-07,
+      "loss": 0.6921,
+      "step": 4860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7586819939520395e-07,
+      "loss": 0.7342,
+      "step": 4861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.744286242493721e-07,
+      "loss": 0.7607,
+      "step": 4862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.729927627775086e-07,
+      "loss": 0.6583,
+      "step": 4863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.715606155278905e-07,
+      "loss": 0.7139,
+      "step": 4864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7013218304737797e-07,
+      "loss": 0.718,
+      "step": 4865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6870746588141374e-07,
+      "loss": 0.662,
+      "step": 4866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.672864645740203e-07,
+      "loss": 0.7366,
+      "step": 4867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6586917966780143e-07,
+      "loss": 0.7431,
+      "step": 4868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.644556117039432e-07,
+      "loss": 0.6809,
+      "step": 4869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6304576122221035e-07,
+      "loss": 0.7205,
+      "step": 4870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.616396287609491e-07,
+      "loss": 0.6696,
+      "step": 4871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602372148570864e-07,
+      "loss": 0.7415,
+      "step": 4872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.588385200461307e-07,
+      "loss": 0.6235,
+      "step": 4873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.574435448621648e-07,
+      "loss": 0.6827,
+      "step": 4874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.560522898378559e-07,
+      "loss": 0.6894,
+      "step": 4875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5466475550445346e-07,
+      "loss": 0.6902,
+      "step": 4876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5328094239177813e-07,
+      "loss": 0.6537,
+      "step": 4877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5190085102823615e-07,
+      "loss": 0.6816,
+      "step": 4878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505244819408081e-07,
+      "loss": 0.6759,
+      "step": 4879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.491518356550571e-07,
+      "loss": 0.7911,
+      "step": 4880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.477829126951248e-07,
+      "loss": 0.6963,
+      "step": 4881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4641771358372537e-07,
+      "loss": 0.7444,
+      "step": 4882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4505623884215844e-07,
+      "loss": 0.7218,
+      "step": 4883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4369848899029715e-07,
+      "loss": 0.717,
+      "step": 4884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.423444645465933e-07,
+      "loss": 0.7592,
+      "step": 4885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.409941660280768e-07,
+      "loss": 0.739,
+      "step": 4886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3964759395035176e-07,
+      "loss": 0.5793,
+      "step": 4887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3830474882760357e-07,
+      "loss": 0.6971,
+      "step": 4888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3696563117259542e-07,
+      "loss": 0.7095,
+      "step": 4889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3563024149665936e-07,
+      "loss": 0.7305,
+      "step": 4890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3429858030971308e-07,
+      "loss": 0.6728,
+      "step": 4891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329706481202443e-07,
+      "loss": 0.6491,
+      "step": 4892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3164644543531957e-07,
+      "loss": 0.7639,
+      "step": 4893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3032597276058222e-07,
+      "loss": 0.6791,
+      "step": 4894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2900923060024783e-07,
+      "loss": 0.7562,
+      "step": 4895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2769621945711084e-07,
+      "loss": 0.7384,
+      "step": 4896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2638693983254023e-07,
+      "loss": 0.8121,
+      "step": 4897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.250813922264783e-07,
+      "loss": 0.8556,
+      "step": 4898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2377957713744513e-07,
+      "loss": 0.7045,
+      "step": 4899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2248149506253204e-07,
+      "loss": 0.6995,
+      "step": 4900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2118714649740912e-07,
+      "loss": 0.6391,
+      "step": 4901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1989653193631667e-07,
+      "loss": 0.6757,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1860965187207262e-07,
+      "loss": 0.6711,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1732650679606727e-07,
+      "loss": 0.7916,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1604709719826422e-07,
+      "loss": 0.6833,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1477142356720048e-07,
+      "loss": 0.7243,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1349948638999084e-07,
+      "loss": 0.7243,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.122312861523168e-07,
+      "loss": 0.6926,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1096682333843655e-07,
+      "loss": 0.6844,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.097060984311805e-07,
+      "loss": 0.6665,
+      "step": 4910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0844911191195249e-07,
+      "loss": 0.6517,
+      "step": 4911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0719586426072858e-07,
+      "loss": 0.656,
+      "step": 4912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0594635595605593e-07,
+      "loss": 0.6569,
+      "step": 4913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 0.7026,
+      "step": 4914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0345855929341797e-07,
+      "loss": 0.747,
+      "step": 4915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0222027188540828e-07,
+      "loss": 0.7576,
+      "step": 4916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0098572572386343e-07,
+      "loss": 0.7643,
+      "step": 4917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9975492128018858e-07,
+      "loss": 0.799,
+      "step": 4918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9852785902436112e-07,
+      "loss": 0.7479,
+      "step": 4919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9730453942493176e-07,
+      "loss": 0.8137,
+      "step": 4920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9608496294902023e-07,
+      "loss": 0.7093,
+      "step": 4921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9486913006231846e-07,
+      "loss": 0.665,
+      "step": 4922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9365704122908503e-07,
+      "loss": 0.6883,
+      "step": 4923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9244869691215418e-07,
+      "loss": 0.7099,
+      "step": 4924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9124409757292683e-07,
+      "loss": 0.7261,
+      "step": 4925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9004324367137505e-07,
+      "loss": 0.7062,
+      "step": 4926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8884613566604205e-07,
+      "loss": 0.7143,
+      "step": 4927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8765277401403438e-07,
+      "loss": 0.7383,
+      "step": 4928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8646315917103862e-07,
+      "loss": 0.7901,
+      "step": 4929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8527729159130147e-07,
+      "loss": 0.758,
+      "step": 4930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.840951717276429e-07,
+      "loss": 0.7516,
+      "step": 4931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8291680003145074e-07,
+      "loss": 0.6884,
+      "step": 4932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.817421769526828e-07,
+      "loss": 0.7317,
+      "step": 4933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.805713029398637e-07,
+      "loss": 0.7309,
+      "step": 4934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7940417844008906e-07,
+      "loss": 0.7017,
+      "step": 4935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.78240803899018e-07,
+      "loss": 0.6997,
+      "step": 4936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.770811797608829e-07,
+      "loss": 0.8114,
+      "step": 4937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7592530646848295e-07,
+      "loss": 0.7051,
+      "step": 4938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7477318446318282e-07,
+      "loss": 0.7064,
+      "step": 4939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7362481418491617e-07,
+      "loss": 0.7346,
+      "step": 4940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7248019607218337e-07,
+      "loss": 0.7067,
+      "step": 4941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7133933056205366e-07,
+      "loss": 0.7315,
+      "step": 4942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7020221809016412e-07,
+      "loss": 0.6355,
+      "step": 4943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6906885909071303e-07,
+      "loss": 0.7102,
+      "step": 4944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6793925399647303e-07,
+      "loss": 0.7102,
+      "step": 4945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6681340323877581e-07,
+      "loss": 0.7043,
+      "step": 4946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6569130724752635e-07,
+      "loss": 0.7446,
+      "step": 4947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6457296645119302e-07,
+      "loss": 0.6431,
+      "step": 4948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.634583812768098e-07,
+      "loss": 0.7709,
+      "step": 4949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623475521499751e-07,
+      "loss": 0.7551,
+      "step": 4950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6124047949485634e-07,
+      "loss": 0.6731,
+      "step": 4951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.601371637341864e-07,
+      "loss": 0.6328,
+      "step": 4952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5903760528926172e-07,
+      "loss": 0.7127,
+      "step": 4953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5794180457994303e-07,
+      "loss": 0.7902,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5684976202465786e-07,
+      "loss": 0.7215,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5576147804040154e-07,
+      "loss": 0.5874,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5467695304272944e-07,
+      "loss": 0.7491,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5359618744576476e-07,
+      "loss": 0.7228,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.525191816621896e-07,
+      "loss": 0.6409,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5144593610326051e-07,
+      "loss": 0.7028,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5037645117878975e-07,
+      "loss": 0.6328,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.49310727297155e-07,
+      "loss": 0.744,
+      "step": 4962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.482487648653008e-07,
+      "loss": 0.776,
+      "step": 4963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.471905642887328e-07,
+      "loss": 0.7446,
+      "step": 4964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4613612597152104e-07,
+      "loss": 0.7203,
+      "step": 4965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.450854503163013e-07,
+      "loss": 0.6893,
+      "step": 4966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.440385377242659e-07,
+      "loss": 0.7241,
+      "step": 4967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4299538859517847e-07,
+      "loss": 0.7,
+      "step": 4968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4195600332735927e-07,
+      "loss": 0.6912,
+      "step": 4969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.409203823176941e-07,
+      "loss": 0.6765,
+      "step": 4970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3988852596163337e-07,
+      "loss": 0.7625,
+      "step": 4971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3886043465318522e-07,
+      "loss": 0.6942,
+      "step": 4972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3783610878492226e-07,
+      "loss": 0.6899,
+      "step": 4973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3681554874798163e-07,
+      "loss": 0.6824,
+      "step": 4974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.357987549320583e-07,
+      "loss": 0.6334,
+      "step": 4975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3478572772541388e-07,
+      "loss": 0.7136,
+      "step": 4976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3377646751486672e-07,
+      "loss": 0.6866,
+      "step": 4977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.327709746857997e-07,
+      "loss": 0.6347,
+      "step": 4978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3176924962215787e-07,
+      "loss": 0.6211,
+      "step": 4979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3077129270644528e-07,
+      "loss": 0.7108,
+      "step": 4980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2977710431972822e-07,
+      "loss": 0.7901,
+      "step": 4981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2878668484163303e-07,
+      "loss": 0.7365,
+      "step": 4982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.278000346503494e-07,
+      "loss": 0.7764,
+      "step": 4983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2681715412262596e-07,
+      "loss": 0.7188,
+      "step": 4984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.258380436337703e-07,
+      "loss": 0.7128,
+      "step": 4985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2486270355765328e-07,
+      "loss": 0.733,
+      "step": 4986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2389113426670597e-07,
+      "loss": 0.6311,
+      "step": 4987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.229233361319171e-07,
+      "loss": 0.6891,
+      "step": 4988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2195930952283775e-07,
+      "loss": 0.6549,
+      "step": 4989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.209990548075768e-07,
+      "loss": 0.7702,
+      "step": 4990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2004257235280536e-07,
+      "loss": 0.6939,
+      "step": 4991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1908986252375243e-07,
+      "loss": 0.7126,
+      "step": 4992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1814092568420588e-07,
+      "loss": 0.7807,
+      "step": 4993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1719576219651585e-07,
+      "loss": 0.7307,
+      "step": 4994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1625437242158588e-07,
+      "loss": 0.6653,
+      "step": 4995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1531675671888621e-07,
+      "loss": 0.7315,
+      "step": 4996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1438291544644042e-07,
+      "loss": 0.7411,
+      "step": 4997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1345284896083109e-07,
+      "loss": 0.7666,
+      "step": 4998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1252655761720299e-07,
+      "loss": 0.7311,
+      "step": 4999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1160404176925543e-07,
+      "loss": 0.7432,
+      "step": 5000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1068530176924775e-07,
+      "loss": 0.7169,
+      "step": 5001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.097703379679993e-07,
+      "loss": 0.687,
+      "step": 5002
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0885915071488396e-07,
+      "loss": 0.6972,
+      "step": 5003
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0795174035783452e-07,
+      "loss": 0.734,
+      "step": 5004
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0704810724334491e-07,
+      "loss": 0.7677,
+      "step": 5005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0614825171646248e-07,
+      "loss": 0.6417,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0525217412079458e-07,
+      "loss": 0.8006,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0435987479850306e-07,
+      "loss": 0.6471,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0347135409031094e-07,
+      "loss": 0.6301,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0258661233549793e-07,
+      "loss": 0.6795,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0170564987189602e-07,
+      "loss": 0.7649,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0082846703590055e-07,
+      "loss": 0.7624,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.995506416245915e-08,
+      "loss": 0.7294,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.908544158507726e-08,
+      "loss": 0.6269,
+      "step": 5014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.821959963581818e-08,
+      "loss": 0.7295,
+      "step": 5015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.735753864530073e-08,
+      "loss": 0.7575,
+      "step": 5016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.649925894269718e-08,
+      "loss": 0.7386,
+      "step": 5017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.564476085574203e-08,
+      "loss": 0.7695,
+      "step": 5018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.479404471071874e-08,
+      "loss": 0.68,
+      "step": 5019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.39471108324741e-08,
+      "loss": 0.7709,
+      "step": 5020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.310395954440276e-08,
+      "loss": 0.7368,
+      "step": 5021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.226459116846054e-08,
+      "loss": 0.6491,
+      "step": 5022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.142900602515881e-08,
+      "loss": 0.7213,
+      "step": 5023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.059720443356124e-08,
+      "loss": 0.7501,
+      "step": 5024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.97691867112882e-08,
+      "loss": 0.6587,
+      "step": 5025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.894495317451456e-08,
+      "loss": 0.6901,
+      "step": 5026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.812450413797192e-08,
+      "loss": 0.8039,
+      "step": 5027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.730783991494629e-08,
+      "loss": 0.6972,
+      "step": 5028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.649496081727715e-08,
+      "loss": 0.6551,
+      "step": 5029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.568586715535843e-08,
+      "loss": 0.6385,
+      "step": 5030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.488055923814186e-08,
+      "loss": 0.7002,
+      "step": 5031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.407903737312929e-08,
+      "loss": 0.7226,
+      "step": 5032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.328130186638029e-08,
+      "loss": 0.7167,
+      "step": 5033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.248735302250565e-08,
+      "loss": 0.6736,
+      "step": 5034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.1697191144674e-08,
+      "loss": 0.8004,
+      "step": 5035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.091081653460287e-08,
+      "loss": 0.6992,
+      "step": 5036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.012822949256981e-08,
+      "loss": 0.7529,
+      "step": 5037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.934943031740139e-08,
+      "loss": 0.6973,
+      "step": 5038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.857441930647858e-08,
+      "loss": 0.7495,
+      "step": 5039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.780319675573689e-08,
+      "loss": 0.6704,
+      "step": 5040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.70357629596663e-08,
+      "loss": 0.6935,
+      "step": 5041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.627211821130576e-08,
+      "loss": 0.5963,
+      "step": 5042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.55122628022531e-08,
+      "loss": 0.6937,
+      "step": 5043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.475619702265403e-08,
+      "loss": 0.6817,
+      "step": 5044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.400392116121091e-08,
+      "loss": 0.7504,
+      "step": 5045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.325543550517732e-08,
+      "loss": 0.7312,
+      "step": 5046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.251074034036021e-08,
+      "loss": 0.7641,
+      "step": 5047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.176983595111653e-08,
+      "loss": 0.693,
+      "step": 5048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.103272262036109e-08,
+      "loss": 0.6771,
+      "step": 5049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.029940062955542e-08,
+      "loss": 0.6963,
+      "step": 5050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.956987025871776e-08,
+      "loss": 0.6875,
+      "step": 5051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.884413178641414e-08,
+      "loss": 0.7295,
+      "step": 5052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.812218548976734e-08,
+      "loss": 0.6615,
+      "step": 5053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.74040316444502e-08,
+      "loss": 0.7504,
+      "step": 5054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.668967052468445e-08,
+      "loss": 0.6589,
+      "step": 5055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.597910240324967e-08,
+      "loss": 0.6733,
+      "step": 5056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.527232755147106e-08,
+      "loss": 0.7195,
+      "step": 5057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.456934623922939e-08,
+      "loss": 0.6951,
+      "step": 5058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.387015873495439e-08,
+      "loss": 0.7568,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.317476530563027e-08,
+      "loss": 0.6376,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.248316621678907e-08,
+      "loss": 0.6188,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.179536173251399e-08,
+      "loss": 0.7314,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.111135211544272e-08,
+      "loss": 0.7923,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.04311376267619e-08,
+      "loss": 0.6425,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.975471852620817e-08,
+      "loss": 0.7574,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.908209507207052e-08,
+      "loss": 0.683,
+      "step": 5066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.84132675211857e-08,
+      "loss": 0.7776,
+      "step": 5067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7748236128946087e-08,
+      "loss": 0.677,
+      "step": 5068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7087001149288555e-08,
+      "loss": 0.7236,
+      "step": 5069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6429562834705576e-08,
+      "loss": 0.7389,
+      "step": 5070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5775921436236335e-08,
+      "loss": 0.6783,
+      "step": 5071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5126077203471186e-08,
+      "loss": 0.7299,
+      "step": 5072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.448003038455163e-08,
+      "loss": 0.7167,
+      "step": 5073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3837781226168115e-08,
+      "loss": 0.7355,
+      "step": 5074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.319932997356114e-08,
+      "loss": 0.663,
+      "step": 5075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2564676870520136e-08,
+      "loss": 0.7039,
+      "step": 5076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.19338221593868e-08,
+      "loss": 0.6166,
+      "step": 5077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 0.6559,
+      "step": 5078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0683508874945774e-08,
+      "loss": 0.702,
+      "step": 5079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.006405077906729e-08,
+      "loss": 0.7709,
+      "step": 5080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.944839202995044e-08,
+      "loss": 0.7734,
+      "step": 5081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.883653286268164e-08,
+      "loss": 0.7363,
+      "step": 5082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.822847351089843e-08,
+      "loss": 0.6665,
+      "step": 5083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7624214206785094e-08,
+      "loss": 0.6925,
+      "step": 5084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.702375518107483e-08,
+      "loss": 0.7381,
+      "step": 5085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6427096663053116e-08,
+      "loss": 0.6812,
+      "step": 5086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.583423888055105e-08,
+      "loss": 0.6152,
+      "step": 5087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5245182059947543e-08,
+      "loss": 0.6709,
+      "step": 5088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.465992642617378e-08,
+      "loss": 0.6796,
+      "step": 5089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4078472202706556e-08,
+      "loss": 0.7846,
+      "step": 5090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.350081961157271e-08,
+      "loss": 0.6092,
+      "step": 5091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.292696887334691e-08,
+      "loss": 0.745,
+      "step": 5092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.235692020714943e-08,
+      "loss": 0.6537,
+      "step": 5093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.179067383065505e-08,
+      "loss": 0.8047,
+      "step": 5094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1228229960080804e-08,
+      "loss": 0.7096,
+      "step": 5095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0669588810193784e-08,
+      "loss": 0.5686,
+      "step": 5096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0114750594308913e-08,
+      "loss": 0.6291,
+      "step": 5097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9563715524290056e-08,
+      "loss": 0.671,
+      "step": 5098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.901648381054779e-08,
+      "loss": 0.739,
+      "step": 5099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.847305566203941e-08,
+      "loss": 0.6783,
+      "step": 5100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.793343128627225e-08,
+      "loss": 0.677,
+      "step": 5101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7397610889300384e-08,
+      "loss": 0.615,
+      "step": 5102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.686559467572348e-08,
+      "loss": 0.689,
+      "step": 5103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.633738284869015e-08,
+      "loss": 0.7742,
+      "step": 5104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.581297560989683e-08,
+      "loss": 0.7661,
+      "step": 5105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.529237315958556e-08,
+      "loss": 0.763,
+      "step": 5106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.477557569654733e-08,
+      "loss": 0.7212,
+      "step": 5107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.426258341811983e-08,
+      "loss": 0.8416,
+      "step": 5108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.375339652018639e-08,
+      "loss": 0.7751,
+      "step": 5109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.324801519717813e-08,
+      "loss": 0.6681,
+      "step": 5110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2746439642072916e-08,
+      "loss": 0.746,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.224867004639642e-08,
+      "loss": 0.7483,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1754706600219955e-08,
+      "loss": 0.645,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.12645494921604e-08,
+      "loss": 0.7039,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.077819890938471e-08,
+      "loss": 0.7245,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.029565503760212e-08,
+      "loss": 0.7735,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9816918061070786e-08,
+      "loss": 0.6225,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9341988162595593e-08,
+      "loss": 0.7802,
+      "step": 5118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8870865523525916e-08,
+      "loss": 0.7392,
+      "step": 5119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8403550323757856e-08,
+      "loss": 0.8217,
+      "step": 5120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.794004274173423e-08,
+      "loss": 0.7174,
+      "step": 5121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7480342954444572e-08,
+      "loss": 0.7586,
+      "step": 5122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7024451137422912e-08,
+      "loss": 0.6833,
+      "step": 5123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6572367464750004e-08,
+      "loss": 0.8344,
+      "step": 5124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6124092109052202e-08,
+      "loss": 0.6985,
+      "step": 5125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.567962524150147e-08,
+      "loss": 0.6645,
+      "step": 5126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5238967031816498e-08,
+      "loss": 0.7517,
+      "step": 5127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.480211764826157e-08,
+      "loss": 0.747,
+      "step": 5128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.436907725764437e-08,
+      "loss": 0.7623,
+      "step": 5129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3939846025321512e-08,
+      "loss": 0.8273,
+      "step": 5130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3514424115191892e-08,
+      "loss": 0.7091,
+      "step": 5131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.309281168970223e-08,
+      "loss": 0.7099,
+      "step": 5132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2675008909843754e-08,
+      "loss": 0.7905,
+      "step": 5133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2261015935152175e-08,
+      "loss": 0.6241,
+      "step": 5134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.185083292370993e-08,
+      "loss": 0.7246,
+      "step": 5135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1444460032142843e-08,
+      "loss": 0.7111,
+      "step": 5136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.104189741562457e-08,
+      "loss": 0.685,
+      "step": 5137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0643145227871032e-08,
+      "loss": 0.7765,
+      "step": 5138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024820362114488e-08,
+      "loss": 0.6919,
+      "step": 5139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.985707274625326e-08,
+      "loss": 0.6924,
+      "step": 5140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.94697527525467e-08,
+      "loss": 0.7491,
+      "step": 5141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9086243787922453e-08,
+      "loss": 0.719,
+      "step": 5142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8706545998824488e-08,
+      "loss": 0.7457,
+      "step": 5143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8330659530234606e-08,
+      "loss": 0.7053,
+      "step": 5144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7958584525687995e-08,
+      "loss": 0.6372,
+      "step": 5145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.759032112725656e-08,
+      "loss": 0.6958,
+      "step": 5146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7225869475563372e-08,
+      "loss": 0.6298,
+      "step": 5147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6865229709770447e-08,
+      "loss": 0.656,
+      "step": 5148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6508401967588738e-08,
+      "loss": 0.7281,
+      "step": 5149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6155386385270367e-08,
+      "loss": 0.7578,
+      "step": 5150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5806183097613058e-08,
+      "loss": 0.828,
+      "step": 5151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5460792237960154e-08,
+      "loss": 0.7412,
+      "step": 5152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5119213938196155e-08,
+      "loss": 0.6767,
+      "step": 5153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.478144832875228e-08,
+      "loss": 0.7254,
+      "step": 5154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.444749553860314e-08,
+      "loss": 0.6224,
+      "step": 5155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4117355695267843e-08,
+      "loss": 0.6487,
+      "step": 5156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.379102892480777e-08,
+      "loss": 0.737,
+      "step": 5157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3468515351831024e-08,
+      "loss": 0.7252,
+      "step": 5158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3149815099487984e-08,
+      "loss": 0.7931,
+      "step": 5159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2834928289472415e-08,
+      "loss": 0.7745,
+      "step": 5160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2523855042022582e-08,
+      "loss": 0.7767,
+      "step": 5161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2216595475921245e-08,
+      "loss": 0.6631,
+      "step": 5162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1913149708494553e-08,
+      "loss": 0.7878,
+      "step": 5163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1613517855612045e-08,
+      "loss": 0.6783,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.131770003168775e-08,
+      "loss": 0.7149,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1025696349676872e-08,
+      "loss": 0.7471,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0737506921080221e-08,
+      "loss": 0.7464,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.045313185594421e-08,
+      "loss": 0.6411,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0172571262853092e-08,
+      "loss": 0.6305,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.89582524894006e-09,
+      "loss": 0.7404,
+      "step": 5170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.622893919879472e-09,
+      "loss": 0.6618,
+      "step": 5171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.353777379889073e-09,
+      "loss": 0.7206,
+      "step": 5172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.088475731728886e-09,
+      "loss": 0.7973,
+      "step": 5173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.82698907670454e-09,
+      "loss": 0.7532,
+      "step": 5174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.569317514662834e-09,
+      "loss": 0.7364,
+      "step": 5175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.315461143996173e-09,
+      "loss": 0.7435,
+      "step": 5176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.065420061638129e-09,
+      "loss": 0.746,
+      "step": 5177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.819194363064552e-09,
+      "loss": 0.7137,
+      "step": 5178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.576784142298011e-09,
+      "loss": 0.8142,
+      "step": 5179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.338189491900016e-09,
+      "loss": 0.7008,
+      "step": 5180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.103410502977692e-09,
+      "loss": 0.7178,
+      "step": 5181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8724472651815474e-09,
+      "loss": 0.6621,
+      "step": 5182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.645299866703259e-09,
+      "loss": 0.678,
+      "step": 5183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.421968394277889e-09,
+      "loss": 0.7657,
+      "step": 5184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.202452933184999e-09,
+      "loss": 0.7331,
+      "step": 5185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.986753567244208e-09,
+      "loss": 0.6533,
+      "step": 5186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.774870378820741e-09,
+      "loss": 0.7387,
+      "step": 5187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.566803448819879e-09,
+      "loss": 0.7203,
+      "step": 5188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.362552856692515e-09,
+      "loss": 0.7901,
+      "step": 5189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.162118680431816e-09,
+      "loss": 0.7114,
+      "step": 5190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.965500996571004e-09,
+      "loss": 0.6893,
+      "step": 5191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.772699880187804e-09,
+      "loss": 0.7242,
+      "step": 5192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.583715404904432e-09,
+      "loss": 0.7602,
+      "step": 5193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.398547642882056e-09,
+      "loss": 0.6179,
+      "step": 5194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2171966648285604e-09,
+      "loss": 0.794,
+      "step": 5195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.039662539989664e-09,
+      "loss": 0.7588,
+      "step": 5196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.865945336156695e-09,
+      "loss": 0.6941,
+      "step": 5197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.696045119664371e-09,
+      "loss": 0.6611,
+      "step": 5198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.529961955387462e-09,
+      "loss": 0.8525,
+      "step": 5199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3676959067441282e-09,
+      "loss": 0.6351,
+      "step": 5200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.209247035694807e-09,
+      "loss": 0.6985,
+      "step": 5201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.054615402743322e-09,
+      "loss": 0.6274,
+      "step": 5202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9038010669346637e-09,
+      "loss": 0.7411,
+      "step": 5203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.756804085857212e-09,
+      "loss": 0.7157,
+      "step": 5204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.613624515641622e-09,
+      "loss": 0.655,
+      "step": 5205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4742624109586054e-09,
+      "loss": 0.6614,
+      "step": 5206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.338717825025594e-09,
+      "loss": 0.6854,
+      "step": 5207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.206990809597853e-09,
+      "loss": 0.6361,
+      "step": 5208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.079081414975148e-09,
+      "loss": 0.6927,
+      "step": 5209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9549896900006305e-09,
+      "loss": 0.6467,
+      "step": 5210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8347156820563983e-09,
+      "loss": 0.6609,
+      "step": 5211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7182594370701577e-09,
+      "loss": 0.6987,
+      "step": 5212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6056209995096717e-09,
+      "loss": 0.6964,
+      "step": 5213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4968004123849799e-09,
+      "loss": 0.7573,
+      "step": 5214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.39179771724951e-09,
+      "loss": 0.7006,
+      "step": 5215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2906129541978563e-09,
+      "loss": 0.7752,
+      "step": 5216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.19324616186689e-09,
+      "loss": 0.762,
+      "step": 5217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0996973774368702e-09,
+      "loss": 0.8608,
+      "step": 5218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0099666366281125e-09,
+      "loss": 0.6901,
+      "step": 5219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.240539737032095e-10,
+      "loss": 0.6682,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.419594214692517e-10,
+      "loss": 0.7115,
+      "step": 5221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.636830112733862e-10,
+      "loss": 0.6993,
+      "step": 5222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.892247730039269e-10,
+      "loss": 0.6208,
+      "step": 5223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.185847350936858e-10,
+      "loss": 0.6869,
+      "step": 5224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.517629245155309e-10,
+      "loss": 0.7044,
+      "step": 5225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.887593667857183e-10,
+      "loss": 0.7505,
+      "step": 5226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.295740859627806e-10,
+      "loss": 0.6622,
+      "step": 5227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7420710464419754e-10,
+      "loss": 0.6815,
+      "step": 5228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2265844397305625e-10,
+      "loss": 0.7077,
+      "step": 5229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7492812363250074e-10,
+      "loss": 0.802,
+      "step": 5230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.310161618490625e-10,
+      "loss": 0.7022,
+      "step": 5231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9092257538932956e-10,
+      "loss": 0.7116,
+      "step": 5232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5464737956438768e-10,
+      "loss": 0.722,
+      "step": 5233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2219058822315888e-10,
+      "loss": 0.6582,
+      "step": 5234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.355221376239343e-11,
+      "loss": 0.6885,
+      "step": 5235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.873226711445746e-11,
+      "loss": 0.7716,
+      "step": 5236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.773075775843517e-11,
+      "loss": 0.6944,
+      "step": 5237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.054769371346744e-11,
+      "loss": 0.7702,
+      "step": 5238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.718308154208259e-11,
+      "loss": 0.6941,
+      "step": 5239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.63692634464519e-12,
+      "loss": 0.7083,
+      "step": 5240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9092317682378738e-12,
+      "loss": 0.7102,
+      "step": 5241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6889,
+      "step": 5242
+    },
+    {
+      "epoch": 1.0,
+      "step": 5242,
+      "total_flos": 6.260178673760993e+18,
+      "train_loss": 0.787337068063255,
+      "train_runtime": 73991.8237,
+      "train_samples_per_second": 9.07,
+      "train_steps_per_second": 0.071
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5242,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 24000,
+  "total_flos": 6.260178673760993e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}